Using AI to Determine Factors Related to Frequent HC Attendance v1
Research type
Research Study
Full title
Using Artificial Intelligence to Determine the Factors Associated with Frequent Attendance at Healthcare Service Facilities.
IRAS ID
329712
Contact name
S. Josephine Pravinkumar
Contact email
Sponsor organisation
NHS Lanarkshire
Duration of Study in the UK
0 years, 11 months, 30 days
Research summary
This research targets the use of state-of-the-art artificial intelligence (AI) models to support the prediction and analysis of Frequent Attenders. Frequent Attenders (FA) are patients who attend a health care facility multiple times per year. The specific objectives are: to conduct a rapid literature review to understand the state-of-the-art in FA analysis and risk factors and interventions previously studied; to apply explainable AI prediction model that incorporates causality structure learning to support prediction and analysis of FA in healthcare systems and carry out technical validations.
This research will have access to datasets to train and validate the AI models. This will require a pseudonymised dataset that covers the known risk factors of patients attending healthcare facilities from previous studies, including clinical, social care and demographic information.
The core AI model will allow analysts to gain quantitative insight of causal relations between the targeted FA risk and its causes and predict potential outcomes for hypothetical interventions. We will apply the model to identify the true cause of high resource use by analysing a wide range of data attributions from the medical records and social/other data, and study the effect of potential interventions to reduce the risks. The causal model will abstract away from features that are only spuriously correlated with the targeted FA risk, hence contributing to improved clarity in the explanation. Technical validation will be performed as a confirmatory study, which is designed to create an AI model to model the relations between the FA and its known risk factors.
Based on the outcomes and conclusion from this project, future work beyond will be designed to explore a wider range of data variables and look for new causal risk factors. Understanding these FA factors will better inform clinical teams and help those patients with most need.REC name
Wales REC 4
REC reference
24/WA/0041
Date of REC Opinion
15 Feb 2024
REC opinion
Favourable Opinion