The value of an AI decision support tool for community mental health services in Greater Manchester

Summary and Key objectives
Mental illness accounts for 23% of the UK illness burden with estimated societal costs of over £100 billion per year (Baker and Kirk-Wade, 2023). Three quarters of service users waiting for care experience a mental health crisis due to treatment delays leading to hospital admissions and increased suicide risk that are distressing for service users and relatives, contribute to increased risk of mortality, and are costly to the NHS (Royal College of Psychiatrists, 2022).

GMMH has around 10,000 people on their CMHT caseloads. CMHT staff manage large caseloads and have limited time to identify people at highest risk of using crisis services, which lead to reactive rather than preventative interventions. Over a 12-month period (June 2020 to May 2021), 2767 crisis events were recorded amongst people on CMHT caseloads which was similar to the previous year (GMMH 2022a). This resulted in high costs associated in crises care, increased burden on care coordinators, while causing significant impact on an individual’s journey to recovery.

Recent research has shown that Artificial intelligence (AI) technologies can be used in the development of prediction, detection and treatment solutions for mental health care (D’Alfonso 2020; Graham et al., 2019) that can significantly reduce the burden of staff and improve the management of mental health crises. However, although it is estimated that there are over 10,000 mental health AI applications currently available, only as few as 2% of these are supported by empirical evidence on the efficacy and acceptability of these applications (Lau et al., 2020). Within the English NHS there are currently limited AI technologies available to NHS mental health providers that can help identify and support the care of service users.

With growing evidence and guidance that using AI technologies can improve care while reducing the burden on staff, the Greater Manchester Mental Health NHS Foundation Trust (GMMH) has partnered with the University of Manchester and Holmusk UK, a data science company, to examine the value (the efficacy and acceptability) of an AI-enabled decision support tool called MaST within their adult Community Mental Health Teams (CMHTs). MaST is already being used in adult (18 to 65) CMHTs in Greater Manchester boroughs of Bolton, Salford, Trafford and Manchester.

In line with the NHS Long Term Plan, GMMH has set out a Mental Health and Wellbeing Strategy to redesign and reorganise its community services with the help of AI with the aim of reducing waiting times by 13% and improving recovery rates by 24%. This approach also aligns with the GMMH Green Plan highlighting the importance of improving pathways through digital transformation: “Improving clinical decision-making in the selection of interventions will reduce lower value activities and hence their associated environmental impacts” (GMMH, 2022b).

The key objective of the research is to examine the efficacy and acceptability of MaST within GMMH’s CMHTs. The efficacy of MaST will be examined using a cost effectiveness analysis (CEA) and cost comparison analysis (CCA) to evaluate the costs and effects of alternative mental health interventions adopted in GMMH. We will also apply relevant statistics, machine learning and AI algorithms on aggregate MaST data, (e.g., Frequency Analysis, Correlation Analysis, Multivariate and Logistic Regression Analysis, Data stratification and Causal Inference), to evaluate the associations between Risk of Crisis and predictor variables. The acceptability of MaST will be examined using interviews with CMHT staff to determine how MaST fits the existing decision-making practices of CMHT staff and impacts the patient pathway. The qualitative research will also involve observations of supervisions by clinical supervisors over CMHT staff, and observations of the use of MaST by multidisciplinary teams.

Examining the efficacy and acceptability of MaST can produce the following potential benefits and impact to research participants:
1) Increase the explainability of how MaST works in a way that is understandable by non-technical users (i.e. CMHT staff)
2) Improve the transparency of how patient data is used and for what purpose in producing AI-enabled recommendations to CMHT staff
3) (1) and (2) will ensure the responsibility and trustworthiness of the AI components of MaST to CMHT staff.
4) Reduce the amount of time CMHT staff spend manually reviewing the electronic patient records by implementing improvements in MaST that helps to find information which is often buried deep in the records. This will help to increase the amount of time available for direct clinical care to improve operational efficiency.
5) Provide opportunities to share good practice of MaST use and digital adoption in MH services, which may be shared within the Trust and more widely
6) Show how AI driven dashboards can inform clinical decision making to improve job satisfaction and more objective (data driven) supervision, which may in turn improve the quality of care. So that this learning will be shared and improve service mental health delivery.

These benefits will be discussed and evaluated with CMHT staff over repeat interviews, but also with consultations with the patient representative group (PRG).

This research has received a funding of £278,176 for two years by the Faculty of the Humanities, University of Manchester. The funds will be used to recruit two postdoctoral research associates that will work together with the chief investigator and co-investigator on evaluating the acceptability and efficacy of MaST respectively.

University of Manchester Privacy Notice
The project follows the University of Manchester's central privacy notice, which sets out how personal information is being used and for which purposes. Information about the privacy notice can be found here https://documents.manchester.ac.uk/display.aspx?DocID=37095

References
•Baker, C., and Kirk-Wade, Esme. 2023. Mental health statistics: prevalence, services and funding in England. House of Commons Library. Number CBP-06988
•D’Alfonso, S., 2020. AI in mental health. Current Opinion in Psychology, 36, pp.112-117.
•GMMH. 2022a. Internal Report on Mental Health Services.
•GMMH. 2022b. Green Plan.
•Graham, S., Depp, C., Lee, E.E., Nebeker, C., Tu, X., Kim, H.C. and Jeste, D.V., 2019. Artificial intelligence for mental health and mental illnesses: an overview. Current psychiatry reports, 21, pp.1-18.
•Lau, N., O'Daffer, A., Colt, S., Joyce, P., Palermo, T.M., McCauley, E. and Rosenberg, A.R., 2020. Android and iPhone mobile apps for psychosocial wellness and stress management: systematic search in app stores and literature review. JMIR mHealth and uHealth, 8(5), p.e17798.
•Royal College of Psychiatrists. 2022. Hidden waits force more than three quarters of mental health patients to seek help from emergency services. Press release, 10 October 2022. www.rcpsych.ac.uk/news-and-features/latest-news/detail/2022/10/10/hidden-waits-force-more-than-three-quarters-of-mental-health-patients-to-seek-help-from-emergency-services