Parkinson’s Disease (PD) is a progressive neurodegenerative disorder that causes uncontrollable
movements and difficulty with balance and coordination. It is highly important for early detection of
Parkinson’s Disease for patients to receive proper treatment. This paper aims to present a preliminary
data mining procedure that help Parkinson’s Disease patients slow down their progression of the disease
while helping early detection of the disease. For early non-invasive treatment, our research first analyses
the early symptoms of Parkinson’s Disease, designs/selects a proper demo video, let the user follow the
demo to exercise and upload his exercise video to our deep learning APP: LaBelle. LaBelle utilizing
MediaPipe Pose to identify, analyze, and store data about the poses and movements of both demo and the
user, calculates the angles created between different joints and major body parts. LaBelle’s AI model uses
a K-means clustering algorithm to create a group of clusters for both demo and the user dataset. Using the
two sets of clusters, LaBelle identifies the key frames in the user video and searches the demo cluster set
for a matching set of properties and frames. It evaluates the differences between the paired frames and
produces a final score as well as feedback on the poses that need improving. Meanwhile, if the user is
willing to donate their exercise data, he can simply input his age, whether he is a PD patient (maybe for
how long) anonymously. Then his data can be stored into our customized dataset, used in data mining for
Parkinson’s Disease prediction, which involves building/training our deep learning CNN model and help
early detection of Parkinson’s Disease.
International Journal of Engineering and Science Invention (IJESI) is an international journal intended for professionals and researchers in all fields of computer science and electronics. IJESI publishes research articles and reviews within the whole field Engineering Science and Technology, new teaching methods, assessment, validation and the impact of new technologies and it will continue to provide information on the latest trends and developments in this ever-expanding subject. The publications of papers are selected through double peer reviewed to ensure originality, relevance, and readability. The articles published in our journal can be accessed online.
Bidirectional Recurrent Network and Neuro‑fuzzy Frequent Pattern Mining for H...BASMAJUMAASALEHALMOH
This document summarizes a research paper that proposes a novel method called Bi-directional Recurrent Network and Neuro-Fuzzy-based (BRN-NF) for heart disease prediction. The method uses a bi-directional recurrent neural network to mine frequent patterns from heart disease data. It then employs a neuro-fuzzy inference algorithm using the mined patterns to predict heart disease. Experiments on cardiovascular disease data show the proposed method improves prediction accuracy, sensitivity and specificity compared to existing methods. The accurate predictions indicate the notable performance of the BRN-NF method.
Top Cited Articles in Data Mining - International Journal of Data Mining & Kn...IJDKP
Data mining and knowledge discovery in databases have been attracting a significant amount of research, industry, and media attention of late. There is an urgent need for a new generation of computational theories and tools to assist researchers in extracting useful information from the rapidly growing volumes of digital data.
This Journal provides a forum for researchers who address this issue and to present their work in a peer-reviewed open access forum. Authors are solicited to contribute to the Journal by submitting articles that illustrate research results, projects, surveying works and industrial experiences that describe significant advances in the following areas, but are not limited to these topics only.
Model for autism disorder detection using deep learningIAESIJAI
Autism spectrum disorder (ASD) is a neurodegenerative illness that impacts individuals' social abilities. The majority of available approaches rely on structural and resting-state functional magnetic resonance imaging (fMRI) to detect ASD with a small dataset, resulting in high accuracy but low generality. To detect ASD with a limited dataset, the bulk of known technologies involve Machine Learning, pattern recognition, and other techniques, leading to high accuracy but moderate generality. To address this constraint and improve the efficacy of the automated autism diagnosis model, an ASD detection model based on deep learning (DL) is provided in this work. The classification challenge is solved using a convolutional neural network classifier. The suggested model beats state-of-the-art methodologies in terms of accuracy, according to simulation findings. The proposed approach investigates how anatomical and functional connectivity indicators can be used to determine whether or not a person is autistic. The proposed method delivers state-of-the-art results, with the classification of Autistic patients achieving 93.41% accuracy and the localization of the classified data regressed to 0.29 mean absolute error (MAE).
diabetic Retinopathy. Eye detection of diseaseshivubhavv
This document discusses using convolutional neural networks (CNNs) for automated detection of diabetic retinopathy (DR) from retinal images. DR is a leading cause of blindness that can be prevented if detected early. Traditional screening requires manual examination by experts and is inefficient. The study aims to investigate whether CNNs can achieve high accuracy in detecting and classifying DR from a publicly available dataset of retinal images. Results showed the proposed CNN model outperformed existing methods, suggesting deep learning can help improve early detection of DR and reduce its burden on healthcare systems.
Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral...ijtsrd
This paper presents the development of a hybrid dynamic expert system for the diagnosis of peripheral diabetes and remedies using a rule based machine learning technique. The aim was to develop a solution to the risk factors of peripheral diabetes. The methodology applied in this study is the experimental method, and the software design methodology used was the agile methodology. Data was collected from Nnamdi Azikiwe University Teaching Hospitals NAUTH and the Lagos State University Teaching Hospital LASUTH for patients between the ages of 28 87years suffering from peripheral neuropathy. Other methods used were data integration by applying uniform data access UDA technique, data processing using Infinite Impulse Response Filter IIRF , data extraction with a computerized approach, machine learning algorithm with Dynamic Feed Forward Neural Network DFNN , rule base algorithm. The modeling of the hybrid dynamic expert system and remedies was achieved using the DFNN for the detection of DPN and a rule based model for remedies and recommendations. The models were implemented with MATLAB and Java programming languages. The result when evaluated achieved a Mean Square Error MSE of 4.9392e 11 and Regression R of 0.99823. The implication of the result showed that the peripheral diabetes detection model correctly learns the peripheral diabetes attributes and was also able to correctly detect peripheral diabetes in patients. The model when compared with other sophisticated models also showed that it achieved a better regression score. The reason was due to the appropriate steps used in the data preparation such as integration and the use of IIFR filter, feature extraction, and the deep configuration of the regression model. Omeye Emmanuel C. | Ngene John N. | Dr. Anyaragbu Hope U. | Dr. Ozioko Ekene | Dr. Iloka Bethram C. | Prof. Inyiama Hycent C. "Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral Diabetes and Remedies using a Rule-Based Machine Learning Technique" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-6 | Issue-7 , December 2022, URL: https://www.ijtsrd.com/papers/ijtsrd52356.pdf Paper URL: https://www.ijtsrd.com/computer-science/other/52356/development-of-a-hybrid-dynamic-expert-system-for-the-diagnosis-of-peripheral-diabetes-and-remedies-using-a-rulebased-machine-learning-technique/omeye-emmanuel-c
A transfer learning with deep neural network approach for diabetic retinopath...IJECEIAES
Diabetic retinopathy is an eye disease caused by high blood sugar and pressure which damages the blood vessels in the eye. Diabetic retinopathy is the root cause of more than 1% of the blindness worldwide. Early detection of this disease is crucial as it prevents it from progressing to a more severe level. However, the current machine learning-based approaches for detecting the severity level of diabetic retinopathy are either, i) rely on manually extracting features which makes an approach unpractical, or ii) trained on small dataset thus cannot be generalized. In this study, we propose a transfer learning-based approach for detecting the severity level of the diabetic retinopathy with high accuracy. Our model is a deep learning model based on global average pooling (GAP) technique with various pre-trained convolutional neural net- work (CNN) models. The experimental results of our approach, in which our best model achieved 82.4% quadratic weighted kappa (QWK), corroborate the ability of our model to detect the severity level of diabetic retinopathy efficiently.
International Journal of Engineering and Science Invention (IJESI) is an international journal intended for professionals and researchers in all fields of computer science and electronics. IJESI publishes research articles and reviews within the whole field Engineering Science and Technology, new teaching methods, assessment, validation and the impact of new technologies and it will continue to provide information on the latest trends and developments in this ever-expanding subject. The publications of papers are selected through double peer reviewed to ensure originality, relevance, and readability. The articles published in our journal can be accessed online.
Bidirectional Recurrent Network and Neuro‑fuzzy Frequent Pattern Mining for H...BASMAJUMAASALEHALMOH
This document summarizes a research paper that proposes a novel method called Bi-directional Recurrent Network and Neuro-Fuzzy-based (BRN-NF) for heart disease prediction. The method uses a bi-directional recurrent neural network to mine frequent patterns from heart disease data. It then employs a neuro-fuzzy inference algorithm using the mined patterns to predict heart disease. Experiments on cardiovascular disease data show the proposed method improves prediction accuracy, sensitivity and specificity compared to existing methods. The accurate predictions indicate the notable performance of the BRN-NF method.
Top Cited Articles in Data Mining - International Journal of Data Mining & Kn...IJDKP
Data mining and knowledge discovery in databases have been attracting a significant amount of research, industry, and media attention of late. There is an urgent need for a new generation of computational theories and tools to assist researchers in extracting useful information from the rapidly growing volumes of digital data.
This Journal provides a forum for researchers who address this issue and to present their work in a peer-reviewed open access forum. Authors are solicited to contribute to the Journal by submitting articles that illustrate research results, projects, surveying works and industrial experiences that describe significant advances in the following areas, but are not limited to these topics only.
Model for autism disorder detection using deep learningIAESIJAI
Autism spectrum disorder (ASD) is a neurodegenerative illness that impacts individuals' social abilities. The majority of available approaches rely on structural and resting-state functional magnetic resonance imaging (fMRI) to detect ASD with a small dataset, resulting in high accuracy but low generality. To detect ASD with a limited dataset, the bulk of known technologies involve Machine Learning, pattern recognition, and other techniques, leading to high accuracy but moderate generality. To address this constraint and improve the efficacy of the automated autism diagnosis model, an ASD detection model based on deep learning (DL) is provided in this work. The classification challenge is solved using a convolutional neural network classifier. The suggested model beats state-of-the-art methodologies in terms of accuracy, according to simulation findings. The proposed approach investigates how anatomical and functional connectivity indicators can be used to determine whether or not a person is autistic. The proposed method delivers state-of-the-art results, with the classification of Autistic patients achieving 93.41% accuracy and the localization of the classified data regressed to 0.29 mean absolute error (MAE).
diabetic Retinopathy. Eye detection of diseaseshivubhavv
This document discusses using convolutional neural networks (CNNs) for automated detection of diabetic retinopathy (DR) from retinal images. DR is a leading cause of blindness that can be prevented if detected early. Traditional screening requires manual examination by experts and is inefficient. The study aims to investigate whether CNNs can achieve high accuracy in detecting and classifying DR from a publicly available dataset of retinal images. Results showed the proposed CNN model outperformed existing methods, suggesting deep learning can help improve early detection of DR and reduce its burden on healthcare systems.
Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral...ijtsrd
This paper presents the development of a hybrid dynamic expert system for the diagnosis of peripheral diabetes and remedies using a rule based machine learning technique. The aim was to develop a solution to the risk factors of peripheral diabetes. The methodology applied in this study is the experimental method, and the software design methodology used was the agile methodology. Data was collected from Nnamdi Azikiwe University Teaching Hospitals NAUTH and the Lagos State University Teaching Hospital LASUTH for patients between the ages of 28 87years suffering from peripheral neuropathy. Other methods used were data integration by applying uniform data access UDA technique, data processing using Infinite Impulse Response Filter IIRF , data extraction with a computerized approach, machine learning algorithm with Dynamic Feed Forward Neural Network DFNN , rule base algorithm. The modeling of the hybrid dynamic expert system and remedies was achieved using the DFNN for the detection of DPN and a rule based model for remedies and recommendations. The models were implemented with MATLAB and Java programming languages. The result when evaluated achieved a Mean Square Error MSE of 4.9392e 11 and Regression R of 0.99823. The implication of the result showed that the peripheral diabetes detection model correctly learns the peripheral diabetes attributes and was also able to correctly detect peripheral diabetes in patients. The model when compared with other sophisticated models also showed that it achieved a better regression score. The reason was due to the appropriate steps used in the data preparation such as integration and the use of IIFR filter, feature extraction, and the deep configuration of the regression model. Omeye Emmanuel C. | Ngene John N. | Dr. Anyaragbu Hope U. | Dr. Ozioko Ekene | Dr. Iloka Bethram C. | Prof. Inyiama Hycent C. "Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral Diabetes and Remedies using a Rule-Based Machine Learning Technique" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-6 | Issue-7 , December 2022, URL: https://www.ijtsrd.com/papers/ijtsrd52356.pdf Paper URL: https://www.ijtsrd.com/computer-science/other/52356/development-of-a-hybrid-dynamic-expert-system-for-the-diagnosis-of-peripheral-diabetes-and-remedies-using-a-rulebased-machine-learning-technique/omeye-emmanuel-c
A transfer learning with deep neural network approach for diabetic retinopath...IJECEIAES
Diabetic retinopathy is an eye disease caused by high blood sugar and pressure which damages the blood vessels in the eye. Diabetic retinopathy is the root cause of more than 1% of the blindness worldwide. Early detection of this disease is crucial as it prevents it from progressing to a more severe level. However, the current machine learning-based approaches for detecting the severity level of diabetic retinopathy are either, i) rely on manually extracting features which makes an approach unpractical, or ii) trained on small dataset thus cannot be generalized. In this study, we propose a transfer learning-based approach for detecting the severity level of the diabetic retinopathy with high accuracy. Our model is a deep learning model based on global average pooling (GAP) technique with various pre-trained convolutional neural net- work (CNN) models. The experimental results of our approach, in which our best model achieved 82.4% quadratic weighted kappa (QWK), corroborate the ability of our model to detect the severity level of diabetic retinopathy efficiently.
Effect of Data Size on Feature Set Using Classification in Health Domaindbpublications
In health domain, the major critical issue is prediction of disease in early stage. Prediction of disease is mainly based on the experience of physician so many machine learning approach contribute their work in the prediction of disease. In existing approaches, either prediction or feature selection has been concentrated. The aim of this paper is to present the effect of data size and set of features in the prediction of disease in health domain using Naïve Bayes. This shows how each attribute or combination of attribute behaves on different size of dataset.
This document describes a proposed system to automate the diagnosis of Autism Spectrum Disorder (ASD) using machine learning techniques. The system uses a recurrent neural network classifier trained on EEG data to predict whether a patient has ASD or not. Preprocessing steps like feature selection and engineering are used to clean the data before training. The proposed system aims to provide faster and more accurate diagnosis of ASD compared to existing methods while using less memory and data. Key advantages include reduced data loss and processing time while maintaining high accuracy levels.
DATA MINING CLASSIFICATION ALGORITHMS FOR KIDNEY DISEASE PREDICTION IJCI JOURNAL
Data mining is a non-trivial process of categorizing valid, novel, potentially useful and ultimately understandable patterns in data. In terms, it accurately state as the extraction of information from a huge database. Data mining is a vital role in several applications such as business organizations, educational institutions, government sectors, health care industry, scientific and engineering. . In the health care
industry, the data mining is predominantly used for disease prediction. Enormous data mining techniques are existing for predicting diseases namely classification, clustering, association rules, summarizations, regression and etc. The main objective of this research work is to predict kidney diseases using classification algorithms such as Naïve Bayes and Support Vector Machine. This research work mainly
focused on finding the best classification algorithm based on the classification accuracy and execution time performance factors. From the experimental results it is observed that the performance of the SVM is better than the Naive Bayes classifier algorithm.
This document outlines a project to develop a deep learning model for detecting the severity of diabetic retinopathy using retinal images. It discusses the challenges of limited data, annotation variability, lack of interpretability in models, and ensuring generalizability and clinical integration. The motivation is to make diabetic retinopathy screening more accessible and affordable through an automated solution, in order to improve patient outcomes. The objectives are to develop a deep learning model for automated detection of diabetic retinopathy severity from retinal images.
Predicting Chronic Kidney Disease using Data Mining Techniquesijtsrd
Kidney is a significant aspect of a human body. Kidney infection or disappointments are expanded in every year. Presently a day’s chronic kidney infection is the most well known disease for the individuals. Today numerous individuals pass on due to chronic kidney disease. The principle issue of CKD is, it will influence the kidney gradually. A few people dont have side effects at all and are analysed by a lab test. It depicts the steady loss of kidney work. Early recognition and therapy are viewed as basic variables in the management and control of chronic kidney disease. Data mining techniques is utilized to extract data from clinical and laboratory, which can be useful to help doctors to recognize the seriousness stage of patients. Using Probabilistic Neural Networks PNN algorithm will get better prediction for determining the severity stage of chronic kidney disease. Seethal V | Kuldeep Baban Vayadande "Predicting Chronic Kidney Disease using Data Mining Techniques" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-5 | Issue-1 , December 2020, URL: https://www.ijtsrd.com/papers/ijtsrd37974.pdf Paper URL : https://www.ijtsrd.com/computer-science/data-miining/37974/predicting-chronic-kidney-disease-using-data-mining-techniques/seethal-v
This document summarizes a presentation on artificial intelligence in medical research given by Dr. Ahmed Elngar. It discusses the history and foundations of AI, as well as key research directions like machine learning, expert systems, computer vision, neural networks, and deep learning. Deep learning is seen as core to AI and medicine, with applications in medical imaging, video analysis, and clinical tasks like screening, diagnosis and monitoring. Challenges remain around data scarcity, bias and ethical deployment, but AI has great potential to improve healthcare accuracy, efficiency and access worldwide.
Sleep quality prediction from wearable data using deep learningLuis Fernandez Luque
Deep learning models were better than linear regression at predicting sleep quality from physical activity data collected by wearables. Specifically, a Time-batched Long Short-Term Memory recurrent neural network model achieved the best performance, with high sensitivity for correctly identifying individuals with good sleep patterns. This research demonstrated the potential for using machine learning on raw sensor data to gain insights into sleep science and develop electronic health applications to monitor and improve sleep quality.
Alzheimer Disease Prediction using Machine Learning AlgorithmsIRJET Journal
This document summarizes research on using machine learning algorithms to predict Alzheimer's disease. It discusses collecting data from the Alzheimer's Disease Neuroimaging Initiative database and preprocessing MRI images. Feature extraction is performed using techniques like volume and thickness measurements. Machine learning models like CNNs and SVMs are trained on the data and tested to distinguish between patient groups and predict progression. The research aims to more accurately diagnose Alzheimer's at earlier stages by combining clinical assessments with structural neuroimaging data and machine learning. Accuracy of over 90% was achieved in some cases at distinguishing between patient classifications like AD vs normal control.
The document discusses research on developing machine learning and deep learning models to detect mental health issues in university students. It introduces the CASTLE framework, which uses multi-modal data like social interactions, academic performance, and appearance to detect issues. It also discusses using EEG data and features to recognize multiple anxiety levels in students. The objectives are to improve CASTLE to use more student data and analyze results with causal learning techniques. Coursework focuses on machine learning, deep learning, and related topics to support the research.
A comparative study on remote tracking of parkinson’s disease progression usi...ijfcstjournal
In recent years, applications of data mining method
s are become more popular in many fields of medical
diagnosis and evaluations. The data mining methods
are appropriate tools for discovering and extractin
g
of available knowledge in medical databases. In thi
s study, we divided 11 data mining algorithms into
five
groups which are applied to a dataset of patient’s
clinical variables data with Parkinson’s Disease (P
D) to
study the disease progression. The dataset includes
22 properties of 42 people that all of our algorit
hms
are applied to this dataset. The Decision Table wit
h 0.9985 correlation coefficients has the best accu
racy
and Decision Stump with 0.7919 correlation coeffici
ents has the lowest accuracy.
The document discusses a hybrid CNN and LSTM network for heart disease prediction. It begins by noting heart disease is a leading cause of death worldwide and that traditional machine learning methods have achieved only 65-85% accuracy in prediction. The proposed method uses a convolutional neural network to extract features from heart disease data, and then an LSTM network to classify the data as normal or abnormal based on those features. When tested on heart disease data, the hybrid model achieved 89% accuracy, outperforming other machine learning algorithms like SVM, Naive Bayes and decision trees.
Deep Learning for Predictive Analytics in Healthcare – Pubrica.pptxPubrica
Predictive analytics services helps healthcare life sciences and providers by utilising various approaches such as data mining, statistics, modelling, machine learning, and artificial intelligence to explore current results and generate future predictions.
Learn More : https://bit.ly/3qWpOjV
Reference: https://pubrica.com/services/data-analytics-machine-learning/predictive-analytics/
Why Pubrica:
When you order our services, we promise you the following – Plagiarism free | always on Time | 24*7 customer support | Written to international Standard | Unlimited Revisions support | Medical writing Expert | Publication Support | Bio statistical experts | High-quality Subject Matter Experts.
Contact us:
Web: https://pubrica.com/
Blog: https://pubrica.com/academy/
Email: sales@pubrica.com
WhatsApp : +91 9884350006
United Kingdom: +44-1618186353
Discovering Abnormal Patches and Transformations of Diabetics Retinopathy in ...cscpconf
Diabetic retinopathy (DR) is one of the retinal diseases due to long-term effect of diabetes.
Early detection for diabetic retinopathy is crucial since timely treatment can prevent
progressive loss of vision. The most common diagnosis technique of diabetic retinopathy is to
screen abnormalities through retinal fundus images by clinicians. However, limited number of
well-trained clinicians increase the possibilities of misdiagnosing. In this work, we propose a
big-data-driven automatic computer-aided diagnosing (CAD) system for diabetic retinopathy
severity regression based on transfer learning, which starts from a deep convolutional neural
network pre-trained on generic images, and adapts it to large-scale DR datasets. From images
in the training set, we also automatically segment the abnormal patches with an occlusion test,
and model the transformations and deterioration process of DR. Our results can be widely used
for fast diagnosis of DR, medical education and public-level healthcare propagation.
DISCOVERING ABNORMAL PATCHES AND TRANSFORMATIONS OF DIABETICS RETINOPATHY IN ...csandit
Diabetic retinopathy (DR) is one of the retinal diseases due to long-term effect of diabetes.Early detection for diabetic retinopathy is crucial since timely treatment can prevent
progressive loss of vision. The most common diagnosis technique of diabetic retinopathy is to screen abnormalities through retinal fundus images by clinicians. However, limited number of well-trained clinicians increase the possibilities of misdiagnosing. In this work, we propose a big-data-driven automatic computer-aided diagnosing (CAD) system for diabetic retinopathy severity regression based on transfer learning, which starts from a deep convolutional neural
network pre-trained on generic images, and adapts it to large-scale DR datasets. From images in the training set, we also automatically segment the abnormal patches with an occlusion test,and model the transformations and deterioration process of DR. Our results can be widely used for fast diagnosis of DR, medical education and public-level healthcare propagation.
Gears are the most essential and commonly utilised power transmission components. It is really essential to operate machines involving different weights and speeds. When a load is increased beyond a certain limit, gear teeth frequently fail. Composite materials, in comparison to other metallic gears, offer significantly better mechanical qualities, such as a higher strength-to-weight ratio, increased hardness, and hence a lower risk of failure. Al6063 and SiC were employed to build a metal matrix composite for spur gear production in this work.
Sound plays a crucial part in every element of human life. Sound is a crucial component in the development of automated systems in a variety of domains, from personal security to essential monitoring. There are a few systems on the market now, but their efficiency is a worry for their use in real-world circumstances. Image classification and feature classification are the same as sound classification, just like other classification algorithms like machine learning.
Heaviness has been related to stroke, depression, and cancer are some of the most serious dangers to human existence. Heart disease, stroke, obesity, and type II diabetes are all disorders that have an impact on our way of life. Using data mining and machine learning approaches to forecast disease based on patient treatment history and health data has been a battle for decades.
javed_prethesis2608 on predcition of heart diseasejaved75
This document presents a thesis proposal for using deep neural networks to predict heart disease from patient data. The motivation is to improve prediction accuracy rates compared to previous machine learning models. The objectives are to explore patient data, preprocess the data, create a neural network model in TensorFlow, train the model, and predict heart disease. The methodology involves feature processing, feature selection, data preprocessing, creating a neural network, training the model, and predicting outcomes. The expected outcome is a prediction accuracy rate of over 75% using this deep learning approach.
This document discusses neural networks and their applications in signal processing. It begins with an introduction to neural networks and their biological inspiration. Applications of neural networks in signal processing are then outlined, including EEG, EMG, medical diagnosis, and more. The document reviews two papers on deep learning applications in biomedical fields and identifies challenges around model building and interpretability. It sets research objectives to optimize mathematical models and improve result interpretation. In conclusion, deep learning has advanced pattern recognition but challenges remain around data availability and model methodology.
Effect of Data Size on Feature Set Using Classification in Health Domaindbpublications
In health domain, the major critical issue is prediction of disease in early stage. Prediction of disease is mainly based on the experience of physician so many machine learning approach contribute their work in the prediction of disease. In existing approaches, either prediction or feature selection has been concentrated. The aim of this paper is to present the effect of data size and set of features in the prediction of disease in health domain using Naïve Bayes. This shows how each attribute or combination of attribute behaves on different size of dataset.
This document describes a proposed system to automate the diagnosis of Autism Spectrum Disorder (ASD) using machine learning techniques. The system uses a recurrent neural network classifier trained on EEG data to predict whether a patient has ASD or not. Preprocessing steps like feature selection and engineering are used to clean the data before training. The proposed system aims to provide faster and more accurate diagnosis of ASD compared to existing methods while using less memory and data. Key advantages include reduced data loss and processing time while maintaining high accuracy levels.
DATA MINING CLASSIFICATION ALGORITHMS FOR KIDNEY DISEASE PREDICTION IJCI JOURNAL
Data mining is a non-trivial process of categorizing valid, novel, potentially useful and ultimately understandable patterns in data. In terms, it accurately state as the extraction of information from a huge database. Data mining is a vital role in several applications such as business organizations, educational institutions, government sectors, health care industry, scientific and engineering. . In the health care
industry, the data mining is predominantly used for disease prediction. Enormous data mining techniques are existing for predicting diseases namely classification, clustering, association rules, summarizations, regression and etc. The main objective of this research work is to predict kidney diseases using classification algorithms such as Naïve Bayes and Support Vector Machine. This research work mainly
focused on finding the best classification algorithm based on the classification accuracy and execution time performance factors. From the experimental results it is observed that the performance of the SVM is better than the Naive Bayes classifier algorithm.
This document outlines a project to develop a deep learning model for detecting the severity of diabetic retinopathy using retinal images. It discusses the challenges of limited data, annotation variability, lack of interpretability in models, and ensuring generalizability and clinical integration. The motivation is to make diabetic retinopathy screening more accessible and affordable through an automated solution, in order to improve patient outcomes. The objectives are to develop a deep learning model for automated detection of diabetic retinopathy severity from retinal images.
Predicting Chronic Kidney Disease using Data Mining Techniquesijtsrd
Kidney is a significant aspect of a human body. Kidney infection or disappointments are expanded in every year. Presently a day’s chronic kidney infection is the most well known disease for the individuals. Today numerous individuals pass on due to chronic kidney disease. The principle issue of CKD is, it will influence the kidney gradually. A few people dont have side effects at all and are analysed by a lab test. It depicts the steady loss of kidney work. Early recognition and therapy are viewed as basic variables in the management and control of chronic kidney disease. Data mining techniques is utilized to extract data from clinical and laboratory, which can be useful to help doctors to recognize the seriousness stage of patients. Using Probabilistic Neural Networks PNN algorithm will get better prediction for determining the severity stage of chronic kidney disease. Seethal V | Kuldeep Baban Vayadande "Predicting Chronic Kidney Disease using Data Mining Techniques" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-5 | Issue-1 , December 2020, URL: https://www.ijtsrd.com/papers/ijtsrd37974.pdf Paper URL : https://www.ijtsrd.com/computer-science/data-miining/37974/predicting-chronic-kidney-disease-using-data-mining-techniques/seethal-v
This document summarizes a presentation on artificial intelligence in medical research given by Dr. Ahmed Elngar. It discusses the history and foundations of AI, as well as key research directions like machine learning, expert systems, computer vision, neural networks, and deep learning. Deep learning is seen as core to AI and medicine, with applications in medical imaging, video analysis, and clinical tasks like screening, diagnosis and monitoring. Challenges remain around data scarcity, bias and ethical deployment, but AI has great potential to improve healthcare accuracy, efficiency and access worldwide.
Sleep quality prediction from wearable data using deep learningLuis Fernandez Luque
Deep learning models were better than linear regression at predicting sleep quality from physical activity data collected by wearables. Specifically, a Time-batched Long Short-Term Memory recurrent neural network model achieved the best performance, with high sensitivity for correctly identifying individuals with good sleep patterns. This research demonstrated the potential for using machine learning on raw sensor data to gain insights into sleep science and develop electronic health applications to monitor and improve sleep quality.
Alzheimer Disease Prediction using Machine Learning AlgorithmsIRJET Journal
This document summarizes research on using machine learning algorithms to predict Alzheimer's disease. It discusses collecting data from the Alzheimer's Disease Neuroimaging Initiative database and preprocessing MRI images. Feature extraction is performed using techniques like volume and thickness measurements. Machine learning models like CNNs and SVMs are trained on the data and tested to distinguish between patient groups and predict progression. The research aims to more accurately diagnose Alzheimer's at earlier stages by combining clinical assessments with structural neuroimaging data and machine learning. Accuracy of over 90% was achieved in some cases at distinguishing between patient classifications like AD vs normal control.
The document discusses research on developing machine learning and deep learning models to detect mental health issues in university students. It introduces the CASTLE framework, which uses multi-modal data like social interactions, academic performance, and appearance to detect issues. It also discusses using EEG data and features to recognize multiple anxiety levels in students. The objectives are to improve CASTLE to use more student data and analyze results with causal learning techniques. Coursework focuses on machine learning, deep learning, and related topics to support the research.
A comparative study on remote tracking of parkinson’s disease progression usi...ijfcstjournal
In recent years, applications of data mining method
s are become more popular in many fields of medical
diagnosis and evaluations. The data mining methods
are appropriate tools for discovering and extractin
g
of available knowledge in medical databases. In thi
s study, we divided 11 data mining algorithms into
five
groups which are applied to a dataset of patient’s
clinical variables data with Parkinson’s Disease (P
D) to
study the disease progression. The dataset includes
22 properties of 42 people that all of our algorit
hms
are applied to this dataset. The Decision Table wit
h 0.9985 correlation coefficients has the best accu
racy
and Decision Stump with 0.7919 correlation coeffici
ents has the lowest accuracy.
The document discusses a hybrid CNN and LSTM network for heart disease prediction. It begins by noting heart disease is a leading cause of death worldwide and that traditional machine learning methods have achieved only 65-85% accuracy in prediction. The proposed method uses a convolutional neural network to extract features from heart disease data, and then an LSTM network to classify the data as normal or abnormal based on those features. When tested on heart disease data, the hybrid model achieved 89% accuracy, outperforming other machine learning algorithms like SVM, Naive Bayes and decision trees.
Deep Learning for Predictive Analytics in Healthcare – Pubrica.pptxPubrica
Predictive analytics services helps healthcare life sciences and providers by utilising various approaches such as data mining, statistics, modelling, machine learning, and artificial intelligence to explore current results and generate future predictions.
Learn More : https://bit.ly/3qWpOjV
Reference: https://pubrica.com/services/data-analytics-machine-learning/predictive-analytics/
Why Pubrica:
When you order our services, we promise you the following – Plagiarism free | always on Time | 24*7 customer support | Written to international Standard | Unlimited Revisions support | Medical writing Expert | Publication Support | Bio statistical experts | High-quality Subject Matter Experts.
Contact us:
Web: https://pubrica.com/
Blog: https://pubrica.com/academy/
Email: sales@pubrica.com
WhatsApp : +91 9884350006
United Kingdom: +44-1618186353
Discovering Abnormal Patches and Transformations of Diabetics Retinopathy in ...cscpconf
Diabetic retinopathy (DR) is one of the retinal diseases due to long-term effect of diabetes.
Early detection for diabetic retinopathy is crucial since timely treatment can prevent
progressive loss of vision. The most common diagnosis technique of diabetic retinopathy is to
screen abnormalities through retinal fundus images by clinicians. However, limited number of
well-trained clinicians increase the possibilities of misdiagnosing. In this work, we propose a
big-data-driven automatic computer-aided diagnosing (CAD) system for diabetic retinopathy
severity regression based on transfer learning, which starts from a deep convolutional neural
network pre-trained on generic images, and adapts it to large-scale DR datasets. From images
in the training set, we also automatically segment the abnormal patches with an occlusion test,
and model the transformations and deterioration process of DR. Our results can be widely used
for fast diagnosis of DR, medical education and public-level healthcare propagation.
DISCOVERING ABNORMAL PATCHES AND TRANSFORMATIONS OF DIABETICS RETINOPATHY IN ...csandit
Diabetic retinopathy (DR) is one of the retinal diseases due to long-term effect of diabetes.Early detection for diabetic retinopathy is crucial since timely treatment can prevent
progressive loss of vision. The most common diagnosis technique of diabetic retinopathy is to screen abnormalities through retinal fundus images by clinicians. However, limited number of well-trained clinicians increase the possibilities of misdiagnosing. In this work, we propose a big-data-driven automatic computer-aided diagnosing (CAD) system for diabetic retinopathy severity regression based on transfer learning, which starts from a deep convolutional neural
network pre-trained on generic images, and adapts it to large-scale DR datasets. From images in the training set, we also automatically segment the abnormal patches with an occlusion test,and model the transformations and deterioration process of DR. Our results can be widely used for fast diagnosis of DR, medical education and public-level healthcare propagation.
Gears are the most essential and commonly utilised power transmission components. It is really essential to operate machines involving different weights and speeds. When a load is increased beyond a certain limit, gear teeth frequently fail. Composite materials, in comparison to other metallic gears, offer significantly better mechanical qualities, such as a higher strength-to-weight ratio, increased hardness, and hence a lower risk of failure. Al6063 and SiC were employed to build a metal matrix composite for spur gear production in this work.
Sound plays a crucial part in every element of human life. Sound is a crucial component in the development of automated systems in a variety of domains, from personal security to essential monitoring. There are a few systems on the market now, but their efficiency is a worry for their use in real-world circumstances. Image classification and feature classification are the same as sound classification, just like other classification algorithms like machine learning.
Heaviness has been related to stroke, depression, and cancer are some of the most serious dangers to human existence. Heart disease, stroke, obesity, and type II diabetes are all disorders that have an impact on our way of life. Using data mining and machine learning approaches to forecast disease based on patient treatment history and health data has been a battle for decades.
javed_prethesis2608 on predcition of heart diseasejaved75
This document presents a thesis proposal for using deep neural networks to predict heart disease from patient data. The motivation is to improve prediction accuracy rates compared to previous machine learning models. The objectives are to explore patient data, preprocess the data, create a neural network model in TensorFlow, train the model, and predict heart disease. The methodology involves feature processing, feature selection, data preprocessing, creating a neural network, training the model, and predicting outcomes. The expected outcome is a prediction accuracy rate of over 75% using this deep learning approach.
This document discusses neural networks and their applications in signal processing. It begins with an introduction to neural networks and their biological inspiration. Applications of neural networks in signal processing are then outlined, including EEG, EMG, medical diagnosis, and more. The document reviews two papers on deep learning applications in biomedical fields and identifies challenges around model building and interpretability. It sets research objectives to optimize mathematical models and improve result interpretation. In conclusion, deep learning has advanced pattern recognition but challenges remain around data availability and model methodology.
Similar to LINKING EARLY DETECTION/TREATMENT OF PARKINSON’S DISEASE USING DEEP LEARNING TECHNIQUES (20)
ACEP Magazine edition 4th launched on 05.06.2024Rahul
This document provides information about the third edition of the magazine "Sthapatya" published by the Association of Civil Engineers (Practicing) Aurangabad. It includes messages from current and past presidents of ACEP, memories and photos from past ACEP events, information on life time achievement awards given by ACEP, and a technical article on concrete maintenance, repairs and strengthening. The document highlights activities of ACEP and provides a technical educational article for members.
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECTjpsjournal1
The rivalry between prominent international actors for dominance over Central Asia's hydrocarbon
reserves and the ancient silk trade route, along with China's diplomatic endeavours in the area, has been
referred to as the "New Great Game." This research centres on the power struggle, considering
geopolitical, geostrategic, and geoeconomic variables. Topics including trade, political hegemony, oil
politics, and conventional and nontraditional security are all explored and explained by the researcher.
Using Mackinder's Heartland, Spykman Rimland, and Hegemonic Stability theories, examines China's role
in Central Asia. This study adheres to the empirical epistemological method and has taken care of
objectivity. This study analyze primary and secondary research documents critically to elaborate role of
china’s geo economic outreach in central Asian countries and its future prospect. China is thriving in trade,
pipeline politics, and winning states, according to this study, thanks to important instruments like the
Shanghai Cooperation Organisation and the Belt and Road Economic Initiative. According to this study,
China is seeing significant success in commerce, pipeline politics, and gaining influence on other
governments. This success may be attributed to the effective utilisation of key tools such as the Shanghai
Cooperation Organisation and the Belt and Road Economic Initiative.
International Conference on NLP, Artificial Intelligence, Machine Learning an...gerogepatton
International Conference on NLP, Artificial Intelligence, Machine Learning and Applications (NLAIM 2024) offers a premier global platform for exchanging insights and findings in the theory, methodology, and applications of NLP, Artificial Intelligence, Machine Learning, and their applications. The conference seeks substantial contributions across all key domains of NLP, Artificial Intelligence, Machine Learning, and their practical applications, aiming to foster both theoretical advancements and real-world implementations. With a focus on facilitating collaboration between researchers and practitioners from academia and industry, the conference serves as a nexus for sharing the latest developments in the field.
Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapte...University of Maribor
Slides from talk presenting:
Aleš Zamuda: Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapter and Networking.
Presentation at IcETRAN 2024 session:
"Inter-Society Networking Panel GRSS/MTT-S/CIS
Panel Session: Promoting Connection and Cooperation"
IEEE Slovenia GRSS
IEEE Serbia and Montenegro MTT-S
IEEE Slovenia CIS
11TH INTERNATIONAL CONFERENCE ON ELECTRICAL, ELECTRONIC AND COMPUTING ENGINEERING
3-6 June 2024, Niš, Serbia
Batteries -Introduction – Types of Batteries – discharging and charging of battery - characteristics of battery –battery rating- various tests on battery- – Primary battery: silver button cell- Secondary battery :Ni-Cd battery-modern battery: lithium ion battery-maintenance of batteries-choices of batteries for electric vehicle applications.
Fuel Cells: Introduction- importance and classification of fuel cells - description, principle, components, applications of fuel cells: H2-O2 fuel cell, alkaline fuel cell, molten carbonate fuel cell and direct methanol fuel cells.
DEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODELgerogepatton
As digital technology becomes more deeply embedded in power systems, protecting the communication
networks of Smart Grids (SG) has emerged as a critical concern. Distributed Network Protocol 3 (DNP3)
represents a multi-tiered application layer protocol extensively utilized in Supervisory Control and Data
Acquisition (SCADA)-based smart grids to facilitate real-time data gathering and control functionalities.
Robust Intrusion Detection Systems (IDS) are necessary for early threat detection and mitigation because
of the interconnection of these networks, which makes them vulnerable to a variety of cyberattacks. To
solve this issue, this paper develops a hybrid Deep Learning (DL) model specifically designed for intrusion
detection in smart grids. The proposed approach is a combination of the Convolutional Neural Network
(CNN) and the Long-Short-Term Memory algorithms (LSTM). We employed a recent intrusion detection
dataset (DNP3), which focuses on unauthorized commands and Denial of Service (DoS) cyberattacks, to
train and test our model. The results of our experiments show that our CNN-LSTM method is much better
at finding smart grid intrusions than other deep learning algorithms used for classification. In addition,
our proposed approach improves accuracy, precision, recall, and F1 score, achieving a high detection
accuracy rate of 99.50%.
Redefining brain tumor segmentation: a cutting-edge convolutional neural netw...IJECEIAES
Medical image analysis has witnessed significant advancements with deep learning techniques. In the domain of brain tumor segmentation, the ability to
precisely delineate tumor boundaries from magnetic resonance imaging (MRI)
scans holds profound implications for diagnosis. This study presents an ensemble convolutional neural network (CNN) with transfer learning, integrating
the state-of-the-art Deeplabv3+ architecture with the ResNet18 backbone. The
model is rigorously trained and evaluated, exhibiting remarkable performance
metrics, including an impressive global accuracy of 99.286%, a high-class accuracy of 82.191%, a mean intersection over union (IoU) of 79.900%, a weighted
IoU of 98.620%, and a Boundary F1 (BF) score of 83.303%. Notably, a detailed comparative analysis with existing methods showcases the superiority of
our proposed model. These findings underscore the model’s competence in precise brain tumor localization, underscoring its potential to revolutionize medical
image analysis and enhance healthcare outcomes. This research paves the way
for future exploration and optimization of advanced CNN models in medical
imaging, emphasizing addressing false positives and resource efficiency.
A review on techniques and modelling methodologies used for checking electrom...nooriasukmaningtyas
The proper function of the integrated circuit (IC) in an inhibiting electromagnetic environment has always been a serious concern throughout the decades of revolution in the world of electronics, from disjunct devices to today’s integrated circuit technology, where billions of transistors are combined on a single chip. The automotive industry and smart vehicles in particular, are confronting design issues such as being prone to electromagnetic interference (EMI). Electronic control devices calculate incorrect outputs because of EMI and sensors give misleading values which can prove fatal in case of automotives. In this paper, the authors have non exhaustively tried to review research work concerned with the investigation of EMI in ICs and prediction of this EMI using various modelling methodologies and measurement setups.
A SYSTEMATIC RISK ASSESSMENT APPROACH FOR SECURING THE SMART IRRIGATION SYSTEMSIJNSA Journal
The smart irrigation system represents an innovative approach to optimize water usage in agricultural and landscaping practices. The integration of cutting-edge technologies, including sensors, actuators, and data analysis, empowers this system to provide accurate monitoring and control of irrigation processes by leveraging real-time environmental conditions. The main objective of a smart irrigation system is to optimize water efficiency, minimize expenses, and foster the adoption of sustainable water management methods. This paper conducts a systematic risk assessment by exploring the key components/assets and their functionalities in the smart irrigation system. The crucial role of sensors in gathering data on soil moisture, weather patterns, and plant well-being is emphasized in this system. These sensors enable intelligent decision-making in irrigation scheduling and water distribution, leading to enhanced water efficiency and sustainable water management practices. Actuators enable automated control of irrigation devices, ensuring precise and targeted water delivery to plants. Additionally, the paper addresses the potential threat and vulnerabilities associated with smart irrigation systems. It discusses limitations of the system, such as power constraints and computational capabilities, and calculates the potential security risks. The paper suggests possible risk treatment methods for effective secure system operation. In conclusion, the paper emphasizes the significant benefits of implementing smart irrigation systems, including improved water conservation, increased crop yield, and reduced environmental impact. Additionally, based on the security analysis conducted, the paper recommends the implementation of countermeasures and security approaches to address vulnerabilities and ensure the integrity and reliability of the system. By incorporating these measures, smart irrigation technology can revolutionize water management practices in agriculture, promoting sustainability, resource efficiency, and safeguarding against potential security threats.
A SYSTEMATIC RISK ASSESSMENT APPROACH FOR SECURING THE SMART IRRIGATION SYSTEMS
LINKING EARLY DETECTION/TREATMENT OF PARKINSON’S DISEASE USING DEEP LEARNING TECHNIQUES
1. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
DOI:10.5121/ijdkp.2022.12601 1
LINKING EARLY DETECTION/TREATMENT
OF PARKINSON’S DISEASE USING DEEP
LEARNING TECHNIQUES
Sarah Fan1
and Yu Sun2
1
Sage Hill School, 20402 Newport Coast Dr, Newport Beach, CA 92657
2
California State Polytechnic University, Pomona, CA, 91768, Irvine, CA 92620
ABSTRACT
Parkinson’s Disease (PD) is a progressive neurodegenerative disorder that causes uncontrollable
movements and difficulty with balance and coordination. It is highly important for early detection of
Parkinson’s Disease for patients to receive proper treatment. This paper aims to present a preliminary
data mining procedure that help Parkinson’s Disease patients slow down their progression of the disease
while helping early detection of the disease. For early non-invasive treatment, our research first analyses
the early symptoms of Parkinson’s Disease, designs/selects a proper demo video, let the user follow the
demo to exercise and upload his exercise video to our deep learning APP: LaBelle. LaBelle utilizing
MediaPipe Pose to identify, analyze, and store data about the poses and movements of both demo and the
user, calculates the angles created between different joints and major body parts. LaBelle’s AI model uses
a K-means clustering algorithm to create a group of clusters for both demo and the user dataset. Using the
two sets of clusters, LaBelle identifies the key frames in the user video and searches the demo cluster set
for a matching set of properties and frames. It evaluates the differences between the paired frames and
produces a final score as well as feedback on the poses that need improving. Meanwhile, if the user is
willing to donate their exercise data, he can simply input his age, whether he is a PD patient (maybe for
how long) anonymously. Then his data can be stored into our customized dataset, used in data mining for
Parkinson’s Disease prediction, which involves building/training our deep learning CNN model and help
early detection of Parkinson’s Disease.
KEYWORDS
Deep Learning, K-means Clustering, Computer Vision, Parkinson’s Disease, Data Mining.
1. INTRODUCTION
Parkinson’s Disease (PD) is a progressive disorder of the nervous system marked by tremors,
muscular rigidity, and slow, imprecise movement, chiefly affecting middle-aged and elderly
people [1]. It is associated with degeneration of the brain's basal ganglia and a deficiency of the
neurotransmitter dopamine. Worldwide, around 7-10 million people have Parkinson’s Disease
[2], making it highly important to diagnose PD accurately in the early stage so that patients can
receive proper treatment [3]. Parkinson’s disease (PD) is difficult to diagnose, particularly in its
early stages, because the symptoms of other neurologic disorders can be similar to those found in
PD. Meanwhile, early non-motor symptoms of PD may be mild and can be caused by many other
conditions. Therefore, these symptoms are often overlooked, making the diagnosis of PD at an
early stage more challenging [4]. To address these difficulties and refine the early detection of
PD, different neuroimaging techniques (such as magnetic resonance imaging (MRI), computed
tomography (CT) and positron emission tomography (PET)) and deep learning-based analysis
methods have been developed [5].
2. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
2
Data mining techniques[6] have taken a significant role in the early diagnosis and prognosis of
many health diseases, including Parkinson’s Disease. Data mining is the process of extracting
previously unknown knowledge from a large collection of data. The knowledge mined is often
referred to as patterns or models, such as clusters, classification rules, linear models, etc. Data
mining involves many stages including data collection, data processing, classifier build,
prediction etc. For all different stages work together in a cohesive way, it needs huge resource
and funds. The progressive in nature gives data mining for PD even more difficulties because all
the tasks are time-consuming.
Meanwhile, once the patient is detected with PD, early non-invasive exercises/treatment is very
helpful. Movement, especially exercises that encourage balance and reciprocal patterns
(movements that require coordination of both sides of the body), can slow down the progression
of the disease. Deep learning methods, especially in computer vision, are beginning to gain more
and more popularity as technology advances their ability to produce accurate models. Human
Pose Estimation (HPE)[7] addresses this particularly interesting area of deep learning research. It
has versatile real-world application in healthcare.
In this research, we are trying to link the early detection and treatment for PD using deep learning
techniques because these two stages can help each other. The early detection can let us provide
the uses with some proper exercises to slow down the progression of the disease while PD
patient’s exercise progress can help to build/train our AI model to make early detection more
precise.
The rest of the paper is organized as follows: Section 2 describes our research background,
direction, and the crucial elements needed in our research in detail; Section 3 presents our
approach to solve the problem and relevant details about the experiment we did; Section 4
presents the results and analysis; Section 5 gives a brief summary of other work that tackles a
similar problem; finally, Section 6 gives the conclusion remarks and discusses the future work of
this project.
2. CHALLENGES
Data mining is the process of analyzing large blocks of information to identify patterns, to extract
valuable information. The knowledge mined is often referred to as patterns or models, such as
clusters, classification rules, linear models, and time-trends. Figure 1 shows the steps of data
mining. Successful data mining takes all the resources, cooperation, and funds to work together
and takes many iterations. This is extremely difficult because not only so many different parties
are involved but also a huge amount of data is needed.
3. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
3
Figure 1. Steps of Data Mining (Source: http://www2.uregina.ca/).
2.1. Machine learning
Machine learning is a subset of data mining that encompasses supervised and unsupervised
learning techniques. It is the technology of developing systems that can learn and draw inferences
from patterns in data which can be applied to many different fields, from data analytics to
predictive analytics, from service personalization to natural language processing, and so on [8].
According to Shalev-Shwartz, S. et al. [9], Machine learning can be defined as “using the
experience to gain expertise.” The learning could be supervised learning, unsupervised learning,
etc. Supervised learning is the most common approach and is the approach we utilize in our
research.
Supervised learning algorithms try to model relationships between the target prediction output
and input features to predict output values for new data based on the relationships learned from
the prior data sets. This type of learning is normally related to classification tasks, which is the
process of teaching a classifier the relationship between the model’s input and output to use this
expertise later for un-seen input [10].
2.2. Deep Learning
Because machine learning is unable to meet the requirements due to the complexity of the
problems in certain areas, Deep Learning (DL) is gaining popularity due to its supremacy in
terms of accuracy. It is an advanced level of machine learning which includes a hierarchical
function that enables machines to process data with a nonlinear approach. The deep learning
networks are built with neuron nodes connected like the human brain and have many layers, each
layer receiving information from the previous layer, trained to perform the desired tasks, and then
passing on the information to the next layer [11].
knowledge
patterns
Data
Target
Data
Processed
Data
Transformed
Data
Selection
Pre-processing
Transformation
Data Mining
Interpretation/
evaluation
4. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
4
2.3. Convolutional Neural Network
A convolutional neural network (CNN) can be made up of many layers of models, where each
layer takes input from the previous layer, applies a filter to the data, and outputs it to the next
layer. CNNs run much faster on GPU, and the huge stockpiles of data that have been collected
can improve the accuracy of computer vision and NLP algorithms. A CNN consists of several
convolutional layers, with each layer including three major stages: convolution, nonlinear
activation (non linearity transform), and pooling (sub-sampling) [12].
2.4. Datasets
Datasets are fundamental in a deep learning system. An extensive and diverse dataset is a crucial
requirement for the successful training of a deep neural network. In our research, we explore
different CNNs using datasets we downloaded from HandPD [13].
3. SOLUTION
Figure 2 shows our solution. By study early “Parkinsonism”, also known as “Parkinsonian
Syndrome”, we plan demo exercises through data collection preparation. Then using our mobile
APP: LaBelle, user can mimic/learn basic movements that will help calm the shakiness/tremor,
relax the stiffness, mitigate the slow down movement and improve user’s ability to maintain
balance. By doing the exercises, the user can record the history scores of their exercise to track
their progress. Meanwhile, if they are willing to share their information anonymously, they can
input their age, whether they are Parkinson’s Disease patients, this information can be collected
and further used to build/update/train deep learning model to help early detection of Parkinson’s
Disease.
Figure 2. Our solution
5. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
5
3.1. Data Collection Planning
Parkinson’s Disease is a progressive and neurodegenerative disease. It involves the loss of
dopamine neuron, which make a little signalling chemical. When there is a reduction in
dopamine, we will see the many movement signs of Parkinson’s Disease.
MediaPipe developed by Google offers cross-platform, customizable machine learning solutions
for live and streaming media. It has a lot of real time detection features. These features are
examples of machine learning can be helpful in data collection. Table 1 shows some of early
symptoms of PD. Not all the symptoms, but sooner or later, one or more symptoms will appear.
For this research, we pick whole body related symptoms and use MediaPipe Pose as our main
real time detection solution.
Symptom Example Real Time Detection Demo Exercises
Shakiness Hard to pick up small
objects
MediaPipe Hands Hand exercises for PD
Tremor Hand tremor MediaPipe Hands Hand exercises for PD
Stiffness Difficult to bend
arm/leg
MediaPipe Pose Sit down, get up, pick
up plate from desk
Shuffling Steps Small step length MediaPipe Pose Turn around exercises
Stooped posture Difficult to have good
posture
MediaPipe Pose Posture Exercises
Slow movement Slow Down Movement MediaPipe Pose Straight Walking
Masked Faces diminished facial
expressivity
MediaPipe Face Mesh Facial Exercises
Asymmetric Arms swing differently
while walking
MediaPipe Pose Walking Exercises
Freeze Sudden stop while
walking
MediaPipe Pose Walking Exercises
Balance Trouble Difficult to keep
balance
MediaPipe Pose Walking Exercises
Table 1. Data Collection Preparation
3.2. A Deep Learning APP: LaBelle
Our solution to the data collection is a deep learning App: LaBelle. LaBelle gives one demo
exercise clip, takes in one of a user. LaBelle is structured into seven layers of processing. The
first layer (L1) consists of two Mediapipe Pose models (M1 and M2 ) to process the demo and
input video clip. The second layer (L2) consists of two blocks (B1 and B2) that calculate the
major angles for important joints in the skeleton-based demo and exerciser model. The third
layer(L3) consists of two blocks (C1 and C2) that group the angle-based properties into clusters
for both demo and APP user. The fourth layer (L4) consists of one block (D11) to detect the key
frames for APP user’s video clip. The fifth layer (L5) consists of one block (D12) to find the
centroids’ feature property for each cluster of user’s video clip based on the starting and ending
frame. The sixth layer (L6) has one block F2 to find the matching property in the demo video clip
for each centroid’s feature property generated from the user’s video. The last layer (L7) has one
block S12 to compare matching properties of demo and user, generating a score to send to the
user along with the frames corresponding to the centroids bearing lowest score to improve on.
6. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
6
Figure 3. Data Collection: LaBelle
3.2.1. MediaPipe Pose (L1 Layer: M1 and M2)
Mediapipe is a cross-platform library developed by Google. It provides high quality ready-to-use
Machine Learning solutions for computer vision tasks. Mediapipe Pose is one of the solutions we
use for high-accuracy body pose detection and tracking [15]. The Pose Landmark Model
(BlazePose GHUM 3D) allows us to predict the location of 33 pose landmarks.
Figure 4. MediaPipe Pose Landmarks [6]
3.2.2. Property Extraction: Angle Calculation (L2 Layer: B2 and B1)
Mediapipe Pose detects/tracks 33 landmarks, outputs 33 landmark connection. However, to
improve the performance, we define a set of key angles between two landmark connections. For
example, we use (∠16, 14 ,12) for the right elbow. We use (∠14, 12, 24) for the right shoulder.
We use (∠24, 26, 28) for the right knee. Same for the left side of the body. As shown in Figure 5
and Figure 6, we use Mediapipe’s built in libraries to plot the human’s pose on a 3D graph.
7. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
7
The coordinates of 3 adjacent joints are called into the plot_angle function of our algorithm,
which calculates and returns the angle between the three landmarks. This process is repeated until
all desired joint angles are returned. This process is carried out for each video frame (or every
other, depending on how many frames are present), resulting in a final tuple array consisting of
all major body angles for individual video frames.
Figure 5. MediaPipe code implementing key points into angle calculation
Figure 6. MediaPipe code to generate angles given key joints
3.2.3. K-means Clustering (L3 Layer: C2 and C1)
We limit our demo video length to about 3 minutes. For each second, we can generalize about 30
frames. That’s approximately 5400 frames in total. For the users’ video, 30 second long, that’s
approximately 900 frames. We want to cluster our large number of frames into similar groups
based on their properties, using our key angles as properties. The videos uploaded by the user
will change each time, so it’s impossible to preset cluster/centroid properties. To solve this
problem, we need an unsupervised and autonomous clustering algorithm.
K-means (Macqueen, 1967)[15] is widely used in unsupervised learning algorithms that tackle
clustering tasks. It uses “centroids”, different randomly initiated points in the data, and assigns
every data point to the nearest centroid (distance is dependent on similarity between data point
and centroid), Once every data point has been assigned, the centroid is adjusted to the average of
all the points assigned to it. Sklearn.cluster.KMeans works with KMeans using Lloyd’s or
Elkan’s algorithm.
8. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
8
Figure 7. K-means Clustering: unsupervised clustering algorithm
3.2.4. K-mean Hyper Parameter Tuning using Silhouette Score
Both video clips need to be processed using clustering. To determine the optimal number of
clusters for both video clips, we use silhouette scoring to tune the hyper parameters. One way we
can do this is by assigning a random number of clusters ranging from 3 to 30 together with
random initial centroid values. Alternatively, we can assign meaningful initial centroid values: by
studying demo exercise and movements.
3.2.5. Key Frame Identification (L4 and L5 Layer: D11 & D12)
Based on the clusters generated from the user’s video clip, each cluster will determine its own
key frame, then the middle frame is identified and used as key frame.
3.2.6. Compare and Generate Final Score (L6 and L7 Layer: F2 & C12)
To compare the two videos, we go through the label 11 of all the centroids in the user video clip
and find the modified label 12 by fitting the centroids within the demo clusters [16]. Then we
find of the most similar property from demo video clip, producing the most similar poses for both
user and demo, allowing us to calculate the score. Show in Figure 8 is part of the algorithm we
use to find the most similar video frame, given a predetermined video frame. We use matrix
norms from linear algebra to help us find the smallest distance (least difference between frames).
We conclude the final score and output the pose that generates the lowest score, which means
least matching, for user to improve in the future practice.
Figure 8. Portion of algorithm used to find the two most similar frames
9. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
9
3.3. Data Mining
The problem this research trying to solve can be summarized as the following
Figure 9. Data Mining CL: Convolutional Layer, FCL: Fully Connected Layer, PL: Pooling Layer, RL:
ReLU Layer
Our model consists of 2 convolutional layers, 2 max-pooling layers, 2 dropout layers, 2 dense
layers, and one flattened layer. All the activation functions are ReLU. Figure 10 shows our
model. We chose this particular CNN architecture since it gives good results [17] [25].
The dropout layer is a technique introduced by Srivastava et al. [20]. This layer aims to avoid
overfitting by randomly ignoring randomly some neurons from the previous layer. We inserted
the dropout layers to improve the performance of our model.
Figure 10. The Proposed CNN Architecture
The data Even though we published our deep learning APP LaBelle at Google Play, at this time,
we have the algorithm and flow ready, but we don’t have not collected enough data to build and
train a CNN model. Instead, to verify our methodology, we choose to use dataset from the open
library.
The dataset consists of hand drawn images (spiral/meander/wave) drawn by healthy people and
Parkinson’s disease patients. The model learns through training and uses a CNN to predict
whether the person who drew the image has Parkinson’s disease or not. The CNN model has one
convolutional layer in front, a fully connected layer at the end, and a variable number of
convolutional layers, max-pooling layers, and ReLU layers in between.
We downloaded HandPD dataset from [13]. The dataset contains 92 individuals, divided into 18
healthy people (Healthy Group) and 74 patients (Patients Group). Some examples are shown
below. The brief description is the following:
10. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
10
● Healthy Group: 6 male and 12 female individuals with ages ranging from 19 to 79 years old
(average age of 44.22±16.53 years). Among those individuals, 2 are left-handed and 16 are right-
handed.
● Patient Group: 59 male and 15 female individuals with ages ranging from 38 to 78 years old
(average age of 58.75±7.51 years). Among those individuals, 5 are left-handed and 69 are right-
handed.
Therefore, each spiral and meander dataset is labeled in two groups: the healthy group
containing 72 images, and the patient group containing 296 images. The images are labeled as
follows: ID_EXAM-ID_IMAGE.jpg, in which ID_EXAM stands for the exam’s identifier, and
ID_IMAGE denotes the number of the image of the exam.
Figure 11 Some examples of spirals extracted from the HandPD dataset [14]
Figure 11 shows (a) 58-year-old males (b) 28-year-old female individuals of the control group,
(c) 56-year-old males, and (d) 65 -year old female individuals of the patient group.
Figure 12. Some examples of meanders extracted from HandPD dataset [14]
Figure 12 shows (a) 58-years old male (b) 28-years old female individuals of a control group and
(c) 56-years old mail and (d) 65-years old female individuals of a patient group.
All the data in HandPD dataset is in *.jpg format. For the exploration, we did some pre-
processing including resizing, blurring, eroding, diluting, and color space converting
(cv2.cvtColor() method).
4. EXPERIMENT
4.1. Experiment For Data COLLECTION: Labelle
In this section we give some examples with numerical and real data sets to demonstrate the
performance of the proposed k-means algorithm. We show these unsupervised learning
behaviours to get the best number of clusters for our algorithm to matching the video between
demo and user. The dataset we are using is from the real data we collected through the training.
To find the best K to using in the production level we try to train the dataset for both 40 k-means
random state and 80 random states, run several k-means, increment k with each iteration, and
record the SSE.
11. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
11
Figure 13. SSE curve with N Cluster changes for tested with 40 and 80 random states
The diagram shows K bigger than 3 has the acceptable SSE control. The silhouette coefficient is
a measure of cluster cohesion and separation. It quantifies how well a data point fits into its
assigned cluster. instead of computing SSE, compute the silhouette coefficient, to determine the
best k for our algorithm:
Figure 14. Silhouette scores for k shows that the best k is 10 since it has the maximum score
Ground truth labels categorize data points into group. These types of metrics do their best to
suggest the correct number of clusters. To determine the best number of clusters, we fitted our
dataset and get the plot.
12. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
12
Figure 15. Ground truth labels for clusters 2, 4, 6, 8, 10
Ground truth labels also shows that cluster with 10 (orange points) has the best error value. The
comparison shows that clusters with 8 has the best result to matching the user video and demo
video.
4.2. Experiment for Data Mining
Figure 16 shows the training and validation accuracy and loss. The dataset we used is the spiral
dataset downloaded from HandPD [13] without pre-processing. The CNN we used is the one
shown in Figure 10. As we can see, it has a severe overfitting problem. To resolve this issue, we
added dropout [20] after max-pooling. Figure 17 shows the validation accuracy and loss after
dropout was added, preventing the model from overfitting and minimizing validation loss.
Figure 16. Pre-processing spiral data from HandPD using the model shown in Figure 10
13. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
13
Figure 17. Spiral data from HandPD with two dropout layers added after max-pooling
To see the effects of different drawing patterns, we used two different patterns from the same
dataset with the same CNN, with dropout added after max-pooling. Figure 17 uses spiral data
from HandPD while Figure 18 uses meander data from HandPD. Comparing Figure 17 and
Figure 18 we can see that both patterns generate similar validation accuracy and validation loss
results, with the spiral slightly more accurate.
Figure 18. Meander data from HandPD with two dropout layers added after max-pooling
5. RELATED WORK
Several researchers have worked on the K-means algorithm. In the paper “Unsupervised K-means
clustering algorithm.", Sinaga, Kristina P., and Miin-Shen Yang construct an unsupervised
learning schema for the k-means algorithm so that it is free of initialization without parameter
selection [21]. They also simultaneously find an optimal number of clusters. In the paper The k-
means Algorithm: A Comprehensive Survey and Performance Evaluation, Syed Islam explain the
advance of k-means algorithm and discuss the using of comparison among different k-means
clustering algorithms.
In the paper "Real-time comparison of movement of Bulgarian folk dances.", Uzunova, Zlatka.
explain a method recording the dancing with two Kinetic sensors and analyze the data with
machine learning algorithm. To understand how to applying and picking the data, we also check
the research paper related to the machine learning filed, In the paper " Machine learning
14. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
14
algorithms-a review” Mahesh, Batta published, he explained how different machine learning
algorithm applying in data analyzing.
Several researchers have worked on the diagnosis of Parkinson’s Disease by using machine
learning methods, e.g. diagnosis using voice, diagnosis using brain scan images, diagnosis
drawings such as meander patterns, spirals, waves, etc.
J. Mei et al. [22] did a review of literature on machine learning for the diagnosis of Parkinson’s
disease, using sound, MRI images, and hand-drawn images. It searches IEEE Xplore and
PubMed. It reviewed research articles published from the year 2009 onwards and summarized
data sources and sample size.
The public repositories and databases include HandPD [13], Kaggle dataset [23], the University
of California at Irvine (UCI) Machine Learning Repository [24], Parkinson’s Progression
Markers Initiative (PPMI) [25], PhysioNet [26], etc.
Quite a few researchers use magnetic resonance images (MRI) or their variations as their research
dataset. Noor et al. [5] surveyed the application of deep learning in detecting neurological
disorders from magnetic resonance images (MRI) in the detection of Parkinson’s disease,
Alzheimer’s disease, and schizophrenia.
E. Huseyn et al. [27] [28] used MRI images as their dataset. S. Chakraborty [29] and X. Zhang
[30] has used a dataset from Parkinson’s Progression Markers Initiative (PPMI). Z.Cai et al. [31]
used an enhanced fuzzy k-nearest neighbor (FKNN) method for the early detection of
Parkinson’s Disease based on vocal measurements. L. Badea et al. [32] explored the
reproducibility of functional connectivity alterations in Parkinson’s Disease based on resting-
state fMRI scans images.
Pereira et al. did a series of research on automatic detecting Classify Parkinson’s disease for
many years. At first, they used non-deep learning algorithms in diagnosing PD [33]. They
collected/constructed a public dataset called “HandPD” [13]. Based on this dataset, they
compared the efficiency of different hand drawn patterns in the diagnosis of PD [11]. Their
results show that the meander pattern generates more accurate results compared to the spiral
pattern. However, in our research, the spiral and meander patterns generate similar results when
they are trained and tested through the same CNN.
Later, they explored the use of CNN on the images extracted from time-series signals and used
three different CNN architectures, ImageNet, CIFAR-10, and LeNet as baseline approach [18].
In her master project, M. Alissa [19] used non-public datasets (spiral pentagon dataset) to
evaluate the efficiency of two different neural networks (Recursive Neural Networks(RNN) and
Convolutional Neural Networks (CNN)). We built a CNN similar to hers and used the dataset
from HandPD [13] and Kaggle [23] to evaluate different CNNs and different datasets.
Gil-Martin et al. [34] presented a method to detect Parkinson’s Disease from drawing movements
using Convolutional Neural Networks. He used the dataset from the UCI machine learning
repository as input data, applied signal-processing (sampling with 100Hz and 140 Hz, resampling
with 110 Hz, perform Hamming windowing and FFT ) to generate pre-processed data, and used
this data to train/validate the CNNs.
M.E. Isenkul et al. [35] designed an improved spiral test dataset using a digitized graphics tablet
for monitoring Parkinson’s Disease. Digitized graphics have more information, including
15. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
15
timestamps, grip angles, and hand pressure, etc. The significance of that can be investigated in
future work.
P.Zham [36] presented a dataset at Kaggle [23] with waves and spirals. He used a composite
index of speed and pen-pressure to distinguish different stages of Parkinson’s Disease.
6. CONCLUSIONS
In this paper, we present a preliminary data mining procedure that help Parkinson’s Disease
patients slow down their progression of the disease while helping early detection of the disease.
By linking the early detection and treatment together, the two stages can help each other, provide
information back and forth, enlarge our database, improve the accuracy of deep learning models
because deep learning is all about data.
Future work can be cooperated with therapists, provide their patients free APP and ask those
patients to donate their exercise video clips.
REFERENCES
[1] M. Bhat, "Parkinson's Disease prediction based on hand tremor analysis," IEEE 2017 International
Conference, 2017.
[2] "Parkinson's Disease Statistics," Parkinson's News Today, 2021. [Online].
Available: https://parkinsonsnewstoday.com/parkinsons-disease-statistics/.
[3] Editorial Team, "Diagnosis-Early Symptoms & Early Diagnosis," parkinsonsdisease.net, 8 March
2017. [Online]. Available: https://parkinsonsdisease.net/diagnosis/early-symptoms-signs.
[4] Blog post, "You could have Parkinson’s disease symptoms in your 30s or 40s and not know it,"
health direct, 11 April 2019. [Online]. Available: https://www.healthdirect.gov.au/blog/parkinsons-
disease-symptoms-in-your-30s-40s. [Accessed 3 September 2022].
[5] M. Noor, "Application of deep learning in detecting neurological disorders from magnetic resonance
images: a survey on the detection of Alzheimer’s disease, Parkinson’s disease and schizophrenia,"
Brain Informatics, 9th October 2020.
[6] AK.Kumar Srivastava, "A systematic review on data mining application n Parkinson’s disease,"
Neuroscience Informatics, Dec. 2022.
[7] Andriluka, Mykhaylo, et al. "2d human pose estimation: New benchmark and state of the art
analysis." Proceedings of the IEEE Conference on computer Vision and Pattern Recognition. 2014.
[8] T. blog, "Guide to machine learning applications: 7 Major Fields," The APP solution, 2021. [Online].
Available: https://theappsolutions.com/blog/development/machine-learning-applications-guide/.
[9] S. Shalev-Shwartz, "Understanding machine learning: From theory to algorithms," Cambridge
University Press, 2014.
[10] I. H. Witten, "Data Mining: Practical Machine Learning Tools and Techniques, Third Edition,"
Morgan Kaufmann Series in Data Management Systems, 2011.
[11] Trending Blog, "What is Deep Learning? and What are its Significance Deep Learning Trends,"
ALPHA Information Systems INDIA PVT LTD, 15 September 2019. [Online].
Available: https://www.aalpha.net/blog/what-is-deep-learning-and-what-are-its-significance/.
[12] H. Wang, "On the origin of deep learning," arXiv preprint arXiv, 2017.
[13] C. Pereira, "Welcome to the HandPD dataset," HandPD, 2017. [Online].
Available: https://wwwp.fc.unesp.br/~papa/pub/datasets/Handpd/.
[14] Zhang, Fan, et al. "MediaPipe hands: On-device real-time hand tracking." arXiv preprint
arXiv:2006.10214 (2020).
[15] Likas, Aristidis, Nikos Vlassis, and Jakob J. Verbeek. "The global k-means clustering algorithm."
Pattern recognition 36.2 (2003): 451-461.
[16] Yuan, Fang, et al. "A new algorithm to get the initial centroids." Proceedings of 2004 International
Conference on Machine Learning and Cybernetics (IEEE Cat. No. 04EX826). Vol. 2. IEEE, 2004.
[17] C. Pereira, "Deep Learning-Aided Parkinson's Disease Diagnosis from Handwritten Dynamics.,"
Graphics. Patterns and Images (SIBGRAPI) 2016 29th SIBGRAPI Conference IEEE, 2016.
16. International Journal of Data Mining & Knowledge Management Process (IJDKP),
Vol.12, No.5/6, November 2022
16
[18] C. Pereira, "Convolutional neural networks applied for Parkinson's disease identification," Machine
Learning for Health Informatics, 2017.
[19] M. Alisssa, "Master's Project: Parkinson's Disease Diagnosis Using Deep Learning," Heriot Watt
University, 2018.
[20] N. Srivastava, "Dropout: A Simple Way to Prevent Neural Networks from Overfitting.," The Journal
of Machine Learning Research., 2014.
[21] Sinaga, Kristina P., and Miin-Shen Yang. "Unsupervised K-means clustering algorithm." IEEE
access 8 (2020): 80716-80727.
[22] J. Mei, "Machine Learning for the Diagnosis of Parkinson’s disease: a review of literature," Frontiers
in Aging Neuroscience, 2021.
[23] K. Mader, "Parkinson's Drawings," Kaggle.com, 2019. [Online].
Available: https://www.kaggle.com/kmader/parkinsons-drawings.
[24] D. Graff, "UCI Machine Learning Repository," University of California, Irvine.
[25] K. Marek, "The Parkinson Progression Marker Initiative (PPMI)," Progress Neurobiol, 2011.
[26] A. L. Goldberger, "PhysioBank, Physio Toolkit and PhysioNet: Components of a new research
resource for complex physiologic signals," Circulation, 2000.
[27] E. Huseyn, "Deep Learning Based Early Diagnostis of Parkinsons Disease," Cornell University
arXiv.org, Auguest 2020.
[28] S. Shinde, "Predictive markers for Parkinson's disease using deep neural nets on neuromelanin
sensitive MRI," bioRxiv the preprint server for biology, 2019.
[29] S. Chakraborty, "Detection of Parkinson's Disease from 3T T1 weighted MRI scans using 3D
convolutional neural network," diagnostics (MDPI), 2020.
[30] X. Zhang, "Multi-View graph convolutional network and its applications on neuroimage analysis for
Parkinson's disease," Amia annual symposium proceedings archive, 2018.
[31] Z. Cai, "An Intelligent Parkinson's Disease Diagnostic," Hindawi.com, Computational and
Mathematical Methods in Medicine, 2018.
[32] L. Badea, "Exploring the reproducibility of functional connectivity alterations in Parkinson’s
disease," National Institute for Research and Development in Informatics, Bucharest, Romania.
[33] C. Pereira, "A step towards the automated diagnosis of parkinson's disease: Analyzing handwriting
movements," in 28th International Symposium on Computer-Based Medical Systems (Sao Carlos:
IEEE), 2015.
[34] M. Gil-Martin, "Parkinson's Disease Detection from Drawing movements using convolutional neural
networks," in Electronics (MDPI), 2019.
[35] M Isenkul, "Improved Spiral Test using a Digitized Graphics Tablet for Monitoring Parkinson's
Disease," in International Conference on e-Health and Telemedicine, 2014.
[36] P. Zham, "Distinguishing Different Stages of Parkinson's Disease Using Composite Index of Speed
and Pen-Pressure of Sketching a Spiral," in Frontiers in Neurology, 2017.