This document describes a study that uses Hidden Markov Models to detect exacerbations in COPD patients based on daily self-reported health scores. The HMM is tuned to balance high recall and precision. Results show the HMM outperforms an existing method, and Poisson regression on detected exacerbations finds Drug A reduces frequency more than Drug B. Further improvements could overcome data constraints and better model exacerbation events.
The document discusses dose-response relationships, which relate the amount of a drug administered (dose) to its biological effect (response). It defines key terms like dose, response, dose-response curve, potency, efficacy, and therapeutic index. Dose-response curves can take different shapes depending on whether response is measured quantally or gradually, and can be used to determine a drug's potency, efficacy, and margin of safety between therapeutic and toxic doses.
Art therapy uses creative activities like painting, drawing, and sculpting to increase well-being and help people overcome challenges. It is used in mental health settings to help people explore feelings, develop skills, and boost self-esteem. Research shows art therapy can reduce stress, depression, and other symptoms in people with conditions like cancer, heart disease, and HIV/AIDS. However, more extensive research is still needed to demonstrate art therapy's full health benefits.
This document is a resume for Jayant Sharma. It summarizes his contact information, objective, educational qualifications, areas of interest, projects undertaken, computer skills, publications, trainings, workshops, skills, and personal details. The resume shows that Jayant has an M-Tech in Digital Electronics and Advanced Communication from Sikkim Manipal Institute of Technology with good grades. He has work experience in maintenance and production. His areas of interest include pulse and digital circuits, mobile communication, and computer networks.
Customer relationship management (CRM) is a method used by companies to analyze customer data and improve business relationships with existing customers to drive sales growth. CRM systems compile customer information from different channels such as phone, website, email, marketing materials and social media. By taking a CRM approach and using these systems, businesses can learn more about their target audiences and how to best cater to their needs, though it risks favoring some customers over others.
1) The story describes a woman who was assaulted by a man she met. She tried to forget what happened but was continually haunted by memories of the attack.
2) She decided to hunt down her attacker and kill him to stop him from hurting other women. She stalked him and eventually shot and killed him.
3) However, the authorities saw it as premeditated murder and she received a life sentence. In prison, she continues to feel like both the hunter and the hunted as she tries to survive her sentence.
Este documento apresenta fotos em preto e branco de várias figuras históricas e celebridades antes de se tornarem famosas, incluindo políticos, atores, músicos e outros. As fotos mostram essas figuras quando crianças, adolescentes ou jovens adultos, demonstrando como eles eram antes de alcançarem o estrelato ou fama. Muitas das fotos fornecem detalhes interessantes, como idades, atividades ou situações em que foram tiradas.
The document discusses dose-response relationships, which relate the amount of a drug administered (dose) to its biological effect (response). It defines key terms like dose, response, dose-response curve, potency, efficacy, and therapeutic index. Dose-response curves can take different shapes depending on whether response is measured quantally or gradually, and can be used to determine a drug's potency, efficacy, and margin of safety between therapeutic and toxic doses.
Art therapy uses creative activities like painting, drawing, and sculpting to increase well-being and help people overcome challenges. It is used in mental health settings to help people explore feelings, develop skills, and boost self-esteem. Research shows art therapy can reduce stress, depression, and other symptoms in people with conditions like cancer, heart disease, and HIV/AIDS. However, more extensive research is still needed to demonstrate art therapy's full health benefits.
This document is a resume for Jayant Sharma. It summarizes his contact information, objective, educational qualifications, areas of interest, projects undertaken, computer skills, publications, trainings, workshops, skills, and personal details. The resume shows that Jayant has an M-Tech in Digital Electronics and Advanced Communication from Sikkim Manipal Institute of Technology with good grades. He has work experience in maintenance and production. His areas of interest include pulse and digital circuits, mobile communication, and computer networks.
Customer relationship management (CRM) is a method used by companies to analyze customer data and improve business relationships with existing customers to drive sales growth. CRM systems compile customer information from different channels such as phone, website, email, marketing materials and social media. By taking a CRM approach and using these systems, businesses can learn more about their target audiences and how to best cater to their needs, though it risks favoring some customers over others.
1) The story describes a woman who was assaulted by a man she met. She tried to forget what happened but was continually haunted by memories of the attack.
2) She decided to hunt down her attacker and kill him to stop him from hurting other women. She stalked him and eventually shot and killed him.
3) However, the authorities saw it as premeditated murder and she received a life sentence. In prison, she continues to feel like both the hunter and the hunted as she tries to survive her sentence.
Este documento apresenta fotos em preto e branco de várias figuras históricas e celebridades antes de se tornarem famosas, incluindo políticos, atores, músicos e outros. As fotos mostram essas figuras quando crianças, adolescentes ou jovens adultos, demonstrando como eles eram antes de alcançarem o estrelato ou fama. Muitas das fotos fornecem detalhes interessantes, como idades, atividades ou situações em que foram tiradas.
This document discusses the importance of quality control for echocardiography interpretation and summarizes several studies on improving inter-observer reliability and accuracy. The first study found low agreement between experts in assessing diastolic function from the same patient data due to a lack of hierarchy for integrating parameters. A later study developed a consensus strategy to improve grading of aortic regurgitation severity from 70% to over 90% agreement. Another study reduced inter-observer variability in left ventricular ejection fraction measurements from echocardiograms by having readers review their measurements alongside cardiac MRI images. A final study showed quantitative measurements of right ventricular function improved accuracy and agreement compared to qualitative assessments alone. The studies demonstrate the value of quality improvement efforts like developing consensus guidelines
When designing a clinical study, a fundamental aspect is the sample size. In this article, we describe the rationale for sample size calculations, when it should be calculated and describe the components necessary to calculate it. For simple studies, standard formulae can be
used; however, for more advanced studies, it is generally necessary to use specialized statistical software programs and consult a biostatistician. Sample size calculations for non-randomized studies are also discussed and two clinical examples are used for illustration
850-0490-00 Rev A-Oscillatory MAP ABP Case Study Christina Mason
Mary Jones underwent a cardiac procedure and her blood pressure is being monitored. Her nurse noticed a difference between the mean arterial pressure (MAP) displayed on the monitor and the MAP she calculated using the usual (SBP + 2DBP)/3 formula. There are a few reasons why the values may differ: 1) The monitor displays the MAP directly measured by the cuff, while the formula provides an estimate. 2) True MAP can only be determined invasively, while noninvasive methods like oscillometry and auscultation may vary from invasive readings by up to 5 mmHg. 3) Research shows oscillometric MAP more closely matches invasive MAP than SBP or DBP. Clinicians can generally trust the MAP
This document outlines a study on jointly modeling multivariate longitudinal measures of hypertension (blood pressure and pulse rate) and time to develop cardiovascular disease among hypertensive outpatients in Ethiopia. The study aims to identify factors affecting changes in blood pressure and pulse rate over time as well as time to develop cardiovascular complications. The study will collect longitudinal data on blood pressure, pulse rate and cardiovascular events from 178 hypertensive patients and analyze it using joint longitudinal-survival models. Preliminary results show changes in blood pressure and pulse rate over time differ between patients who did and did not develop cardiovascular events. Key factors like diabetes, family history of hypertension and clinical stage of hypertension affect both longitudinal outcomes and survival.
The study reviewed the relationship between dietary and supplemental antioxidants and prostate cancer risk. Antioxidants examined included vitamin E, selenium, vitamin C, carotenoids, and polyphenols from coffee and tea. The evidence for effects of vitamin E and selenium on prostate cancer risk was inconsistent. While some studies found protective effects of selenium at low baseline levels, others found no effect. Studies of vitamin C, carotenoids, and polyphenols like green tea provided inconclusive or no evidence of relationships with prostate cancer risk.
This study evaluated the performance of three pneumonia severity scores - CURB-65, CRB-65, and CURB-age - in predicting 30-day mortality in 559 adult inpatients with community-acquired pneumonia. The study found that CURB-age stratified patients into risk groups that were most closely associated with mortality outcomes. Specifically, CURB-age identified more low-risk patients who had lower mortality than CRB-65, and sorted more patients who died within 30 days into the high-risk group compared to the other scores. Analysis of receiver operating characteristics further indicated CURB-age had better ability to predict 30-day mortality compared to the other scores. The study concludes CURB-age may provide
This document discusses case-control studies and their design and analysis. It begins by defining case-control studies as observational studies where subjects are sampled based on disease presence or absence and their prior exposure is then determined. It describes key features, need, steps in design including case and control selection. It then covers statistical analysis including odds ratios to measure risk associated with exposure and interpretations. It discusses effect modification and confounding, and analytical tools like stratification and multivariate modeling to control for confounding.
Epidemiological Approaches for Evaluation of diagnostic tests.pptxBhoj Raj Singh
Diagnosis of a disease or a problem is the first step towards solution/ treatment. Clinical Diagnosis or Provisional Diagnosis is the first step in diagnosis and is done after a physical examination of the patient by a clinician. Clinical diagnosis may or may not be true and to reach Final diagnosis Laboratory Investigations using gross and microscopic pathological observations and determining the disease indicators are required. The diagnostic tests may be Non-dichotomous Diagnostic Tests (when continuous values are given by the test in a range starting from sub-normal to above-normal range) and Dichotomous Diagnostic Tests (when results are given either plus or minus, disease or no-disease). To make non- Dichotomous diagnostic test a Dichotomous one you need to establish the cut-off values based on reference values or Gold Standard test readings or with the use of Receiver operator characteristic (ROC) curves, Precision-Recall Curves, Likelihood Ratios, etc., and finally establishing statistical agreement (using Kappa values, Level of Agreement, χ2 Statistics) between the true diagnosis and laboratory diagnosis. Thereafter, the Accuracy, Precision, Bias, Sensitivity, Specificity, Positive Predictive value, and Negative Predictive value, of a diagnostic test are established for use in clinical practice. Diagnostic tests are also used to determine Prevalence (True prevalence, apparent prevalence) and Incidence of the disease to estimate the disease burden so that control measures can be implemented. There are several Phases in the development and use of a diagnostic assay starting from conceptualization of the diagnostic test, development and evaluation to determine flaws in diagnostic test use and Interpretation influencers. This presentation mainly deals with the epidemiological evaluation procedures for diagnostic tests.
HEART DISEASES PREDICTION USING MACHINE LEARNING ALGORITHMPoojaSri45
Implemented a machine learning project aimed at predicting heart diseases using various algorithms and techniques. Developed as a part of academic or professional endeavor, the project demonstrates proficiency in data preprocessing, feature selection, model training, and evaluation.
VALIDITY AND RELIABLITY OF A SCREENING TEST seminar 2.pptxShaliniPattanayak
A presentation shedding some insight into the tricky concepts of validity and reliability of any screening test, used in day-to-day lives, using easy and understandable language.
David Madigan MedicReS World Congress 2014MedicReS
This document discusses issues with published observational studies and potential ways to improve them. It notes that even small choices in study design can lead to a wide range of results. Applying different analysis methods to large healthcare databases still found many false positives and negatives. However, performance improved by tailoring analyses to specific outcomes and restricting to larger sample sizes. Self-controlled designs like case-crossover consistently outperformed traditional cohort and case-control designs. Further strategies like evaluating results against known null distributions may help address biases from unmeasured confounding.
This document summarizes a study evaluating the use of D-dimer testing and clinical risk algorithms to diagnose pulmonary embolism (PE). The study assessed 627 patients presenting with suspected PE who were stratified into low, intermediate, or high risk groups using the Geneva score. For low and intermediate risk patients, a quantitative D-dimer assay showed 100% sensitivity and negative predictive value, correctly ruling out PE in 25% and 33% of cases respectively. The study supports using the D-dimer assay as a first-line test for low to intermediate risk PE patients to reduce unnecessary CT scans.
This document summarizes a study evaluating the use of D-dimer testing and clinical risk algorithms to diagnose pulmonary embolism (PE). The study assessed 627 patients presenting with suspected PE who were stratified into low, intermediate, or high risk groups using the Geneva score. For low and intermediate risk patients, a quantitative D-dimer assay showed 100% sensitivity and negative predictive value, correctly ruling out PE in 25% and 33% of cases respectively. The study supports using the D-dimer assay as a first-line test for low to intermediate risk PE patients to reduce unnecessary CT scans.
This document summarizes a study evaluating the use of D-dimer testing and clinical risk algorithms to diagnose pulmonary embolism (PE). The study assessed 627 patients presenting with suspected PE who were stratified into low, intermediate, or high risk groups using the Geneva score. For low and intermediate risk patients, a quantitative D-dimer assay achieved 100% sensitivity and negative predictive value, correctly ruling out PE in 25% and 33% of cases respectively. The study supports using the D-dimer assay as a first-line test for low to intermediate risk PE patients to reduce unnecessary CT scans.
This document summarizes a study evaluating the use of D-dimer testing and clinical risk algorithms to diagnose pulmonary embolism (PE). The study assessed 627 patients presenting with suspected PE who were stratified into low, intermediate, or high risk groups using the Geneva score. For low and intermediate risk patients, a quantitative D-dimer assay achieved 100% sensitivity and negative predictive value, correctly ruling out PE in 25% and 33% of cases respectively. The study supports using the D-dimer assay as a first-line test for low to intermediate risk PE patients to reduce unnecessary CT scans.
This document summarizes a study evaluating the use of D-dimer testing and clinical risk algorithms to diagnose pulmonary embolism (PE). The study assessed 627 patients presenting with suspected PE who were stratified into low, intermediate, or high risk groups using the Geneva score. For low and intermediate risk patients, a quantitative D-dimer assay achieved 100% sensitivity and negative predictive value, correctly ruling out PE in 25% and 33% of cases respectively. The study supports using the D-dimer assay as a first-line test for low to intermediate risk PE patients to reduce unnecessary CT scans.
Assignment Pharmacoeconomics Fatma Adel SolimanAsia Smith
This document provides an overview of decision analysis and Markov modeling. It discusses the steps in decision analysis, including identifying the decision, specifying alternatives, drawing the decision tree structure, specifying costs/outcomes/probabilities, performing calculations, and conducting sensitivity analysis. The key steps in Markov modeling are also outlined, such as choosing health states, determining transitions between states, setting the cycle length/number, estimating transition probabilities, and calculating costs and outcomes. The advantages of Markov modeling include flexibility and modeling dynamic systems, while disadvantages include greater complexity and the memoryless assumption.
The document discusses target controlled infusion (TCI) for intravenous anesthesia. TCI uses pharmacokinetic models and computer controlled infusion pumps to maintain stable plasma or effect-site concentrations of anesthetic drugs. It achieves this by simulating drug concentrations based on dosage and continually adjusting the infusion rate. TCI allows titration of drugs to clinical effects and precise control of drug levels throughout a procedure. Accuracy depends on how well the pharmacokinetic model matches an individual patient. Future improvements include accounting for individual patient factors, multidrug interactions, and closed-loop systems.
This document discusses the importance of quality control for echocardiography interpretation and summarizes several studies on improving inter-observer reliability and accuracy. The first study found low agreement between experts in assessing diastolic function from the same patient data due to a lack of hierarchy for integrating parameters. A later study developed a consensus strategy to improve grading of aortic regurgitation severity from 70% to over 90% agreement. Another study reduced inter-observer variability in left ventricular ejection fraction measurements from echocardiograms by having readers review their measurements alongside cardiac MRI images. A final study showed quantitative measurements of right ventricular function improved accuracy and agreement compared to qualitative assessments alone. The studies demonstrate the value of quality improvement efforts like developing consensus guidelines
When designing a clinical study, a fundamental aspect is the sample size. In this article, we describe the rationale for sample size calculations, when it should be calculated and describe the components necessary to calculate it. For simple studies, standard formulae can be
used; however, for more advanced studies, it is generally necessary to use specialized statistical software programs and consult a biostatistician. Sample size calculations for non-randomized studies are also discussed and two clinical examples are used for illustration
850-0490-00 Rev A-Oscillatory MAP ABP Case Study Christina Mason
Mary Jones underwent a cardiac procedure and her blood pressure is being monitored. Her nurse noticed a difference between the mean arterial pressure (MAP) displayed on the monitor and the MAP she calculated using the usual (SBP + 2DBP)/3 formula. There are a few reasons why the values may differ: 1) The monitor displays the MAP directly measured by the cuff, while the formula provides an estimate. 2) True MAP can only be determined invasively, while noninvasive methods like oscillometry and auscultation may vary from invasive readings by up to 5 mmHg. 3) Research shows oscillometric MAP more closely matches invasive MAP than SBP or DBP. Clinicians can generally trust the MAP
This document outlines a study on jointly modeling multivariate longitudinal measures of hypertension (blood pressure and pulse rate) and time to develop cardiovascular disease among hypertensive outpatients in Ethiopia. The study aims to identify factors affecting changes in blood pressure and pulse rate over time as well as time to develop cardiovascular complications. The study will collect longitudinal data on blood pressure, pulse rate and cardiovascular events from 178 hypertensive patients and analyze it using joint longitudinal-survival models. Preliminary results show changes in blood pressure and pulse rate over time differ between patients who did and did not develop cardiovascular events. Key factors like diabetes, family history of hypertension and clinical stage of hypertension affect both longitudinal outcomes and survival.
The study reviewed the relationship between dietary and supplemental antioxidants and prostate cancer risk. Antioxidants examined included vitamin E, selenium, vitamin C, carotenoids, and polyphenols from coffee and tea. The evidence for effects of vitamin E and selenium on prostate cancer risk was inconsistent. While some studies found protective effects of selenium at low baseline levels, others found no effect. Studies of vitamin C, carotenoids, and polyphenols like green tea provided inconclusive or no evidence of relationships with prostate cancer risk.
This study evaluated the performance of three pneumonia severity scores - CURB-65, CRB-65, and CURB-age - in predicting 30-day mortality in 559 adult inpatients with community-acquired pneumonia. The study found that CURB-age stratified patients into risk groups that were most closely associated with mortality outcomes. Specifically, CURB-age identified more low-risk patients who had lower mortality than CRB-65, and sorted more patients who died within 30 days into the high-risk group compared to the other scores. Analysis of receiver operating characteristics further indicated CURB-age had better ability to predict 30-day mortality compared to the other scores. The study concludes CURB-age may provide
This document discusses case-control studies and their design and analysis. It begins by defining case-control studies as observational studies where subjects are sampled based on disease presence or absence and their prior exposure is then determined. It describes key features, need, steps in design including case and control selection. It then covers statistical analysis including odds ratios to measure risk associated with exposure and interpretations. It discusses effect modification and confounding, and analytical tools like stratification and multivariate modeling to control for confounding.
Epidemiological Approaches for Evaluation of diagnostic tests.pptxBhoj Raj Singh
Diagnosis of a disease or a problem is the first step towards solution/ treatment. Clinical Diagnosis or Provisional Diagnosis is the first step in diagnosis and is done after a physical examination of the patient by a clinician. Clinical diagnosis may or may not be true and to reach Final diagnosis Laboratory Investigations using gross and microscopic pathological observations and determining the disease indicators are required. The diagnostic tests may be Non-dichotomous Diagnostic Tests (when continuous values are given by the test in a range starting from sub-normal to above-normal range) and Dichotomous Diagnostic Tests (when results are given either plus or minus, disease or no-disease). To make non- Dichotomous diagnostic test a Dichotomous one you need to establish the cut-off values based on reference values or Gold Standard test readings or with the use of Receiver operator characteristic (ROC) curves, Precision-Recall Curves, Likelihood Ratios, etc., and finally establishing statistical agreement (using Kappa values, Level of Agreement, χ2 Statistics) between the true diagnosis and laboratory diagnosis. Thereafter, the Accuracy, Precision, Bias, Sensitivity, Specificity, Positive Predictive value, and Negative Predictive value, of a diagnostic test are established for use in clinical practice. Diagnostic tests are also used to determine Prevalence (True prevalence, apparent prevalence) and Incidence of the disease to estimate the disease burden so that control measures can be implemented. There are several Phases in the development and use of a diagnostic assay starting from conceptualization of the diagnostic test, development and evaluation to determine flaws in diagnostic test use and Interpretation influencers. This presentation mainly deals with the epidemiological evaluation procedures for diagnostic tests.
HEART DISEASES PREDICTION USING MACHINE LEARNING ALGORITHMPoojaSri45
Implemented a machine learning project aimed at predicting heart diseases using various algorithms and techniques. Developed as a part of academic or professional endeavor, the project demonstrates proficiency in data preprocessing, feature selection, model training, and evaluation.
VALIDITY AND RELIABLITY OF A SCREENING TEST seminar 2.pptxShaliniPattanayak
A presentation shedding some insight into the tricky concepts of validity and reliability of any screening test, used in day-to-day lives, using easy and understandable language.
David Madigan MedicReS World Congress 2014MedicReS
This document discusses issues with published observational studies and potential ways to improve them. It notes that even small choices in study design can lead to a wide range of results. Applying different analysis methods to large healthcare databases still found many false positives and negatives. However, performance improved by tailoring analyses to specific outcomes and restricting to larger sample sizes. Self-controlled designs like case-crossover consistently outperformed traditional cohort and case-control designs. Further strategies like evaluating results against known null distributions may help address biases from unmeasured confounding.
This document summarizes a study evaluating the use of D-dimer testing and clinical risk algorithms to diagnose pulmonary embolism (PE). The study assessed 627 patients presenting with suspected PE who were stratified into low, intermediate, or high risk groups using the Geneva score. For low and intermediate risk patients, a quantitative D-dimer assay showed 100% sensitivity and negative predictive value, correctly ruling out PE in 25% and 33% of cases respectively. The study supports using the D-dimer assay as a first-line test for low to intermediate risk PE patients to reduce unnecessary CT scans.
This document summarizes a study evaluating the use of D-dimer testing and clinical risk algorithms to diagnose pulmonary embolism (PE). The study assessed 627 patients presenting with suspected PE who were stratified into low, intermediate, or high risk groups using the Geneva score. For low and intermediate risk patients, a quantitative D-dimer assay showed 100% sensitivity and negative predictive value, correctly ruling out PE in 25% and 33% of cases respectively. The study supports using the D-dimer assay as a first-line test for low to intermediate risk PE patients to reduce unnecessary CT scans.
This document summarizes a study evaluating the use of D-dimer testing and clinical risk algorithms to diagnose pulmonary embolism (PE). The study assessed 627 patients presenting with suspected PE who were stratified into low, intermediate, or high risk groups using the Geneva score. For low and intermediate risk patients, a quantitative D-dimer assay achieved 100% sensitivity and negative predictive value, correctly ruling out PE in 25% and 33% of cases respectively. The study supports using the D-dimer assay as a first-line test for low to intermediate risk PE patients to reduce unnecessary CT scans.
This document summarizes a study evaluating the use of D-dimer testing and clinical risk algorithms to diagnose pulmonary embolism (PE). The study assessed 627 patients presenting with suspected PE who were stratified into low, intermediate, or high risk groups using the Geneva score. For low and intermediate risk patients, a quantitative D-dimer assay achieved 100% sensitivity and negative predictive value, correctly ruling out PE in 25% and 33% of cases respectively. The study supports using the D-dimer assay as a first-line test for low to intermediate risk PE patients to reduce unnecessary CT scans.
This document summarizes a study evaluating the use of D-dimer testing and clinical risk algorithms to diagnose pulmonary embolism (PE). The study assessed 627 patients presenting with suspected PE who were stratified into low, intermediate, or high risk groups using the Geneva score. For low and intermediate risk patients, a quantitative D-dimer assay achieved 100% sensitivity and negative predictive value, correctly ruling out PE in 25% and 33% of cases respectively. The study supports using the D-dimer assay as a first-line test for low to intermediate risk PE patients to reduce unnecessary CT scans.
Assignment Pharmacoeconomics Fatma Adel SolimanAsia Smith
This document provides an overview of decision analysis and Markov modeling. It discusses the steps in decision analysis, including identifying the decision, specifying alternatives, drawing the decision tree structure, specifying costs/outcomes/probabilities, performing calculations, and conducting sensitivity analysis. The key steps in Markov modeling are also outlined, such as choosing health states, determining transitions between states, setting the cycle length/number, estimating transition probabilities, and calculating costs and outcomes. The advantages of Markov modeling include flexibility and modeling dynamic systems, while disadvantages include greater complexity and the memoryless assumption.
The document discusses target controlled infusion (TCI) for intravenous anesthesia. TCI uses pharmacokinetic models and computer controlled infusion pumps to maintain stable plasma or effect-site concentrations of anesthetic drugs. It achieves this by simulating drug concentrations based on dosage and continually adjusting the infusion rate. TCI allows titration of drugs to clinical effects and precise control of drug levels throughout a procedure. Accuracy depends on how well the pharmacokinetic model matches an individual patient. Future improvements include accounting for individual patient factors, multidrug interactions, and closed-loop systems.
1. Hidden Markov Models for Detecting
Changes in Health Outcomes and
Comparing Groups of Subjects
ZHANG ZEYANG
SEPTEMBER 2015
2. Acknowledgement
Final project for the degree
MSc. Computational Statistics and Machine Learning at UCL.
Many thanks to my supervisors:
Dr. David Barber (UCL)
Dr. Steven Barrett (GSK) and Dr. Maria Costa (GSK)
3. Background: COPD disease
Chronic Obstructive Pulmonary Disease (COPD) can be summarized as:
• A collection of lung diseases including chronic bronchitis and emphysema etc.
• A long-term condition that causes inflammation in the lungs, damaged lung tissue and
a narrowing of the airways, making breathing difficult.
• A life-threatening respiratory disease that is commonly seen both in the UK and
worldwide.
4. Background: COPD Exacerbations
An (acute) exacerbation of COPD is characterized as:
• Worsening of COPD symptoms ( dyspnea, cough, and/or sputum) beyond day-to-day
variations that usually last for a few days.
• Lack of a standardized, consistent and commonly accepted definition.
• The studies of the efficacies of new therapies on COPD have been hampered by the
difficulty in identifying and quantifying exacerbations.
• New approaches are being sought to better recognize and understand exacerbations.
5. Introduction
A Patient-Reported instrument has been employed to monitor the health of COPD
patients, in which the participating patients are divided into two treatment groups (
Drug A and Drug B) and all required to answer 14 questions in an electronic survey,
reflecting their
• Chest symptoms
• Cough and sputum symptoms
• Breathless symptoms
• General well-being
on a daily basis during the clinical trials (around 6 months). For each question, a patient
has to assign a score where a higher score indicates a more severe symptom.
6. Dataset
The sum of the 14 scores of each study day forms a time-series data for each patient.
Meanwhile, the clinical exacerbations of each patient over the same periods are also
recorded together with other individual information such as number of historical
exacerbations, treatment group etc.
7. Objectives
In a nutshell, this project aims to
1. Construct an accurate yet computationally efficient model for detecting COPD
exacerbations based on patients’ self-reported health scores.
2. Develop a systematic method to evaluate the model and benchmark the
detected results against the clinical exacerbations.
3. Based on the detected exacerbations, compare the treatment efficacies of
Drug A and Drug B at cohort level.
8. Hidden Markov Model (HMM)
The Hidden Markov Model (HMM),
which can be represented by a Direct
Acyclic Graph, is an unsupervised
machine learning model used in this
project to find exacerbations.
The HMM in our model consists of:
• Hidden variable h , that represents the
exacerbation status of each individual.
• Observed variable v , that represents the
reported health scores of each individual.
The most likely exacerbation status of an
individual can be inferred via the Viterbi
algorithm.
9. Evolution of Models & Results
Through restructuring and manipulating the HMM, we can adapt the model to
accommodate various assumptions of exacerbations to generate more satisfying results.
11. Evaluation: Precision, Recall Measures
An instrument from Information Retrieval is borrowed to evaluated the performance of the
HMM in detecting the exacerbations.
Clinical Exacerbation Not Clinical Exacerbation
Detected by HMM a (true positives) b (false positives) a + b
Not Detected by HMM c (false negatives) d (true negatives) c + d
a + c b + d a+b+c+d
In measurement of days, we have
Recall =
𝑎
𝑎+𝑐
= P(detected by model |clinical exacerbation)
Precision =
𝑎
𝑎+𝑏
= P(clinical exacerbation |detected by model)
12. Evaluation: Composite Measure - F
Both high Recall and high Precision values are desired; however, it is easy to see that
there is generally a trade-off between Recall and Precision.
A sensitive model tends to over-identify exacerbations, thus generating a high Recall
and low Precision and vice versa.
Therefore, based on Recall and Precision values, a composite measure is created
F-measure =
1
π
1
𝑅
+(1−π)(
1
𝑃
)
,
to strike a balance between these two indices, where π here represents the weightage
we assigned to Recall. A higher F-measure indicates a better performance of the
model in detecting the exacerbations.
13. Evaluation: Parameters
When restructuring and designing HMM, a series of
parameters were created .
The parameters govern the sensitivity of the model in
detecting the exacerbations and embody some basic
assumptions of an exacerbation.
For instance, α represents our belief of how likely a
patient enters an exacerbation from a non-
exacerbation day in general. A Recall-Precision Curve
can be plotted for various values of α.
14. Evaluation: Parameters Tuning
In this project, we believe that the clinically identified exacerbations are not the
‘universal truth’. A significant amount of exacerbations could have been missed by the
clinicians as the patients may not always approach on time when the symptoms
deteriorate.
Therefore, we are more interested in creating a model that can successfully identify
most clinically found exacerbations (high Recall), while being more tolerant to over-
detection of exacerbations (low Precision).
We thus adjust our HMM to the ‘ideal sensitivity’ by setting the parameters to the
values that generate the highest F-measure for π=0.80 and π=0.85 respectively.
15. Evaluation: Comparison of Performances
When comparing with an existing method (Method X), the performances of our HMM
seem encouraging.
16. Cohort Level Analysis: Regression
Based on the exacerbations detected by our HMM (at π=0.80 and π=0.85 respectively), we can
analyze the impacts different factors (treatment groups, historical exacerbations) have on the
exacerbation frequencies of the COPD patients.
It is natural to assume that a patient’s frequency of exacerbation follows a Poisson distribution –
Poisson(t·λ). Hence, a Poisson Regression is then fitted to the data. Using a log link, we have
θ ’β = log(t) + log(λ)
where θ ’ represents the predictor variables (or independent variables, regressors), such as
historical exacerbations and treatment groups.
19. Further Improvements
Overcome the constraints of the data.
Improve design of the HMM (3-order HMM, heterogeneous transition matrix etc.).
When evaluating, take measurements in events of exacerbations instead of
measurements in days for better accuracy.
People are normally diagnosed in their 40s or 50s.
From point 2 to point 3, it makes it difficult when we test new drugs or monitor the progression of COPD disease. Better identify better quantify exacerbations.
I will avoid the details of the algorithms. The Viterbi considers transition and emission distributions.
However, the diagram above is just a basic HMM. In the context of this project, we adapt redesign the HMM.
Bear in mind, this results are generated based on the assumed values of parameters in Viterbi.
To what extent are they in line (in agreement ) with clinical exacerbations??????
It is up to us how sensitive we want the model to be. A composite measure is then desired.