SlideShare a Scribd company logo
1 of 4
Download to read offline
Clinical Research in Diabetes and Endocrinology  •  Vol 3  •  Issue 2  •  2020 16
INTRODUCTION
I
n2017,estimatesofdiabeteshaveshowntheexistenceof425
million diabetics in the world, and this number is expected to
reach to 625 million by 2045 (IDF, 2017; Li et al., 2019).[1-3]
Diabetes mellitus is a category of endocrine disorders
associated with impaired glucose absorption that occurs as a
consequence of the “Insulin” hormone’s absolute or relative
insufficiency. Condition is characterized by a chronic path
and a violation of all forms of metabolism.[4]
Diabetes is usually divided into four categories: Type 1
diabetes (T1D), type 2 diabetes, gestational diabetes mellitus,
and, due to other factors, particular types of diabetes. T1D and
T2D are the two most common forms of the condition (T2D).
The former is caused by the degradation of the beta cells of the
pancreas, which results in insulin deficiency, whereas the latter
is caused by the inadequate transport of insulin to the cells.[5]
Life-threatening complications such as strokes, heart attacks,
chronic renal failure, diabetic foot syndrome, antipathy,
neuropathy, encephalopathy, hyperthyroidism, tumors of the
adrenal gland, liver cirrhosis, glucagonoma, and transient
hyperglycemia and many other complications can contribute
to both forms of disease.[4]
For all people who are predisposed
to diabetes, the prediction and early detection of diabetes are
therefore important. Using artificial intelligence techniques,
multiple diseases can currently be diagnosed, and deep neural
networks have achieved the best results in classification
problems (Miotto etal., 2017; Liu et al., 2019).[3,6,7]
Data mining was used to predict diabetes.[5,8,9]
Ali et al. used
neural network analytics to build a model for classifying
diabetic people based on past medical history and patient
control level. The author developed a new predictive model for
data mining methods that would identify the level of control of
diabetic patients centered on medical reports that are historical.
The analysis was carried out with the use of three techniques
RESEARCH ARTICLE
Prediction of Risk Factors Leading to Diabetes Using
Neural Network Analysis
Ahed J. Alkhatib1,2
, Amer Mahmoud Sindiani3
, Eman Hussein Alshdaifat4
1
Department of Legal Medicine, Toxicology and Forensic Medicine, Jordan University of Science and Technology,
Jordan,2
International Mariinskaya Academy, Department of Medicine and Critical Care, Department of
Philosophy, Academician Secretary of Department of Sociology, Jordan, 3
Department of Obstetrics and
Gynecology, Faculty of Medicine, Jordan University of Science and Technology, Jordan, 4
Department of
Obstetrics and Gynecology, Faculty of medicine, Yarmouk University, Jordan
ABSTRACT
This study presents a prediction of diabetes type 2 using neural network analysis. A dataset of diabetes posted in Kaggle
was used for analysis. The dataset included one dependent variable, outcome of disease, with two numbers, “0” means
no disease, and “1” means disease. There were other eight risk factors for diabetes prediction, glucose, body mass index
(BMI), insulin, skin thickness, no. of pregnancies, diabetes pedigree function, and age. We constructed the model using
SPSS version 21. The results showed that the model prediction was 78.3% for training and 76.9% for testing. The model
showed that the most significant predictors of diabetes were: glucose, BMI, diabetes pedigree function, no. of pregnancies,
age, blood pressure, insulin, and skin thickness. Taken together, this model was effective and leads to a good prediction rate.
Key words: Covariates, dependent variable, diabetes type 2, neural network analysis, outcome
Address for correspondence:
Ahed J. Alkhatib, Department of Legal Medicine, Toxicology and Forensic Medicine, Jordan University of Science and
Technology, Jordan. Mobile: 00962795905145.
https://doi.org/10.33309/2639-832X.030204 www.asclepiusopen.com
© 2020 The Author(s). This open access article is distributed under a Creative Commons Attribution (CC-BY) 4.0 license.
Alkhatib: Relative importance of risk factors in diabetes type 2
 Clinical Research in Diabetes and Endocrinology  •  Vol 3  •  Issue 2  •  2020
of data processing, which are Logistic, Naïve Bayes, and J48.
The analysis was carried out using the WEKA program. The
outcome showed that logistics data mining algorithm gave
an average accuracy of 0.73, a recall of 0.744, a metric of
0.653, and a precision of 74.4%. Naïve Bayes gave an average
accuracy of 0.717, 0.742 recall, 0.653 F-measure, and 74.2%
precision. J48 has given an average accuracy of 0.54, recall
of 0.735, F-measure of 0.623, and precision of 0.54 and 73.5
percentage points. This showed that the logistic algorithm was
more precise than the other two algorithms.
Study objectives
The main objective of this study was to identify the risk
factors associated with diabetes and to determine their
relative importance.
METHODOLOGY
The present study was based on a dataset posted in Kaggle.[3]
The dataset of diabetes taken from India was analyzed in this
study using a neural network.The data consisted of 768 females.
Study variables
The dataset involved one dependent variable, outcome
indicating having diabetes (1) or no diabetes (0). There
are 8 covariates or independent variable: Glucose, insulin,
age, blood pressure, skin thickness, no. of pregnancies, and
diabetes pedigree function.
Statistical analysis
Neural network analytics of data was carried out using SPSS
version 21.
RESULTS
As demonstrated in Table 1, the study sample included 768
cases, 526 (68.5%) in training section, and 242 (31.5%) in
testing section. All cases were included in the study.
Network information
As demonstrated in Table 2, network information included
the following parameters: Input layer with 8 covariates:
Pregnancies number, glucose, blood pressure, skin thickness,
insulin, body mass index (BMI), diabetes pedigree function,
and age. Number of units is 8 and rescaling method for
covariates is standardized. The second parameter was about
the hidden layer (s), there was one hidden layer with five units
in hidden layer, and the activation function was hyperbolic
tangent. The third parameter was the output layer. There was
one variable, the output. It has 2 number of units. Activation
function was softmax and the error function was cross-entropy.
Model summary
Weconstructedthemodelthathadthefollowingcharacteristics
[Table 3]: For training, cross-entropy error was computed as
Table 2: Network information
Input
layer
Covariates 1 Pregnancies No.
2 Glucose
3 Blood pressure
4 Skin thickness
5 Insulin
6 BMI
7 Diabetes
pedigree function
8 Age
Number of unitsa
8
Rescaling method for
covariates
Standardized
Hidden
layer(s)
Number of hidden layers 1
Number of units in hidden
layer 1a
5
Activation function Hyperbolic
tangent
Output
layer
Dependent variables 1 outcome
Number of units 2
Activation function Soft max
Error function Cross-entropy
BMI: Body mass index
Table 3: Model summary
Training ???
Cross-entropy error 241.342
Percent incorrect predictions 21.7
Stopping rule used 1 consecutive step(s)
with no decrease in errora
Training time 0:00:00.16
Testing
Cross-entropy error 111.363
Percent incorrect predictions 23.1
Table 1: Case processing summary
n Percent
Sample
Training 526 68.5
Testing 242 31.5
Valid 768 100.0
Excluded 0
Total 768
Alkhatib: Relative importance of risk factors in diabetes type 2
Clinical Research in Diabetes and Endocrinology  •  Vol 3  •  Issue 2  •  2020 18
241.342, percent incorrect prediction 21.7%, stopping rules
used were 1 consecutive step(s) with no decrease in errora
,
and training time was 0:000:00.16. For testing, cross-entropy
error was 111.363 and percent incorrect prediction was
23.1%. Dependent variable was outcome.
Classification
As seen in Table 4, in this study, the outcome variable
implied two values, “0”: Non-diabetic, and “1”: Diabetic.
In the training part, for 0 output, the system determined 345
cases, 47 cases were classified by mistake in “0” output, but
the system put them in “1” output. Percent correction was
computed as 86.4%. In “1” output, 181 cases were included,
among which 67 cases were predicted to be in “0” output.
Percent correction was computed as 63%. The overall percent
was 78.3%. In the testing part, 155 cases were included in the
output “0,” among which 23 cases were predicted to have the
disease. The percent correction was 85.2%. For the output
“1”, 87 cases were included, of which 33 cases were predicted
to be in the output “0.” Percent correction was found to be
62.1%. The overall percent was 76.9%.
Independent variable importance
As seen in Table 5 and Figure 1, the importance of glucose
level was 0.276 (100%), BMI 0.244 (88.4%), diabetes
pedigree function 0.145 (52.6%), no. of pregnancy 0.101
(36.5%), age 0.08 (29.1%), blood pressure 0.071 (25.7%),
insulin 0.052 (18.8%), and skin thickness 0.030 (10.7%).
DISCUSSION
The results of this study identified diabetes risk factors among
females who had diabetes type 2. Using neural network
analysis usually uses technical language that is not easily
understood by readers. The problem of these studies is being
made by non-medical professional, including those working
in computer engineering or software engineers. According
to this context, the author integrated technical and medical
languages to help the reader to be able understand the topic
of diabetes.
According to our model, covariates or independent variables,
glucose level was the first important predictor of diabetes
type 2 among females, and its importance 100%. BMI was
able to predict 88.4% of diabetes cases [Figure 1]. The
importance of such a figure is that it pointed to points that
we need to control. Another important point is the number
Table 4: Classification
Sample Observed Predicted
0.00 1.00 Percent correct
Training 0.00 298 47 86.4
1.00 67 114 63.0
Overall percent 69.4 30.6 78.3
Testing 0.00 132 23 85.2
1.00 33 54 62.1
Overall percent 68.2 31.8 76.9
Dependent variable: Outcome
Figure 1: Normalized importance of covariates
Alkhatib: Relative importance of risk factors in diabetes type 2
 Clinical Research in Diabetes and Endocrinology  •  Vol 3  •  Issue 2  •  2020
of pregnancies that predicted about 37% of diabetic cases
among females.
CONCLUSIONS
The results of the present study pointed to two important
considerations, the use of neutral network analytics is
effective in the prediction of diabetes and identifying risk
factors with their relative importance.
REFERENCES
1.	 Zhou H, Myrzashova R, Zheng R. Diabetes prediction model
based on an enhanced deep neural network. EURASIP J Wirel
Commun Netw 2020;148:1621.
2.	 Li G, Peng S, Wang C, Niu J, Yuan Y. An energy-efficient data
collection scheme using denoising autoencoder in wireless
sensor networks. Tsinghua Sci Technol 2019;24:86-96.
3.	 Liu H, Kou H, Yan C, Qi L. Link prediction in paper citation
network to construct paper correlated graph. EURASIP J Wirel
Commun Netw 2019;233:2352.
4.	 Available from: https://www.kaggle.com/uciml/pima-indians-
diabetes-database. [Last accessed on 2020 Dec 10].
5.	 American Diabetes Association. Classification and diagnosis
of diabetes: Standards of medical care in diabetes-2018.
Diabetes Care 2018;41:13-27.
6.	 Ali R, Siddiqi MH, Idris M, Kang BH, Lee S. Prediction of
diabetes mellitus based on boosting ensemble modeling.
In: International Conference on Ubiquitous Computing and
Ambient Intelligence. Berlin, Germany: Springer International
Publishing; 2014. p. 25-8.
7.	 Miotto R, Wang F, Wang S, Jiang X, Dudley T. Deep learning
for healthcare: Review, opportunities and challenges. Brief
Bioinform 2017;19:1236-46.
8.	 International Diabetes Federation. IDF Diabetes Atlas. 8th
ed. Brussels, Belgium: International Diabetes Federation;
2017.
9.	 El_Jerjawi NS, Abu-Naser SS. Diabetes prediction
using artificial neural network. Int J Adv Sci Technol
2018;121:55-64.
How to cite this article: Alkhatib AJ, Sindiani AM,
Alshdaifat EH. Prediction of Risk Factors Leading to
Diabetes Using Neural Network Analysis. Clin Res
Diabetes Endocrinol 2020;3(2):16-19.
Table 5: Independent variable importance
Independent variable Importance Normalized
importance %
No. of pregnancies 0.101 36.5
Glucose level 0.276 100.0
Blood pressure 0.071 25.7
Skin thickness 0.030 10.7
Insulin 0.052 18.8
BMI 0.244 88.4
Diabetes pedigree function 0.145 52.6
Age 0.080 29.1
BMI: Body mass index

More Related Content

What's hot

PREDICTIVE MODEL FOR LIKELIHOOD OF DETECTING CHRONIC KIDNEY FAILURE AND DISEA...
PREDICTIVE MODEL FOR LIKELIHOOD OF DETECTING CHRONIC KIDNEY FAILURE AND DISEA...PREDICTIVE MODEL FOR LIKELIHOOD OF DETECTING CHRONIC KIDNEY FAILURE AND DISEA...
PREDICTIVE MODEL FOR LIKELIHOOD OF DETECTING CHRONIC KIDNEY FAILURE AND DISEA...
ijcsitcejournal
 
Total and Cause-Specific Mortality of U.S. Nurses Working Rotating Night Shifts
Total and Cause-Specific Mortality of U.S. Nurses Working Rotating Night ShiftsTotal and Cause-Specific Mortality of U.S. Nurses Working Rotating Night Shifts
Total and Cause-Specific Mortality of U.S. Nurses Working Rotating Night Shifts
Emergency Live
 
Ngai india Stampede 2001-2010 Final Presentation
Ngai india Stampede 2001-2010 Final PresentationNgai india Stampede 2001-2010 Final Presentation
Ngai india Stampede 2001-2010 Final Presentation
Gordon Ngai
 
VCS_StatisticsReport_2013
VCS_StatisticsReport_2013VCS_StatisticsReport_2013
VCS_StatisticsReport_2013
Jennifer Nguyen
 
To Assess the Severity and Mortality among Covid 19 Patients after Having Vac...
To Assess the Severity and Mortality among Covid 19 Patients after Having Vac...To Assess the Severity and Mortality among Covid 19 Patients after Having Vac...
To Assess the Severity and Mortality among Covid 19 Patients after Having Vac...
YogeshIJTSRD
 
The Relation of Obesity and Chronic Diseases among Home Health Care Patients
The Relation of Obesity and Chronic Diseases among Home Health Care PatientsThe Relation of Obesity and Chronic Diseases among Home Health Care Patients
The Relation of Obesity and Chronic Diseases among Home Health Care Patients
ijtsrd
 

What's hot (20)

Deep learning-approach
Deep learning-approachDeep learning-approach
Deep learning-approach
 
E0342023026
E0342023026E0342023026
E0342023026
 
Diabetes tipo 2 insulina 2013
Diabetes tipo 2 insulina 2013Diabetes tipo 2 insulina 2013
Diabetes tipo 2 insulina 2013
 
Comparative evaluation of the effects of artemisinin based combination therap...
Comparative evaluation of the effects of artemisinin based combination therap...Comparative evaluation of the effects of artemisinin based combination therap...
Comparative evaluation of the effects of artemisinin based combination therap...
 
Anti diabetic drugs in patients with diabetes
Anti diabetic drugs in patients with diabetesAnti diabetic drugs in patients with diabetes
Anti diabetic drugs in patients with diabetes
 
Kwon et.al
Kwon et.alKwon et.al
Kwon et.al
 
Performance Analysis of Data Mining Methods for Sexually Transmitted Disease ...
Performance Analysis of Data Mining Methods for Sexually Transmitted Disease ...Performance Analysis of Data Mining Methods for Sexually Transmitted Disease ...
Performance Analysis of Data Mining Methods for Sexually Transmitted Disease ...
 
PREDICTIVE MODEL FOR LIKELIHOOD OF DETECTING CHRONIC KIDNEY FAILURE AND DISEA...
PREDICTIVE MODEL FOR LIKELIHOOD OF DETECTING CHRONIC KIDNEY FAILURE AND DISEA...PREDICTIVE MODEL FOR LIKELIHOOD OF DETECTING CHRONIC KIDNEY FAILURE AND DISEA...
PREDICTIVE MODEL FOR LIKELIHOOD OF DETECTING CHRONIC KIDNEY FAILURE AND DISEA...
 
Estimating the Survival Function of HIV AIDS Patients using Weibull Model
Estimating the Survival Function of HIV AIDS Patients using Weibull ModelEstimating the Survival Function of HIV AIDS Patients using Weibull Model
Estimating the Survival Function of HIV AIDS Patients using Weibull Model
 
Total and Cause-Specific Mortality of U.S. Nurses Working Rotating Night Shifts
Total and Cause-Specific Mortality of U.S. Nurses Working Rotating Night ShiftsTotal and Cause-Specific Mortality of U.S. Nurses Working Rotating Night Shifts
Total and Cause-Specific Mortality of U.S. Nurses Working Rotating Night Shifts
 
Mathematical Model for Drug Therapy in Patients With Diabetes Mellitus
Mathematical Model for Drug Therapy in Patients With Diabetes MellitusMathematical Model for Drug Therapy in Patients With Diabetes Mellitus
Mathematical Model for Drug Therapy in Patients With Diabetes Mellitus
 
Ngai india Stampede 2001-2010 Final Presentation
Ngai india Stampede 2001-2010 Final PresentationNgai india Stampede 2001-2010 Final Presentation
Ngai india Stampede 2001-2010 Final Presentation
 
NEET PG Cutoff 2018 | CareerXpert
NEET PG Cutoff 2018 | CareerXpertNEET PG Cutoff 2018 | CareerXpert
NEET PG Cutoff 2018 | CareerXpert
 
VCS_StatisticsReport_2013
VCS_StatisticsReport_2013VCS_StatisticsReport_2013
VCS_StatisticsReport_2013
 
The relationship between gastrin hormone level and microalbumin urea in patie...
The relationship between gastrin hormone level and microalbumin urea in patie...The relationship between gastrin hormone level and microalbumin urea in patie...
The relationship between gastrin hormone level and microalbumin urea in patie...
 
To Assess the Severity and Mortality among Covid 19 Patients after Having Vac...
To Assess the Severity and Mortality among Covid 19 Patients after Having Vac...To Assess the Severity and Mortality among Covid 19 Patients after Having Vac...
To Assess the Severity and Mortality among Covid 19 Patients after Having Vac...
 
A Study on the Presence of Islets Cell Autoantibodies in NonInsulin Requiring...
A Study on the Presence of Islets Cell Autoantibodies in NonInsulin Requiring...A Study on the Presence of Islets Cell Autoantibodies in NonInsulin Requiring...
A Study on the Presence of Islets Cell Autoantibodies in NonInsulin Requiring...
 
Nutrition status of dm patients in medicine clinic
Nutrition status of dm patients in medicine clinicNutrition status of dm patients in medicine clinic
Nutrition status of dm patients in medicine clinic
 
Inappropriate drug use in hospitalized elderly patients of medicine and cardi...
Inappropriate drug use in hospitalized elderly patients of medicine and cardi...Inappropriate drug use in hospitalized elderly patients of medicine and cardi...
Inappropriate drug use in hospitalized elderly patients of medicine and cardi...
 
The Relation of Obesity and Chronic Diseases among Home Health Care Patients
The Relation of Obesity and Chronic Diseases among Home Health Care PatientsThe Relation of Obesity and Chronic Diseases among Home Health Care Patients
The Relation of Obesity and Chronic Diseases among Home Health Care Patients
 

Similar to Prediction of Risk Factors Leading to Diabetes Using Neural Network Analysis

K-Nearest Neighbours based diagnosis of hyperglycemia
K-Nearest Neighbours based diagnosis of hyperglycemiaK-Nearest Neighbours based diagnosis of hyperglycemia
K-Nearest Neighbours based diagnosis of hyperglycemia
ijtsrd
 
Holistic Forecasting of Onset of Diabetes through Data Mining Techniques
Holistic Forecasting of Onset of Diabetes through Data Mining TechniquesHolistic Forecasting of Onset of Diabetes through Data Mining Techniques
Holistic Forecasting of Onset of Diabetes through Data Mining Techniques
ijcnes
 
A study of modifiable and non-modifiable risk factors associated with diabeti...
A study of modifiable and non-modifiable risk factors associated with diabeti...A study of modifiable and non-modifiable risk factors associated with diabeti...
A study of modifiable and non-modifiable risk factors associated with diabeti...
International Journal of Medical Sciences and Nursing Research
 
Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral...
Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral...Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral...
Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral...
ijtsrd
 
A comparative analysis of biochemical and hematological parameters in diabeti...
A comparative analysis of biochemical and hematological parameters in diabeti...A comparative analysis of biochemical and hematological parameters in diabeti...
A comparative analysis of biochemical and hematological parameters in diabeti...
amsjournal
 

Similar to Prediction of Risk Factors Leading to Diabetes Using Neural Network Analysis (20)

Diabetes Prediction Using ML
Diabetes Prediction Using MLDiabetes Prediction Using ML
Diabetes Prediction Using ML
 
Aa csh diabetespredictioncasestudy
Aa csh diabetespredictioncasestudyAa csh diabetespredictioncasestudy
Aa csh diabetespredictioncasestudy
 
Implementation of a Web Application to Foresee and Pretreat Diabetes Mellitus...
Implementation of a Web Application to Foresee and Pretreat Diabetes Mellitus...Implementation of a Web Application to Foresee and Pretreat Diabetes Mellitus...
Implementation of a Web Application to Foresee and Pretreat Diabetes Mellitus...
 
Analysis and Prediction of Diabetes Diseases using Machine Learning Algorithm...
Analysis and Prediction of Diabetes Diseases using Machine Learning Algorithm...Analysis and Prediction of Diabetes Diseases using Machine Learning Algorithm...
Analysis and Prediction of Diabetes Diseases using Machine Learning Algorithm...
 
Ai in diabetes management
Ai in diabetes managementAi in diabetes management
Ai in diabetes management
 
ML In Predicting Diabetes In The Early Stage
ML In Predicting Diabetes In The Early StageML In Predicting Diabetes In The Early Stage
ML In Predicting Diabetes In The Early Stage
 
K-Nearest Neighbours based diagnosis of hyperglycemia
K-Nearest Neighbours based diagnosis of hyperglycemiaK-Nearest Neighbours based diagnosis of hyperglycemia
K-Nearest Neighbours based diagnosis of hyperglycemia
 
PREDICTION OF DIABETES MELLITUS USING MACHINE LEARNING TECHNIQUES
PREDICTION OF DIABETES MELLITUS USING MACHINE LEARNING TECHNIQUESPREDICTION OF DIABETES MELLITUS USING MACHINE LEARNING TECHNIQUES
PREDICTION OF DIABETES MELLITUS USING MACHINE LEARNING TECHNIQUES
 
Holistic Forecasting of Onset of Diabetes through Data Mining Techniques
Holistic Forecasting of Onset of Diabetes through Data Mining TechniquesHolistic Forecasting of Onset of Diabetes through Data Mining Techniques
Holistic Forecasting of Onset of Diabetes through Data Mining Techniques
 
A study of modifiable and non-modifiable risk factors associated with diabeti...
A study of modifiable and non-modifiable risk factors associated with diabeti...A study of modifiable and non-modifiable risk factors associated with diabeti...
A study of modifiable and non-modifiable risk factors associated with diabeti...
 
AN IMPROVED MODEL FOR CLINICAL DECISION SUPPORT SYSTEM
AN IMPROVED MODEL FOR CLINICAL DECISION SUPPORT SYSTEMAN IMPROVED MODEL FOR CLINICAL DECISION SUPPORT SYSTEM
AN IMPROVED MODEL FOR CLINICAL DECISION SUPPORT SYSTEM
 
DIABETES PREDICTOR USING ENSEMBLE TECHNIQUE
DIABETES PREDICTOR USING ENSEMBLE TECHNIQUEDIABETES PREDICTOR USING ENSEMBLE TECHNIQUE
DIABETES PREDICTOR USING ENSEMBLE TECHNIQUE
 
Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral...
Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral...Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral...
Development of a Hybrid Dynamic Expert System for the Diagnosis of Peripheral...
 
Prognosis of Diabetes by Performing Data Mining of HbA1c
 Prognosis of Diabetes by Performing Data Mining of HbA1c Prognosis of Diabetes by Performing Data Mining of HbA1c
Prognosis of Diabetes by Performing Data Mining of HbA1c
 
A comparative analysis of biochemical and hematological parameters in diabeti...
A comparative analysis of biochemical and hematological parameters in diabeti...A comparative analysis of biochemical and hematological parameters in diabeti...
A comparative analysis of biochemical and hematological parameters in diabeti...
 
A Social-Ecological Approach To Determine Barriers Of DMSM Practice For Patie...
A Social-Ecological Approach To Determine Barriers Of DMSM Practice For Patie...A Social-Ecological Approach To Determine Barriers Of DMSM Practice For Patie...
A Social-Ecological Approach To Determine Barriers Of DMSM Practice For Patie...
 
ppt presentation for diabetes.pptx
ppt presentation for diabetes.pptxppt presentation for diabetes.pptx
ppt presentation for diabetes.pptx
 
Diabetes Prediction by Supervised and Unsupervised Approaches with Feature Se...
Diabetes Prediction by Supervised and Unsupervised Approaches with Feature Se...Diabetes Prediction by Supervised and Unsupervised Approaches with Feature Se...
Diabetes Prediction by Supervised and Unsupervised Approaches with Feature Se...
 
Austin Aging Research
Austin Aging Research Austin Aging Research
Austin Aging Research
 
Neo4j GraphTalk Basel - Using Graph Technology to drive Diabetes Reserach
Neo4j GraphTalk Basel - Using Graph Technology to drive Diabetes ReserachNeo4j GraphTalk Basel - Using Graph Technology to drive Diabetes Reserach
Neo4j GraphTalk Basel - Using Graph Technology to drive Diabetes Reserach
 

More from asclepiuspdfs

Comparison of the Hypocalcemic Effects of Erythropoietin and U-74389G
Comparison of the Hypocalcemic Effects of Erythropoietin and U-74389GComparison of the Hypocalcemic Effects of Erythropoietin and U-74389G
Comparison of the Hypocalcemic Effects of Erythropoietin and U-74389G
asclepiuspdfs
 
Bone Marrow Histology is a Pathognomonic Clue to Each of the JAK2V617F, MPL,5...
Bone Marrow Histology is a Pathognomonic Clue to Each of the JAK2V617F, MPL,5...Bone Marrow Histology is a Pathognomonic Clue to Each of the JAK2V617F, MPL,5...
Bone Marrow Histology is a Pathognomonic Clue to Each of the JAK2V617F, MPL,5...
asclepiuspdfs
 
Helicobacter pylori Frequency in Polycythemia Vera Patients without Dyspeptic...
Helicobacter pylori Frequency in Polycythemia Vera Patients without Dyspeptic...Helicobacter pylori Frequency in Polycythemia Vera Patients without Dyspeptic...
Helicobacter pylori Frequency in Polycythemia Vera Patients without Dyspeptic...
asclepiuspdfs
 
Clinical Significance of Hypocalcemia in COVID-19
Clinical Significance of Hypocalcemia in COVID-19Clinical Significance of Hypocalcemia in COVID-19
Clinical Significance of Hypocalcemia in COVID-19
asclepiuspdfs
 
Excess of Maternal Transmission of Type 2 Diabetes: Is there a Role of Bioche...
Excess of Maternal Transmission of Type 2 Diabetes: Is there a Role of Bioche...Excess of Maternal Transmission of Type 2 Diabetes: Is there a Role of Bioche...
Excess of Maternal Transmission of Type 2 Diabetes: Is there a Role of Bioche...
asclepiuspdfs
 
Predictive and Preventive Care: Metabolic Diseases
Predictive and Preventive Care: Metabolic DiseasesPredictive and Preventive Care: Metabolic Diseases
Predictive and Preventive Care: Metabolic Diseases
asclepiuspdfs
 

More from asclepiuspdfs (20)

Convalescent Plasma and COVID-19: Ancient Therapy Re-emerged
Convalescent Plasma and COVID-19: Ancient Therapy Re-emergedConvalescent Plasma and COVID-19: Ancient Therapy Re-emerged
Convalescent Plasma and COVID-19: Ancient Therapy Re-emerged
 
The Negative Clinical Consequences Due to the Lack of the Elaboration of a Sc...
The Negative Clinical Consequences Due to the Lack of the Elaboration of a Sc...The Negative Clinical Consequences Due to the Lack of the Elaboration of a Sc...
The Negative Clinical Consequences Due to the Lack of the Elaboration of a Sc...
 
Anemias Necessitating Transfusion Support
Anemias Necessitating Transfusion SupportAnemias Necessitating Transfusion Support
Anemias Necessitating Transfusion Support
 
Decreasing or Increasing Role of Autologous Stem Cell Transplantation in Mult...
Decreasing or Increasing Role of Autologous Stem Cell Transplantation in Mult...Decreasing or Increasing Role of Autologous Stem Cell Transplantation in Mult...
Decreasing or Increasing Role of Autologous Stem Cell Transplantation in Mult...
 
Comparison of the Hypocalcemic Effects of Erythropoietin and U-74389G
Comparison of the Hypocalcemic Effects of Erythropoietin and U-74389GComparison of the Hypocalcemic Effects of Erythropoietin and U-74389G
Comparison of the Hypocalcemic Effects of Erythropoietin and U-74389G
 
Refractory Anemia
Refractory AnemiaRefractory Anemia
Refractory Anemia
 
Management of Immunogenic Heparin-induced Thrombocytopenia
Management of Immunogenic Heparin-induced ThrombocytopeniaManagement of Immunogenic Heparin-induced Thrombocytopenia
Management of Immunogenic Heparin-induced Thrombocytopenia
 
Adult Still’s Disease Resembling Drug-related Scratch Dermatitis
Adult Still’s Disease Resembling Drug-related Scratch DermatitisAdult Still’s Disease Resembling Drug-related Scratch Dermatitis
Adult Still’s Disease Resembling Drug-related Scratch Dermatitis
 
Bone Marrow Histology is a Pathognomonic Clue to Each of the JAK2V617F, MPL,5...
Bone Marrow Histology is a Pathognomonic Clue to Each of the JAK2V617F, MPL,5...Bone Marrow Histology is a Pathognomonic Clue to Each of the JAK2V617F, MPL,5...
Bone Marrow Histology is a Pathognomonic Clue to Each of the JAK2V617F, MPL,5...
 
Helicobacter pylori Frequency in Polycythemia Vera Patients without Dyspeptic...
Helicobacter pylori Frequency in Polycythemia Vera Patients without Dyspeptic...Helicobacter pylori Frequency in Polycythemia Vera Patients without Dyspeptic...
Helicobacter pylori Frequency in Polycythemia Vera Patients without Dyspeptic...
 
Lymphoma of the Tonsil in a Developing Community
Lymphoma of the Tonsil in a Developing CommunityLymphoma of the Tonsil in a Developing Community
Lymphoma of the Tonsil in a Developing Community
 
Should Metformin Be Continued after Hospital Admission in Patients with Coron...
Should Metformin Be Continued after Hospital Admission in Patients with Coron...Should Metformin Be Continued after Hospital Admission in Patients with Coron...
Should Metformin Be Continued after Hospital Admission in Patients with Coron...
 
Clinical Significance of Hypocalcemia in COVID-19
Clinical Significance of Hypocalcemia in COVID-19Clinical Significance of Hypocalcemia in COVID-19
Clinical Significance of Hypocalcemia in COVID-19
 
Excess of Maternal Transmission of Type 2 Diabetes: Is there a Role of Bioche...
Excess of Maternal Transmission of Type 2 Diabetes: Is there a Role of Bioche...Excess of Maternal Transmission of Type 2 Diabetes: Is there a Role of Bioche...
Excess of Maternal Transmission of Type 2 Diabetes: Is there a Role of Bioche...
 
The Effect of Demographic Data and Hemoglobin A 1c on Treatment Outcomes in P...
The Effect of Demographic Data and Hemoglobin A 1c on Treatment Outcomes in P...The Effect of Demographic Data and Hemoglobin A 1c on Treatment Outcomes in P...
The Effect of Demographic Data and Hemoglobin A 1c on Treatment Outcomes in P...
 
Self-efficacy Impact Adherence in Diabetes Mellitus
Self-efficacy Impact Adherence in Diabetes MellitusSelf-efficacy Impact Adherence in Diabetes Mellitus
Self-efficacy Impact Adherence in Diabetes Mellitus
 
Uncoiling the Tightening Obesity Spiral
Uncoiling the Tightening Obesity SpiralUncoiling the Tightening Obesity Spiral
Uncoiling the Tightening Obesity Spiral
 
Prevalence of Chronic Kidney disease in Patients with Metabolic Syndrome in S...
Prevalence of Chronic Kidney disease in Patients with Metabolic Syndrome in S...Prevalence of Chronic Kidney disease in Patients with Metabolic Syndrome in S...
Prevalence of Chronic Kidney disease in Patients with Metabolic Syndrome in S...
 
Management Of Hypoglycemia In Patients With Type 2 Diabetes
Management Of Hypoglycemia In Patients With Type 2 DiabetesManagement Of Hypoglycemia In Patients With Type 2 Diabetes
Management Of Hypoglycemia In Patients With Type 2 Diabetes
 
Predictive and Preventive Care: Metabolic Diseases
Predictive and Preventive Care: Metabolic DiseasesPredictive and Preventive Care: Metabolic Diseases
Predictive and Preventive Care: Metabolic Diseases
 

Recently uploaded

Best medicine 100% Effective&Safe Mifepristion ௵+918133066128௹Abortion pills ...
Best medicine 100% Effective&Safe Mifepristion ௵+918133066128௹Abortion pills ...Best medicine 100% Effective&Safe Mifepristion ௵+918133066128௹Abortion pills ...
Best medicine 100% Effective&Safe Mifepristion ௵+918133066128௹Abortion pills ...
Abortion pills in Kuwait Cytotec pills in Kuwait
 
Failure to thrive in neonates and infants + pediatric case.pptx
Failure to thrive in neonates and infants  + pediatric case.pptxFailure to thrive in neonates and infants  + pediatric case.pptx
Failure to thrive in neonates and infants + pediatric case.pptx
claviclebrown44
 
CAS 110-63-4 BDO Liquid 1,4-Butanediol 1 4 BDO Warehouse Supply For Excellent...
CAS 110-63-4 BDO Liquid 1,4-Butanediol 1 4 BDO Warehouse Supply For Excellent...CAS 110-63-4 BDO Liquid 1,4-Butanediol 1 4 BDO Warehouse Supply For Excellent...
CAS 110-63-4 BDO Liquid 1,4-Butanediol 1 4 BDO Warehouse Supply For Excellent...
ocean4396
 

Recently uploaded (20)

CONGENITAL HYPERTROPHIC PYLORIC STENOSIS by Dr M.KARTHIK EMMANUEL
CONGENITAL HYPERTROPHIC PYLORIC STENOSIS  by Dr M.KARTHIK EMMANUELCONGENITAL HYPERTROPHIC PYLORIC STENOSIS  by Dr M.KARTHIK EMMANUEL
CONGENITAL HYPERTROPHIC PYLORIC STENOSIS by Dr M.KARTHIK EMMANUEL
 
DR. Neha Mehta Best Psychologist.in India
DR. Neha Mehta Best Psychologist.in IndiaDR. Neha Mehta Best Psychologist.in India
DR. Neha Mehta Best Psychologist.in India
 
ROSE CASE SPINAL SBRT BY DR KANHU CHARAN PATRO
ROSE  CASE SPINAL SBRT BY DR KANHU CHARAN PATROROSE  CASE SPINAL SBRT BY DR KANHU CHARAN PATRO
ROSE CASE SPINAL SBRT BY DR KANHU CHARAN PATRO
 
Best medicine 100% Effective&Safe Mifepristion ௵+918133066128௹Abortion pills ...
Best medicine 100% Effective&Safe Mifepristion ௵+918133066128௹Abortion pills ...Best medicine 100% Effective&Safe Mifepristion ௵+918133066128௹Abortion pills ...
Best medicine 100% Effective&Safe Mifepristion ௵+918133066128௹Abortion pills ...
 
ANAPHYLAXIS BY DR.SOHAN BISWAS,MBBS,DNB(INTERNAL MEDICINE) RESIDENT.pptx
ANAPHYLAXIS BY DR.SOHAN BISWAS,MBBS,DNB(INTERNAL MEDICINE) RESIDENT.pptxANAPHYLAXIS BY DR.SOHAN BISWAS,MBBS,DNB(INTERNAL MEDICINE) RESIDENT.pptx
ANAPHYLAXIS BY DR.SOHAN BISWAS,MBBS,DNB(INTERNAL MEDICINE) RESIDENT.pptx
 
Unveiling Alcohol Withdrawal Syndrome: exploring it's hidden depths
Unveiling Alcohol Withdrawal Syndrome: exploring it's hidden depthsUnveiling Alcohol Withdrawal Syndrome: exploring it's hidden depths
Unveiling Alcohol Withdrawal Syndrome: exploring it's hidden depths
 
Failure to thrive in neonates and infants + pediatric case.pptx
Failure to thrive in neonates and infants  + pediatric case.pptxFailure to thrive in neonates and infants  + pediatric case.pptx
Failure to thrive in neonates and infants + pediatric case.pptx
 
5cladba raw material 5CL-ADB-A precursor raw
5cladba raw material 5CL-ADB-A precursor raw5cladba raw material 5CL-ADB-A precursor raw
5cladba raw material 5CL-ADB-A precursor raw
 
Premium ℂall Girls In Mumbai👉 Dail ℂALL ME: 📞9833325238 📲 ℂall Richa VIP ℂall...
Premium ℂall Girls In Mumbai👉 Dail ℂALL ME: 📞9833325238 📲 ℂall Richa VIP ℂall...Premium ℂall Girls In Mumbai👉 Dail ℂALL ME: 📞9833325238 📲 ℂall Richa VIP ℂall...
Premium ℂall Girls In Mumbai👉 Dail ℂALL ME: 📞9833325238 📲 ℂall Richa VIP ℂall...
 
TEST BANK For Huether and McCance's Understanding Pathophysiology, Canadian 2...
TEST BANK For Huether and McCance's Understanding Pathophysiology, Canadian 2...TEST BANK For Huether and McCance's Understanding Pathophysiology, Canadian 2...
TEST BANK For Huether and McCance's Understanding Pathophysiology, Canadian 2...
 
Results For Love Spell Is Guaranteed In 1 Day +27834335081 [BACK LOST LOVE SP...
Results For Love Spell Is Guaranteed In 1 Day +27834335081 [BACK LOST LOVE SP...Results For Love Spell Is Guaranteed In 1 Day +27834335081 [BACK LOST LOVE SP...
Results For Love Spell Is Guaranteed In 1 Day +27834335081 [BACK LOST LOVE SP...
 
World Hypertension Day 17th may 2024 ppt
World Hypertension Day 17th may 2024 pptWorld Hypertension Day 17th may 2024 ppt
World Hypertension Day 17th may 2024 ppt
 
SEMESTER-V CHILD HEALTH NURSING-UNIT-1-INTRODUCTION.pdf
SEMESTER-V CHILD HEALTH NURSING-UNIT-1-INTRODUCTION.pdfSEMESTER-V CHILD HEALTH NURSING-UNIT-1-INTRODUCTION.pdf
SEMESTER-V CHILD HEALTH NURSING-UNIT-1-INTRODUCTION.pdf
 
Evidence-based practiceEBP) in physiotherapy
Evidence-based practiceEBP) in physiotherapyEvidence-based practiceEBP) in physiotherapy
Evidence-based practiceEBP) in physiotherapy
 
Is Rheumatoid Arthritis a Metabolic Disorder.pptx
Is Rheumatoid Arthritis a Metabolic Disorder.pptxIs Rheumatoid Arthritis a Metabolic Disorder.pptx
Is Rheumatoid Arthritis a Metabolic Disorder.pptx
 
Dermatome and myotome test & pathology.pdf
Dermatome and myotome test & pathology.pdfDermatome and myotome test & pathology.pdf
Dermatome and myotome test & pathology.pdf
 
Premium ℂall Girls In Mumbai Airport👉 Dail ℂALL ME: 📞9833325238 📲 ℂall Richa ...
Premium ℂall Girls In Mumbai Airport👉 Dail ℂALL ME: 📞9833325238 📲 ℂall Richa ...Premium ℂall Girls In Mumbai Airport👉 Dail ℂALL ME: 📞9833325238 📲 ℂall Richa ...
Premium ℂall Girls In Mumbai Airport👉 Dail ℂALL ME: 📞9833325238 📲 ℂall Richa ...
 
CAS 110-63-4 BDO Liquid 1,4-Butanediol 1 4 BDO Warehouse Supply For Excellent...
CAS 110-63-4 BDO Liquid 1,4-Butanediol 1 4 BDO Warehouse Supply For Excellent...CAS 110-63-4 BDO Liquid 1,4-Butanediol 1 4 BDO Warehouse Supply For Excellent...
CAS 110-63-4 BDO Liquid 1,4-Butanediol 1 4 BDO Warehouse Supply For Excellent...
 
Case presentation on Antibody screening- how to solve 3 cell and 11 cell panel?
Case presentation on Antibody screening- how to solve 3 cell and 11 cell panel?Case presentation on Antibody screening- how to solve 3 cell and 11 cell panel?
Case presentation on Antibody screening- how to solve 3 cell and 11 cell panel?
 
Tips and tricks to pass the cardiovascular station for PACES exam
Tips and tricks to pass the cardiovascular station for PACES examTips and tricks to pass the cardiovascular station for PACES exam
Tips and tricks to pass the cardiovascular station for PACES exam
 

Prediction of Risk Factors Leading to Diabetes Using Neural Network Analysis

  • 1. Clinical Research in Diabetes and Endocrinology  •  Vol 3  •  Issue 2  •  2020 16 INTRODUCTION I n2017,estimatesofdiabeteshaveshowntheexistenceof425 million diabetics in the world, and this number is expected to reach to 625 million by 2045 (IDF, 2017; Li et al., 2019).[1-3] Diabetes mellitus is a category of endocrine disorders associated with impaired glucose absorption that occurs as a consequence of the “Insulin” hormone’s absolute or relative insufficiency. Condition is characterized by a chronic path and a violation of all forms of metabolism.[4] Diabetes is usually divided into four categories: Type 1 diabetes (T1D), type 2 diabetes, gestational diabetes mellitus, and, due to other factors, particular types of diabetes. T1D and T2D are the two most common forms of the condition (T2D). The former is caused by the degradation of the beta cells of the pancreas, which results in insulin deficiency, whereas the latter is caused by the inadequate transport of insulin to the cells.[5] Life-threatening complications such as strokes, heart attacks, chronic renal failure, diabetic foot syndrome, antipathy, neuropathy, encephalopathy, hyperthyroidism, tumors of the adrenal gland, liver cirrhosis, glucagonoma, and transient hyperglycemia and many other complications can contribute to both forms of disease.[4] For all people who are predisposed to diabetes, the prediction and early detection of diabetes are therefore important. Using artificial intelligence techniques, multiple diseases can currently be diagnosed, and deep neural networks have achieved the best results in classification problems (Miotto etal., 2017; Liu et al., 2019).[3,6,7] Data mining was used to predict diabetes.[5,8,9] Ali et al. used neural network analytics to build a model for classifying diabetic people based on past medical history and patient control level. The author developed a new predictive model for data mining methods that would identify the level of control of diabetic patients centered on medical reports that are historical. The analysis was carried out with the use of three techniques RESEARCH ARTICLE Prediction of Risk Factors Leading to Diabetes Using Neural Network Analysis Ahed J. Alkhatib1,2 , Amer Mahmoud Sindiani3 , Eman Hussein Alshdaifat4 1 Department of Legal Medicine, Toxicology and Forensic Medicine, Jordan University of Science and Technology, Jordan,2 International Mariinskaya Academy, Department of Medicine and Critical Care, Department of Philosophy, Academician Secretary of Department of Sociology, Jordan, 3 Department of Obstetrics and Gynecology, Faculty of Medicine, Jordan University of Science and Technology, Jordan, 4 Department of Obstetrics and Gynecology, Faculty of medicine, Yarmouk University, Jordan ABSTRACT This study presents a prediction of diabetes type 2 using neural network analysis. A dataset of diabetes posted in Kaggle was used for analysis. The dataset included one dependent variable, outcome of disease, with two numbers, “0” means no disease, and “1” means disease. There were other eight risk factors for diabetes prediction, glucose, body mass index (BMI), insulin, skin thickness, no. of pregnancies, diabetes pedigree function, and age. We constructed the model using SPSS version 21. The results showed that the model prediction was 78.3% for training and 76.9% for testing. The model showed that the most significant predictors of diabetes were: glucose, BMI, diabetes pedigree function, no. of pregnancies, age, blood pressure, insulin, and skin thickness. Taken together, this model was effective and leads to a good prediction rate. Key words: Covariates, dependent variable, diabetes type 2, neural network analysis, outcome Address for correspondence: Ahed J. Alkhatib, Department of Legal Medicine, Toxicology and Forensic Medicine, Jordan University of Science and Technology, Jordan. Mobile: 00962795905145. https://doi.org/10.33309/2639-832X.030204 www.asclepiusopen.com © 2020 The Author(s). This open access article is distributed under a Creative Commons Attribution (CC-BY) 4.0 license.
  • 2. Alkhatib: Relative importance of risk factors in diabetes type 2 Clinical Research in Diabetes and Endocrinology  •  Vol 3  •  Issue 2  •  2020 of data processing, which are Logistic, Naïve Bayes, and J48. The analysis was carried out using the WEKA program. The outcome showed that logistics data mining algorithm gave an average accuracy of 0.73, a recall of 0.744, a metric of 0.653, and a precision of 74.4%. Naïve Bayes gave an average accuracy of 0.717, 0.742 recall, 0.653 F-measure, and 74.2% precision. J48 has given an average accuracy of 0.54, recall of 0.735, F-measure of 0.623, and precision of 0.54 and 73.5 percentage points. This showed that the logistic algorithm was more precise than the other two algorithms. Study objectives The main objective of this study was to identify the risk factors associated with diabetes and to determine their relative importance. METHODOLOGY The present study was based on a dataset posted in Kaggle.[3] The dataset of diabetes taken from India was analyzed in this study using a neural network.The data consisted of 768 females. Study variables The dataset involved one dependent variable, outcome indicating having diabetes (1) or no diabetes (0). There are 8 covariates or independent variable: Glucose, insulin, age, blood pressure, skin thickness, no. of pregnancies, and diabetes pedigree function. Statistical analysis Neural network analytics of data was carried out using SPSS version 21. RESULTS As demonstrated in Table 1, the study sample included 768 cases, 526 (68.5%) in training section, and 242 (31.5%) in testing section. All cases were included in the study. Network information As demonstrated in Table 2, network information included the following parameters: Input layer with 8 covariates: Pregnancies number, glucose, blood pressure, skin thickness, insulin, body mass index (BMI), diabetes pedigree function, and age. Number of units is 8 and rescaling method for covariates is standardized. The second parameter was about the hidden layer (s), there was one hidden layer with five units in hidden layer, and the activation function was hyperbolic tangent. The third parameter was the output layer. There was one variable, the output. It has 2 number of units. Activation function was softmax and the error function was cross-entropy. Model summary Weconstructedthemodelthathadthefollowingcharacteristics [Table 3]: For training, cross-entropy error was computed as Table 2: Network information Input layer Covariates 1 Pregnancies No. 2 Glucose 3 Blood pressure 4 Skin thickness 5 Insulin 6 BMI 7 Diabetes pedigree function 8 Age Number of unitsa 8 Rescaling method for covariates Standardized Hidden layer(s) Number of hidden layers 1 Number of units in hidden layer 1a 5 Activation function Hyperbolic tangent Output layer Dependent variables 1 outcome Number of units 2 Activation function Soft max Error function Cross-entropy BMI: Body mass index Table 3: Model summary Training ??? Cross-entropy error 241.342 Percent incorrect predictions 21.7 Stopping rule used 1 consecutive step(s) with no decrease in errora Training time 0:00:00.16 Testing Cross-entropy error 111.363 Percent incorrect predictions 23.1 Table 1: Case processing summary n Percent Sample Training 526 68.5 Testing 242 31.5 Valid 768 100.0 Excluded 0 Total 768
  • 3. Alkhatib: Relative importance of risk factors in diabetes type 2 Clinical Research in Diabetes and Endocrinology  •  Vol 3  •  Issue 2  •  2020 18 241.342, percent incorrect prediction 21.7%, stopping rules used were 1 consecutive step(s) with no decrease in errora , and training time was 0:000:00.16. For testing, cross-entropy error was 111.363 and percent incorrect prediction was 23.1%. Dependent variable was outcome. Classification As seen in Table 4, in this study, the outcome variable implied two values, “0”: Non-diabetic, and “1”: Diabetic. In the training part, for 0 output, the system determined 345 cases, 47 cases were classified by mistake in “0” output, but the system put them in “1” output. Percent correction was computed as 86.4%. In “1” output, 181 cases were included, among which 67 cases were predicted to be in “0” output. Percent correction was computed as 63%. The overall percent was 78.3%. In the testing part, 155 cases were included in the output “0,” among which 23 cases were predicted to have the disease. The percent correction was 85.2%. For the output “1”, 87 cases were included, of which 33 cases were predicted to be in the output “0.” Percent correction was found to be 62.1%. The overall percent was 76.9%. Independent variable importance As seen in Table 5 and Figure 1, the importance of glucose level was 0.276 (100%), BMI 0.244 (88.4%), diabetes pedigree function 0.145 (52.6%), no. of pregnancy 0.101 (36.5%), age 0.08 (29.1%), blood pressure 0.071 (25.7%), insulin 0.052 (18.8%), and skin thickness 0.030 (10.7%). DISCUSSION The results of this study identified diabetes risk factors among females who had diabetes type 2. Using neural network analysis usually uses technical language that is not easily understood by readers. The problem of these studies is being made by non-medical professional, including those working in computer engineering or software engineers. According to this context, the author integrated technical and medical languages to help the reader to be able understand the topic of diabetes. According to our model, covariates or independent variables, glucose level was the first important predictor of diabetes type 2 among females, and its importance 100%. BMI was able to predict 88.4% of diabetes cases [Figure 1]. The importance of such a figure is that it pointed to points that we need to control. Another important point is the number Table 4: Classification Sample Observed Predicted 0.00 1.00 Percent correct Training 0.00 298 47 86.4 1.00 67 114 63.0 Overall percent 69.4 30.6 78.3 Testing 0.00 132 23 85.2 1.00 33 54 62.1 Overall percent 68.2 31.8 76.9 Dependent variable: Outcome Figure 1: Normalized importance of covariates
  • 4. Alkhatib: Relative importance of risk factors in diabetes type 2 Clinical Research in Diabetes and Endocrinology  •  Vol 3  •  Issue 2  •  2020 of pregnancies that predicted about 37% of diabetic cases among females. CONCLUSIONS The results of the present study pointed to two important considerations, the use of neutral network analytics is effective in the prediction of diabetes and identifying risk factors with their relative importance. REFERENCES 1. Zhou H, Myrzashova R, Zheng R. Diabetes prediction model based on an enhanced deep neural network. EURASIP J Wirel Commun Netw 2020;148:1621. 2. Li G, Peng S, Wang C, Niu J, Yuan Y. An energy-efficient data collection scheme using denoising autoencoder in wireless sensor networks. Tsinghua Sci Technol 2019;24:86-96. 3. Liu H, Kou H, Yan C, Qi L. Link prediction in paper citation network to construct paper correlated graph. EURASIP J Wirel Commun Netw 2019;233:2352. 4. Available from: https://www.kaggle.com/uciml/pima-indians- diabetes-database. [Last accessed on 2020 Dec 10]. 5. American Diabetes Association. Classification and diagnosis of diabetes: Standards of medical care in diabetes-2018. Diabetes Care 2018;41:13-27. 6. Ali R, Siddiqi MH, Idris M, Kang BH, Lee S. Prediction of diabetes mellitus based on boosting ensemble modeling. In: International Conference on Ubiquitous Computing and Ambient Intelligence. Berlin, Germany: Springer International Publishing; 2014. p. 25-8. 7. Miotto R, Wang F, Wang S, Jiang X, Dudley T. Deep learning for healthcare: Review, opportunities and challenges. Brief Bioinform 2017;19:1236-46. 8. International Diabetes Federation. IDF Diabetes Atlas. 8th ed. Brussels, Belgium: International Diabetes Federation; 2017. 9. El_Jerjawi NS, Abu-Naser SS. Diabetes prediction using artificial neural network. Int J Adv Sci Technol 2018;121:55-64. How to cite this article: Alkhatib AJ, Sindiani AM, Alshdaifat EH. Prediction of Risk Factors Leading to Diabetes Using Neural Network Analysis. Clin Res Diabetes Endocrinol 2020;3(2):16-19. Table 5: Independent variable importance Independent variable Importance Normalized importance % No. of pregnancies 0.101 36.5 Glucose level 0.276 100.0 Blood pressure 0.071 25.7 Skin thickness 0.030 10.7 Insulin 0.052 18.8 BMI 0.244 88.4 Diabetes pedigree function 0.145 52.6 Age 0.080 29.1 BMI: Body mass index