AI assisted triage of UK patients in mental health care services: a qualitative focus group study of patients' attitudes.

backgroundThe referral process between healthcare services can be complex, especially in psychiatry, leading to significant delays and 'hidden waiting lists'. Digital approaches may be helpful. The CHRONOSIG (CHRONOlogical SIGnature) project aims to improve the referral and triage process by applying machine learning (ML) technology to information in electronic health records. We used a focus group methodology to ascertain the views of patients and participants on using CHRONOSIG and similar digital approaches to support decision making in triaging referrals in difficult to treat depression, and the potential benefits and disadvantages of such an approach.

methodsA lived experience participant focus group (N = 16) was held on 25th September, 2024, with a lived experience chair. Participants were recruited by convenience sampling. Data were analysed thematically and managed using the Framework method, with double coding of transcripts, and reported using COREQ guidelines.

resultsMain themes from the analysis were: (i) the complexity of mental health needs assessments; (ii) challenges in the current mental health system; (iii) general challenges of using a computer/artificial intelligence based tool for risk prediction and clinical decision support; (iv) differences and similarities in using a computer-based prediction tool in mental health vs. in physical health; (v) possible benefits and harms; (vi) factors to consider in the future.

conclusionsPatient engagement is a key challenge for digital tools in mental health, but previous studies in digital decision support tools have focussed on clinician feedback. In this study we ascertained the views of lived experience participants in mental healthcare triage and referral in difficult to treat depression. Participants identified delays, errors and confusion in the referral process and expressed positive views on the ability of the CHRONOSIG tool to help to improve waiting times and time spent between services, particularly when used as an addition to a high-quality clinical consultation. In many countries there are shortfalls in mental health care provision with increasing waits in both recorded and unrecorded waiting lists. This study supports a potential route to improve these processes; by more accurately and efficiently identifying the needs of patients and matching these to suitable services and research opportunities.