Assembling the data jigsaw: powering robust population research in MSK disease

This project will bring together existing and innovative health and social care data in Greater Manchester to fill gaps in knowledge around living with a musculoskeletal (MSK) condition.

Research using routinely collected health data can answer key questions, but in MSK disease, progress in answering these is often hampered by availability of the right data. This can be for three key reasons. First, despite all the necessary data items being in existence, these data items are not yet available altogether in the same dataset. Second, data are not formatted in a way that makes them amenable for research. And third, the data items do not yet exist within the health and social care system.  This project will address these challenges by assembling a data jigsaw: collecting, processing and linking data from different sources.

The project will take place in Salford, where the hospital has a well-established electronic health record (EHR). All data assembly work will be designed to be as transferrable as possible to other localities. The project team will build on de-identified primary and secondary care EHR data, which they will link with data from social care and social media with explicit consent. Machine learning will be used to extract data automatically from outpatient letters, providing detailed information on diagnosis, medication, disease severity, daily activities and quality of life. The project will develop methods of collecting structured data from rheumatologists within the EHR and establish touchscreens in outpatient waiting areas to allow patients to become active contributors to musculoskeletal research. The assembled data will be used to answer questions about the prevalence and impact of MSK disease and the effectiveness and safety of pain-relieving medications. Findings will be communicated to policymakers through Versus Arthritis and other routes. Future research will benefit from the resources made available by the project: a suite of open source machine learning models and algorithms for extracting data from free text, and a MSK Common Data Model to support harmonised collection of research quality data within routine NHS rheumatology practice.