The application of Multi-Task Learning (MTL) to the problem of predicting stroke patient rehabilitation outcomes to support rehabilitation programme design on the basis of the Barthel Index (BI) which is commonly used multi-component scoring metric for assessing the independence of patients with stroke in terms of their ability to complete activities of daily living. The models predict the components of the BI and demonstrate benefits of MTL in a health context to predict patient profiles.
Addressing target audiences and expressing needs
- We are sharing our knowledge
- Collaboration
We are specifically looking for clinical and academic research partners who are interested in stroke rehabiliation and reintegration and who wish to work with us to develop out and validate this solution so that this fundamental research can have a real beneficial impact on society.
- Others/ No specific audience
- Research and Technology Organisations
- Academia/ Universities
R&D, Technology and Innovation aspects
The current model is proof of concept AI predictice model that has been tested in lab conditions on retrospective data. The next steps is stress test the model for biases, integrate AI explainability methods, and to validate the model on mutiple datasets and then progess to a prospective data study.
Assuming relevant data is available it is possible to retrain the model to adapt it to different populations, and even to adapt it to different prediction tasks (e.g., other medicial indexes apart from the Barthel Index).
The product is a AI prediction model and so it can be easily replicated and deployed on multiple sites as required. However, prior to this happening its performance should be validated across multiple datasets with patients from multiple ethnicities to test for bias and stability across different demongraphics and ethnicities.
Given the proliferation of multi-component assessment tools across medicine the use of multi-task learning to improve predictions of patient outcomes has the potential to have a significant impact across many areas of health, thereby adding value in all of these areas for stakeholders, and society.
