Incorporating AI into cardiovascular diseases prevention - insights from Singapore

Introduction Improved upstream primary prevention of cardiovascular disease (CVD) would enable more individuals to lead lives free of CVD. However, there remain gaps in provision of CVD primary prevention, namely a) addressing social determinants of health in an equitable way and b) enabling convenient and accessible CVD primary prevention across the spectrum of primary to specialist care. Methods A battle map for CVD prevention- CardioSight. Using the data informatics capabilities at the National University Health System (NUHS), Singapore, empowered by the Endeavour AI system, and combined large language model (LLM) tools, our team has created a real-time dashboard able to capture and showcase information on cardiovascular risk factors at both individual and geographical level- CardioSight. Being Streambased AI, the dashboard can interface in real time with our healthcare cluster’s Electronic Medical Records System (EMR) to achieve live data feeds for real time analysis and visualisation (Figure 1). Real time data is pinned on a geographical heat map alongside nearest primary care providers, providing a battle map to target individuals at various levels of cardiovascular risk. Further insights such as medication records and data on area-level socioeconomic determinants allow a whole-of-systems approach to promote healthcare delivery, while also allowing for outcomes to be tracked effectively. Results Figure 2 shows the Glycated hemoglobin (HBA1c) and low-density lipoprotein- C (LDL-C) values of 93,200 patients with diabetes managed in NUHS from 1/6/23 to 31/10/23, mapped to where they live in Singapore. Light green spots represent HBA1c < 7, yellow spots represent HBA1c 7-10%, and red spots represent HBA1c > 10%. Diamonds represent LDL-C 1.8 to <= 2.6, up-pointing triangles represent LDL-C 2.6 to <= 3, and down-pointing triangles represent LDL-C > 3. 13,728 (14.7%) patients have concurrently HBA1c > 7% and LDL-C > 1.8. Discussion A significant proportion of the population of diabetics has risk factors not at target. There is also a geographical overrepresentation of some of these patients in certain areas. These would represent key groups and individuals to deliver preventive interventions. Conclusion In its complete form, CardioSight supports the execution of personalised yet high-level preventive cardiology initiatives within our healthcare system. Importantly, knowledge of baseline epidemiology of CVD and risk factors will allow assessment of the effectiveness of ongoing interventions over not just a geographical location but also across a timeline. Fig 1.NUHS AI Tribrid Architecture Fig 2.