Problem Scope

Medication errors affect 7 million patients annually in the world and are recognised by the Ministry of Health (MOH) as the 5 th most common Serious Reportable Events in Singapore. The socio-economic impact of medication error is significant, including prolonged hospitalisation, higher risk for morbidity and mortality, and substantial healthcare spending (e.g., up to USD$40 billion in the United States and £750 million in England per year).

Strategies such as Computerised Provider Order Entry (CPOE) systems and pharmacists’ manual review of all prescriptions have achieved limited success in mitigating medication errors. In recent years, AI-related initiatives have emerged as potential solutions. However, few have been clinically adopted and all tackle only isolated domains of the problem.

Design Approach

We will develop a comprehensive, AI-assisted medication safety platform for clinical usage, comprising: (i) a medication safety datapool hosting extensive, labelled prescription and patient data, and (ii) novel, hybrid and explainable AI algorithms with continual learning capability for more accurate medication error detection, adverse drug reaction detection and prediction. As a start, we will focus on the diabetes mellitus and cardiovascular disease patients (in line with national strategic thrusts), before scaling to other patient populations.

Potential Impact/Benefits to Target Sector 

Success would contribute to reducing the medication errors and its associated risks including prolonged hospitalisation, morbidity and mortality, hence translating into healthcare cost savings for both institutions and patients. It would also contribute to the development of next-generation clinical decision support systems that promote greater confidence, safety and effectiveness for medication use in Singapore.