Back

Value of using artificial intelligence derived clusters by health and social care need in Primary Care: A qualitative interview study

Holt, S. Z.; Simpson, G.; Santer, M.; Everitt, H.; Farmer, A.; Zhou, K.; Qian, Z.; Davies, F.; Dambha-Miller, H.; Morrison, L.

2024-10-19 primary care research
10.1101/2024.10.17.24315657 medRxiv
Show abstract

PurposePeople living with MLTCs attending consultations in primary care frequently have unmet social care needs (SCNs), which can be challenging to identify and address. Artificial intelligence (AI) derived clusters could help to identify patients at risk of SCNs. Understanding the views of people living with MLTCs and those involved in their care can help inform the design of effective interventions informed by AI-derived clusters to address SCNs. MethodsQualitative study using semi-structured online and telephone interviews with 24 people living with MLTCs and 20 people involved in the care of MLTCs. Interviews were analysed using Reflexive Thematic Analysis. ResultsPrimary care was viewed as an appropriate place to have conversations about SCNs. However, participants felt health care professionals lack capacity to have these conversations and to identify sources of support. AI was perceived as a tool that could potentially increase capacity for this but only when supplemented with effective, clinical conversations. Interventions harnessing AI should be brief, be easy to use and remain relevant over time, to ensure no additional burden on clinical capacity. Interventions must allow flexibility to be used by multidisciplinary teams within primary care, frame messages positively and facilitate conversations that remain patient centered. ConclusionOur findings suggest that AI-derived clusters to identify and support SCNs in primary care have perceived value, but there were some concerns including the need to consider personal context. AI derived clusters can be used as a tool to inform and prioritise effective clinical conversations. Conference names, dates and locations for any prior presentationsO_LIBritish Journal of General Practice Research Conference, March 2024, London. C_LIO_LISchool of Academic Primary Care Southwest Conference, March 2024. Cardiff. C_LIO_LIFaculty of Medicine Research Conference, June 2024. Southampton. C_LI

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.