NIHR Maudsley Biomedical Research Centre PhD studentships – Translational Research in Mental Health

LSUH-2.04 Using electronic health record data for clinical risk prediction: Application to liver disease

Primary Supervisor: Dr Katherine Morley
Email: katherine.morley[at]kcl.ac.uk
Second Supervisor: Professor Robert Stewart
Email: robert.stewart[at]kcl.ac.uk

Project Description

Background: Electronic health record (EHR) systems contain data from clinical practice for millions of patients across multiple healthcare domains providing a valuable source of data for developing clinical risk prediction tools. However, these data are observational and can be incomplete. Linking data collected via “traditional” epidemiological surveys to EHR data provides an opportunity to compare how patient risk factors, diagnoses, and treatment are characterised in different data sources, and can thus inform both the development of risk prediction tools and clinical practice. This project will use this approach to develop a risk prediction tool for liver disease, an important consequence of lifestyle substance use.

Aim: To combine electronic health records (EHR) with data collected via “traditional” epidemiological surveys to develop computational tools that can assist clinical staff to identify addiction services clients at high risk of liver disease.

Design:
Project 1: Survey a large sample of addiction services clients and link their questionnaire responses to their EHR. The questions will cover lifestyle substance use and associated factors that affect risk of liver disease. The linked data set will be used to determine what is not being captured in the EHR, leading to improvements in extraction of information from EHR data and clinical recording.
Project 2: Using the linked survey-EHR data set from Project 1, develop a risk prediction model for liver disease. The model will initially be developed using only data available in the EHR; further analyses will investigate whether information not currently well-captured in the EHR can improve model performance by including of variables from the survey data.
Project 3: Based on the model from Project 2, develop computer-based decision-support tools that can be used by clinical staff to: (i) estimate an individual client’s risk of developing liver disease; (ii) translate statistical estimates into clinically meaningful information that supports patient education (e.g. graphical tools); (iii) inform discussions between staff and clients about treatment options. Input from staff and clients will be sought via interviews and focus groups to ensure these tools are relevant and useful from their perspectives.

Skills learned: large-scale data analysis; advanced statistical analysis.

To Apply

Go to http://www.maudsleybrc.nihr.ac.uk/training/current-opportunities-events/phd-studentships/

Questions?

Contact Dr Katherine Morley – katherine.morley[at]kcl.ac.uk

Leave a Reply