Bristol Medical School

About the course

Health research is increasingly turning to high-throughput molecular datasets (also known as ‘omic’ datasets) to discover novel biomarkers of disease risk and outcome. Unfortunately, the size and complexity of these datasets makes them difficult to manage and prone to many pitfalls.

In this course, we introduce you to the latest approaches from data science for interpreting and extracting useful and reliable biomarkers from these challenging datasets. It aims provide:

• An overview of the principles and methods of epidemiology and data science that are relevant to high-throughput omic studies
• The knowledge and skills necessary to design and utilize population-based omic studies to gain insight and to derive robust biomarkers of exposures and health outcomes

Attendees may have a background in epidemiology, genetics, statistics, public health or a clinical speciality. A basic knowledge of epidemiology is required and some understanding of molecular epidemiology terminology and machine learning would be advantageous. Practical knowledge of R is required as students will be processing large omic datasets in practical sessions.

What this course will cover

• Examples of published omic analyses and models for epidemiological and medical applications

• Statistical methods for preprocessing, discovering patterns and testing associations in omic datasets

• Interpreting the biological relevance of omic patterns and associations

• Estimating the heritability and proportion of variation explained by omic data

• Approaches from machine learning for deriving reliable omic biomarkers for indexing exposures and predicting health outcomes

• Application and interpretation of appropriate metrics for evaluating biomarker performance

• Ethical challenges of developing, interpreting and applying molecular biomarkers

How to join

Please note, there are pre-requisites for enrolling on this course – details can be found on the full course listing linked below.

Sign up via the University of Bristol Medical School website

About the course

This course introduces the analysis of survival-time outcomes, consisting of the time each person “survives” until some specified event occurs. This course aims to give students:

• A grounding in the theory behind the methods most commonly used to analyse rates and survival-time data
• Extensive hands-on experience of their application in Stata software

The course is intended for researchers and analysts who wish to analyse and understand survival-time data (time to event data). The course focuses on popular methods of analysing these types of data, mainly Poisson and Cox regression.

What this course will cover

• Definition of rates and the relation between risks and rates
• Manipulating person-time data in Stata using the st commands
• Analysis of rates using Mantel-Haenszel methods and Poisson regression
• Splitting follow-up time to allow for exposures that change with time
• Introduction to survival analysis
• Log rank tests and Cox proportional hazards regression
• Testing for proportional hazards and modelling non-proportional hazards
• Survival-time data management
• Choosing between survival analysis methods.

Please note that ‘repeated measures’ analyses, in which multiple events or measurements are recorded in the same person over time, are not covered in this course.

How to join

Please note, there are pre-requisites for enrolling on this course – details can be found on the full course listing linked below.

Sign up via the University of Bristol Medical School website