Further Topics in Health Data Science

Module code: MD7476

This module will extend your knowledge to more advanced data science topics as well as exposing you to a wider variety of topics. 

You will take one mandatory teaching week (Advanced Data Science). You will also select one optional teaching week from a range of different topics. The choice will vary from year to year. 

Advanced Data Science

You will develop your software skills. You will learn how to organise larger code bases, learn more advanced git functionality and learn how to turn single use code scripts into software for others to use. You will also learn how to use high performance computing (super computers) to run larger analyses.

Currently some of the optional topics offered are:

Advanced Survival Analysis

This option builds on the methods covered in the (basic) survival analysis module. The methods you cover will include the use of counting process notation, time-dependent covariates, modelling time-dependent effects, multiple events, cure models and flexible parametric survival models. There will be lectures, group practical work and presentations.

Advanced Evidence Synthesis

The option will start with a review of basic meta-analysis for different outcomes. You will then use a Bayesian implementation of meta-analysis and meta-regression methods and discuss the advantages of using this approach. More advanced methods such as network meta-analysis, a very important generalisation of the standard meta-analysis model increasingly used in HTA when more than two alternative treatment options exist, meta-analysis methods for the evaluation of diagnostic tests, and meta-analysis methods for synthesis of data on multiple outcomes will be introduced.

Statistical Genetics

In this option you will be introduced to the basic biology and genetics, and statistical and epidemiological concepts that underpin genetic epidemiology. Practical sessions will give you experience of running and interpreting a genetic association analysis including multiple genetic variants. The fundamental concepts underpinning analyses of family-based data, recurrence risk ratios, heritability, and variance components analysis will be covered. The use of genetic data in causal inference through “Mendelian Randomisation” will be considered.

Decision Modelling

This option will start by outlining the motivation for and principles of decision modelling. Both clinical decision models (weighing up benefits and side-effects of treatments) and economic decision models (assessing cost-effectiveness) will be covered. Simple decision trees and then Markov models will be described. After the basic principles of decision modelling are covered, methods for estimating individual parameters for such models will be considered. This includes the integration of meta-analyses with decision models. Given there is usually (considerable) uncertainty in any decision problem, value of information methods, to assess whether conducting further studies to reduce this uncertainty would provide good value for money, will be presented briefly.

Statistical Programming in SAS

SAS is an extensive software package used in the pharmaceutical industry, clinical trials units and by some researchers. It has historically been used as the recommended software for data management, data reporting and data analysis of clinical trials. This option will give you the opportunity to learn to use SAS for data handling and statistical analysis through an introductory lecture, a self-learning booklet and support sessions.

Student Selected Reading

For this option you will identify a topic related to medical statistics and find a published article to act as a starting point. Under the supervision of one of the staff, you read around the topic and prepare a report. We will be happy to help you select the topic and the initial reference.