Teacher(s)
Penaloza-Baeza Andrea; Sawadogo Kiswendsida Clovis (coordinator);
Language
French
Main themes
The first part addresses the use and interpretation of diagnostic tests. That includes the basic characteristics (sensitivity, specificity, predictive value) ; the potential biases ; the ROC curves ; the Bayesian analysis ; the threshold probabilities. The second part analyzes the expected outcomes from a test or a treatment : utility concept, life expectancy. The third part includes clinical reasoning, decision-making, and cost-efficacy. These concepts are illustrated by clinical exemples taken from the diagnostic procedures used in medicine.
Learning outcomes
At the end of this learning unit, the student is able to : | |
| 1 | To teach the rationale of the diagnostic procedure and the basis of clinical reasoning, using methods taken from epidemiology and statistics. |
Content
This learning unit provides:
- on the one hand, essential statistical tools for estimating the time to an event (death, recurrence, disease) and to understand the factors influencing it, enabling the estimation of survival probability and the comparison of the effectiveness of treatments or interventions, even with incomplete data (censoring), which is crucial for clinical research, epidemiology and the evaluation of health policies. It helps to quantify the impact of early diagnosis and innovative treatments, and to improve the coordination of patient pathways.
- on the other hand, crucial skills for assessing the validity, reliability and utility of screening tools, including an understanding of concepts such as sensitivity, specificity and predictive value, which are essential for making informed decisions about their implementation and the interpretation of results, thereby reducing the risk of errors (false positives/negatives) and optimising the management of public health programmes.
Contents
I. Survival analysis
I.1 Fundamental concepts
I.2 Survival and hazard
I.3 Kaplan-Meier estimate
I.4 Log-rank test
I.5 Cox regression
I.6 Sample size for a survival study
II. Diagnostic Tests
II.1 Definitions of normality
II.2 Diagnostic qualities of tests
II.3 Discriminatory power of tests
II.4 ROC curves
II.5 Predictive values and the use of odds
II.6 Improving risk prediction with a new model
II.7 Decision trees
- on the one hand, essential statistical tools for estimating the time to an event (death, recurrence, disease) and to understand the factors influencing it, enabling the estimation of survival probability and the comparison of the effectiveness of treatments or interventions, even with incomplete data (censoring), which is crucial for clinical research, epidemiology and the evaluation of health policies. It helps to quantify the impact of early diagnosis and innovative treatments, and to improve the coordination of patient pathways.
- on the other hand, crucial skills for assessing the validity, reliability and utility of screening tools, including an understanding of concepts such as sensitivity, specificity and predictive value, which are essential for making informed decisions about their implementation and the interpretation of results, thereby reducing the risk of errors (false positives/negatives) and optimising the management of public health programmes.
Contents
I. Survival analysis
I.1 Fundamental concepts
I.2 Survival and hazard
I.3 Kaplan-Meier estimate
I.4 Log-rank test
I.5 Cox regression
I.6 Sample size for a survival study
II. Diagnostic Tests
II.1 Definitions of normality
II.2 Diagnostic qualities of tests
II.3 Discriminatory power of tests
II.4 ROC curves
II.5 Predictive values and the use of odds
II.6 Improving risk prediction with a new model
II.7 Decision trees
Teaching methods
Lecture illustrated with real-life examples through tutorials
Evaluation methods
A closed-book written examination focusing on the practical application of the methodological concepts covered in the course, as practised during the tutorials
The mark will be split between the two parts of the course: 50 per cent for the ‘Survival Analysis’ section and 50 per cent for the ‘Diagnostic Tests’ section.
Decimals will be rounded down to the nearest whole number.
AI may be used as a revision aid and support tool, but is effectively excluded from the final assessment.
The mark will be split between the two parts of the course: 50 per cent for the ‘Survival Analysis’ section and 50 per cent for the ‘Diagnostic Tests’ section.
Decimals will be rounded down to the nearest whole number.
AI may be used as a revision aid and support tool, but is effectively excluded from the final assessment.
Other information
Written examination. References : 1. Sox HC, Medical decision making, Butterworths ; 2. Grenier B, Evaluation de la décision médicale, Masson ; 3. Kassirer JP et Kopelman RI, Learning clinical reasoning, Williams & Wilkins ; 4. Friedland DJ et al, Evidence-based medicine, Lange ; 5. Weinstein, Clinical decision analysis, Saunders.
Online resources
The course slides and reference materials are available on Moodle
Bibliography
Références :
Eric Vittinghoff • David V. Glidden • Stephen C. Shiboski • Charles E. McCulloch. Regression Methods in Biostatistics. Linear, Logistic, Survival, and Repeated Measures Models. Springer, 2012 (2nd Edition) (Pages 54-62)
Sox H. C. , Medical decision making, Butterworths Grenier B. , décision médicale, Masson Weinstein, Clinical decision analysis, Sauders
Eric Vittinghoff • David V. Glidden • Stephen C. Shiboski • Charles E. McCulloch. Regression Methods in Biostatistics. Linear, Logistic, Survival, and Repeated Measures Models. Springer, 2012 (2nd Edition) (Pages 54-62)
Sox H. C. , Medical decision making, Butterworths Grenier B. , décision médicale, Masson Weinstein, Clinical decision analysis, Sauders
Teaching materials
- Mis à disposition sur Moodle
Faculty or entity
