How AI is computing the previously uncomputable in healthcare

At the University of Texas Medical Branch (UTMB), artificial intelligence is revolutionizing preventative care in healthcare by analyzing CT scans to assign cardiac risk scores, irrespective of the scan’s initial purpose. This innovation, which leverages AI to sift through existing patient data, has identified numerous high-risk cardiovascular cases that might otherwise have gone unnoticed. Peter McCaffrey, UTMB’s chief AI officer, highlights that this system doesn’t involve AI performing complex tasks; instead, it efficiently processes high volumes of data to uncover significant patient insights previously lost amid the vast amounts of data.

UTMB’s application of AI extends beyond cardiac screening. Their models, designed to scan for incidental coronary artery calcification (iCAC), help identify patients at risk of heart disease without visible symptoms. Consequently, the AI assigns risk tiers based on factors like statin use or cardiologist visits, further enhancing the utility of AI in patient triage. High-score patients receive notifications, ensuring timely intervention, while severe cases warrant direct follow-up calls—a process UTMB aims to automate shortly.

Moreover, AI assists in the quick detection of strokes and pulmonary embolisms, using algorithms trained to identify blockage symptoms in CT scans promptly. This timely identification accelerates critical treatment interventions, potentially saving lives by capitalizing on the small windows where prompt action is crucial. To maintain efficacy, these models undergo rigorous sensitivity and bias testing both before and after deployment. This ensures continuous enhancement in AI’s diagnostic accuracy and reduces instances of oversight due to bias.

UTMB also uses AI to automate routine checks such as validating inpatient admissions through AI-assisted reviews of EHRs, and refining documentation processes, thus alleviating pressure on medical staff. Across their applications, whether in identifying missed joint space narrowings or addressing the broader ‘intellectual bottleneck’, AI provides the computational power previously unavailable. McCaffrey underscores that AI isn’t in competition with healthcare professionals but functions as an essential partner, facilitating the scale and depth of analysis required to manage burgeoning data efficiently.