INTRODUCTION:
South Africa’s public health system, serving over 80% of the population, faces significant challenges due to inadequate healthcare systems with a shortage of healthcare professionals. Unplanned readmissions in South Africa, where 10.5% of patients return to the hospital within 30 days after being discharged (Dreyer et al., 2019) is a major concern, with discharge processes often having poor continuity of care post discharge. The aim of the study was to develop an AI intervention to identify and address unplanned readmission at a selected hospital in KwaZulu-Natal

METHODS:
A transdisciplinary team (university and industry collaboration) was established to develop 1). A readmission risk classification AI tool utilizing predictive analytics to flag patients at risk of complications and readmissions post-discharge; and 2) An evidence-based review of interventions to prevent unplanned readmissions in low resource settings.

RESULTS:
Machine learning using Random Forrest was used to develop an algorithm to flag patients at risk of an unplanned readmission. The review identified four key interventions: 1). Discharge planning; 2) Medication Review; 3) Patient Education; and 4) Follow-up visits with recommendations to leverage AI-powered clinical decision support assistance to facilitate post-discharge recovery and prevent unplanned readmission which will enhance patient care continuity and mitigate workforce shortage.

CONCLUSIONS:
The presentation will present work to date and will discuss the challenges in developing an AI-powered clinical decision support system integrated with our readmission classifier. This project is creative in that it firstly addresses the critical gaps in care continuity of care after hospital discharge in public hospitals in low income settings in South Africa.