Effective integration of health and social care data for improved decision-making

Introduction/background summary: Frontline health and social care professionals and managers co-designed a machine learning and interactive visualisation solution to better understand their datasets of persons affected by cancer. Context: While cancer is primarily a health condition, it is often a life-changing diagnosis that impacts almost every other aspect of a patient’s life, and the lives of those close to them. This demands a holistic and integrated approach towards the care of persons affected by cancer (PABC). However, the data relevant to such care often exists in “silos”, with the separation of medical records from financial information about the PABC being one common example. In cases where different datasets are accessible, the data can remain disjointed, complicating efforts to extract insight from the combined data. This paper presents a possible digital solution to these challenges, applying machine learning and interactive visualisation technologies to enable frontline workers and decision-makers to explore large datasets. Project Partners: The system was developed by Abertay University and evaluated with electronic holistic needs assessment (HNA) data from Macmillan Cancer Support on around 100,000 PABC across the United Kingdom. The Digital Health and Care Innovation Centre facilitated a series of design-driven user engagement workshops with health and social care professionals to provide regular feedback in an iterative, co-design process. Method: Association rule mining is used to identify association “rules” in the data that specify the strength of a connection between one or more “antecedents” and one or more “consequents” [1]. As an unsupervised machine learning technique, this has the cost-saving benefit of not requiring labour-intensive labelled training datasets. The connections within the data are then visualised with a three-dimensional interactive graph powered by the Unity game engine. For accessibility and ease of deployment, the front-end system is hosted as a web app in the Microsoft Azure cloud. Results and Discussion: The tool demonstrated value in areas such as the identification of pressure points in the care system to facilitate optimal reallocation of resources. More importantly, the HNA process has been shown to provide a clinically meaningful and statistically significant improvement in the quality of life of PABC, as measured by EuroQol’s EQ-5D scores [2]. It is hoped that the continued development of artificial intelligence (AI) data integration and visualisation systems such as those reported in this paper will eventually yield a corresponding population-level improvement in quality of life. 1. Agrawal R, Imielinski T and Swami A. Mining Association Rules Between Sets of Items in Large Databases. In: Proceedings of the 1993 ACM SIGMOD International Conference on Management of Data, Washington, D.C., USA, pp.207-216. New York, NY, USA: ACM. 2. Snowden A, Young J and Savinc J. Meeting psychosocial needs to improve health: a prospective cohort study. BMC Cancer 2020. DOI: 10.1186/s12885-020-07022-w.