Regression for prediction
Students learn to treat regression models as tools for estimating outcomes and risks for new patients.
Module 2
Supervised Learning for Clinical and Health Data
This module moves from foundational prediction thinking into supervised learning methods used for clinical and health-data problems: regression, logistic classification, k-nearest neighbours, decision trees and modelling pipelines.
Module aim
Learn core supervised models for health prediction.
The purpose of this module is to connect familiar statistical models and beginner machine learning methods to prediction tasks in medical and public health datasets.
5
Lessons
R
Coding labs
Preparing
Module status
Supervised
Learning focus
Clinical classification
The module introduces probability-based classification, thresholds and supervised learning for health outcomes.
Model workflow
Each method is placed inside a practical clinical-data pipeline rather than taught as an isolated algorithm.
Module lessons
Study the supervised learning sequence in order.
Each lesson introduces a different supervised learning idea, then connects it back to clinical prediction, model interpretation and responsible use with health data.
2.1
Lesson 2.1
Regression as a prediction model
Learn how regression can be used as a prediction tool, not only as an explanatory model, and how fitted values become risk estimates.
Open lesson →
2.2
Lesson 2.2
Logistic regression as a classifier
Understand logistic regression as a clinical classification model for binary outcomes, predicted probabilities and risk thresholds.
Open lesson →
2.3
Lesson 2.3
k-nearest neighbours and distance-based learning
Study how distance-based prediction works, why scaling matters and why local neighbourhood methods can be sensitive in health data.
Open lesson →
2.4
Lesson 2.4
Decision trees and rule-based prediction
Learn how decision trees split data into clinical decision rules, why they are interpretable and why they can easily overfit.
Open lesson →
2.5
Lesson 2.5
Model pipelines for clinical datasets
Bring supervised learning together through a clinical modelling pipeline: preprocessing, fitting, prediction, validation and reporting.
Open lesson →
Learning route
Use Module 2 before studying validation and performance.
Module 3 assumes that students understand how supervised models are trained, how they generate predictions and why prediction pipelines must be built before performance is judged.
Continue to Module 3 →Course pathway
Return to the full ML in Biostatistics course.
Use the course homepage to move between modules, case studies, datasets, scripts and the full medical machine learning pathway.
Back to course →