A recent hospital trial has shown promising results for artificial intelligence (AI) in predicting patient mortality. The AI system analyzed electrocardiogram (ECG) test results, a common heart function test, to identify patients at high risk of death.

The trial, conducted in Taiwan, involved nearly 16,000 patients across two hospitals. Researchers trained the AI on a massive dataset of over 450,000 ECG tests alongside patient survival data. This allowed the AI to learn and assign a percentile score representing each new patient’s risk of death. Patients scoring in the top 5% were flagged as high-risk.

The key finding of the trial was a significant reduction in deaths. When doctors received AI alerts for high-risk patients, it led to a 31% decrease in overall mortality among that group. Deaths specifically related to heart issues saw an even more dramatic drop – exceeding a 90% reduction for high-risk patients.

Experts hail this as a major breakthrough. “This is actually quite extraordinary,” said Eric Topol, a cardiologist not involved in the research. Traditionally, such significant reductions in mortality are rarely seen outside of new drug development. In this case, the AI acted as a non-invasive intervention, simply prompting doctors to prioritize high-risk patients.

While the exact reasons behind the improved outcomes are still under investigation, the success points to the potential of AI in flagging critical cases and prompting earlier intervention. This could revolutionize patient care by enabling more targeted use of medical resources and ultimately saving lives.