Recommendation Systems for Self-regulated Personalised Care Pathways
Research type
Research Study
Full title
To develop a prediction and recommendation system to enhance Healthshare digital pathways screening and self care tool for Musculoskeletal patients using machine learning techniques based on deep neural networks, graph neural networks and natural language processing.
IRAS ID
329367
Contact name
Genghui Qu
Contact email
Sponsor organisation
University of Essex
Duration of Study in the UK
3 years, 7 months, 15 days
Research summary
Shared decision making (between physician, therapist and patient) and accurately informing patients as to their treatment choices is of significant importance to the NHS in what is an environment of increasing costs, limited resources and challenges in consistently delivering high-quality healthcare. Intelligent reliable healthcare recommendation systems have become a hot topic in healthcare management application research due to increasing demand for accessible self-regulated, self-directed, and patient centric therapeutics, health and social care. However, interpretable recommendation system with dynamic recommender capability haven't been employed in most medical areas and scenarios. Therefore, this project will be carried out. I will collect anonymized patient data from NHS SystemOne and Healthshare (a private healthcare provider) digital database and then train and deploy a recommendation system to address a need for an intelligent data-driven screening, digital treatment / therapy recommendation system for more accurately assessing digital pathways and clinical requirements for MSK patients and optimising care management recommendations for self-care (digital pathways) MSK patients. The models will address critical data centric research challenges around how to handle data sparsity, uncertainties, dynamic (changing information), and model interpretability. The models will be optimized to recommend personalized / profile specific therapy and treatment needs. A tailored evaluation framework will be designed for evaluating proposed models based on a number quantitative and qualitative metrics. The research will initially focus on self-regulated musculoskeletal conditions as 40% musculoskeletal issues can be self-managed with exercises and lifestyle changes.
REC name
London - Brent Research Ethics Committee
REC reference
23/PR/0867
Date of REC Opinion
30 Nov 2023
REC opinion
Further Information Favourable Opinion