HSMA: Machine Learning Module – The Healthcare Service Analytics & Decision Intelligence Atlas

Summary

This is a collection of slides, session recordings, code examples and exercises from the machine learning module of the Health Service Modelling Associates Programme, covering 30 hours of content. It teaches classification and regression with scikit-learn and tensorflow, including machine learning ethics, assessing model performance, explainable AI, model optimisation, reinforcement learning and synthetic data.

Rating: Gold

Module written by an experienced team of academics with over a decade in the operational research and data science space, with input from an ex-NHS data scientist.

This module serves as an introduction to machine learning methods in Python. Students make use of packages including scikit-learn and tensorflow to train and assess machine learning models that can classify data into groups or predict a numeric value (e.g. length of stay).

It was delivered to the sixth cohort of the Health Service Modelling Associates (HSMA) programme, a training course aimed at analysts, clinicians, managers and other people working in the NHS and related healthcare organisations.

It covers

Key machine learning concepts, such as train-test splits, accuracy, the bias-variance trade-off, and more
Preparing data for machine learning with one-hot encoding and feature engineering
Machine learning ethics
Tackling classification problems with machine learning
Tackling regression problems (numeric predictions) with machine learning
The theory of, and how to apply, a wide range of algorithms using the scikit-learn package: logistic regression, decision trees, random forests, XGBoost, LightGBM, and more
Creating ensemble models
Assessing the performance of machine learning models in classification and regression problems with accuracy, precision, recall, f1 score, confusion matrices, the receiver operating curve, the ROC AUC, and more
How to explain the activity of black-box models with techniques such as individual conditional expectation (ICE) plots, partial dependence plots (PDPs), SHAP values, and more
Calibrating models for obtaining prediction probabilities
Optimising model parameters efficiently with grid search and the optuna framework
Optimising the selected features with a range of feature selection approaches
Creating neural networks for classification problems with the tensorflow package
The principles of reinforcement learning
Creating synthetic data with SMOTE

All slides, session recordings, code examples, exercises and exercise solutions are available to access on the module page, covering 30 hours of content.

View the module

To view the module, click on the image above or go to https://hsma.co.uk/hsma_content/modules/current_module_details/4_machine_learning.html.

Tip

The guided code examples from this module, which talk through the how and why of each step, are also collected in the HSMA machine learning notebooks:

https://machinelearning.hsma.co.uk

Tip

Brand new to Python? Consider starting with the HSMA book of Python first:

https://python.hsma.co.uk