Exploring the Barriers and Opportunities for a More Predictive Data-Driven Telecare Service: Qualitative Study in Scotland.

Background: Telecare uses technology to help people live more independently at home. When an adverse event (such as someone falling or a bath overflowing) happens, the technology reactively senses this and alerts a call center to respond. If the technology can detect a person's current (and past) states and behaviors, with machine learning, we can more proactively identify potential risks before an adverse event occurs and intervene. Despite social care organizations being data-rich, few predictive analytics are currently routinely applied. There is a need to understand current data management practices before optimizing organizational and technical readiness for proactive data-driven telecare services. Objective: The aim of this study was to understand how specific telecare data (monitoring falls, identifying people at risk of falling, and providing services in response) are collected, managed, and used in the largest health and social care region in Scotland. The objectives were to: (1) map the community alarm data flow to understand what data were being collected, by what services, and where it was stored, linked, and managed; and (2) identify the current barriers and opportunities around staff and organizations using or applying predictive analytics routinely within telecare service provision. Methods: This qualitative study involved interviews with health and social care professionals working in Glasgow City Council (GCC) and a telecare service provider (Tunstall). Interviews explored experiences of the systems and data access, processes for collecting and using the data, and how it might be better used to target services more proactively. Data underwent a thematic framework analysis. Descriptions of the data flow were used to develop visual representations of the sociotechnical system. Results: A total of 14 participants at operational and managerial levels took part. A complex sociotechnical telecare system was identified, involving multiple staff roles, with data exchanged across 11 teams, using 17 systems, with 4 distinct data sources. In total, four themes highlighted key challenges that are: (1) suboptimal systems and equipment; (2) data recording inefficiencies and use; (3) specific patient population barriers and IT literacy; and (4) limited resources and support. Opportunities for more predictive telecare included: establishing a more structured and integrated approach to data management; scope for improved data organization and retrieval; better cross-platform integration and data sharing; and the use of tools or models to support insightful data analysis tailored to the users. Conclusions: Scottish telecare data services require improved infrastructure to be managed in ways that support more predictive telecare services. This includes more structured and linked datasets and greater integration between the services and systems to allow service providers more integrated, up-to-date, and real-time connected data to build accurate and meaningful models.