Predicting Recovery in acute Ischaemic Stroke using Machine learning (PRISM)

Summary

Stroke is the largest cause of adult disability and a leading cause of death worldwide. Treatments to address the causative blocked arteries in the brain include both clot-busting drugs (thrombolysis) and mechanical clot removal (thrombectomy). These have transformed patient outcomes but deciding which therapy is best for each patient is currently imprecise.

Strategies to improve the decision-making process are crucial given the time-sensitive nature of stroke treatments and the large disparities in healthcare provision across regions and diverse patient groups. Brain imaging is a key tool in diagnosing and selecting patients for the treatments above.

Project aims

This project focuses on an innovative integration of multimodality clinical and imaging data to improve patient selection and outcome prediction after an acute stroke. It uses cutting-edge machine learning to improve the precision and speed of clinical decision-making. Leveraging deep learning, the project will integrate several data types—including clinical information, multiple CT scans per subject and blood tests into a multi-modality predictive model.

The model should provide superior prediction than currently possible by considering key factors such as age and severity of the stroke alongside brain imaging. After further validation, this work could also support the customisation of treatment plans tailored to individual needs, which should decrease healthcare inequalities and enhance the effectiveness of stroke treatments across diverse populations. 

A pilot phase of the study, funded by the University of Cambridge C2D3 Accelerate Science programme, has provisioned a successful multimodal data pipeline from Addenbrooke’s Hospital to the High-Performance Computing cluster at Cambridge University. Data on over 700 subjects is ready for analysis, with plans for validation across larger datasets regionally and nationally via the BHF Data Science Centre.

The successful PhD student will gain exceptional experience in stroke medicine, brain imaging, machine learning and data science through frequent interactions with leading researchers and clinicians. The PI has a strong track record in helping students obtain PhDs on time with successful subsequent placements in academia, medicine and industry.

Contact details

Professor James Rudd - jhfr2@cam.ac.uk

Opportunities

This project is open to applicants who want to do a:

• PhD