Mount Sinai researchers unveil InfEHR, an AI system that links unconnected medical events in electronic health records, enhancing diagnostic accuracy and uncovering hidden health patterns. This breakthrough promises significant advancements in personalized medicine and patient care.
A breakthrough in artificial intelligence could soon revolutionize how doctors diagnose diseases. Researchers at the Icahn School of Medicine at Mount Sinai and their collaborators have developed InfEHR, an AI system that connects disparate medical events over time. This innovative technology is capable of revealing hidden patterns within electronic health records (EHRs), transforming millions of fragmented data points into actionable diagnostic insights.
The study, published on Sept. 26 in Nature Communications, highlights InfEHR’s ability to personalize diagnostics. Rather than following a generic diagnostic process, InfEHR customizes its analysis for each patient by constructing a network from individual medical events.
This approach allows the AI to not only provide personalized answers but also pose personalized questions, significantly enhancing the diagnostic process.
“We were intrigued by how often the system rediscovered patterns that clinicians suspected but couldn’t act on because the evidence wasn’t fully established,” senior corresponding author Girish N. Nadkarni, the chair of the Windreich Department of Artificial Intelligence and Human Health, director of the Hasso Plattner Institute for Digital Health, the Irene and Dr. Arthur M. Fishberg Professor of Medicine at the Icahn School of Medicine at Mount Sinai, and the chief AI officer of the Mount Sinai Health System, said in a news release. “By quantifying those intuitions, InfEHR gives us a way to validate what was previously just a hunch and opens the door to entirely new discoveries.”
In the study, the InfEHR system analyzed deidentified electronic records from Mount Sinai Hospital in New York and UC Irvine in California. The AI transformed each patient’s medical timeline into a network illustrating how various medical events were connected over time. By scrutinizing multiple such networks, InfEHR learned to detect patterns typically associated with underlying conditions.
Cost efficiency and accuracy are notable advantages of InfEHR. The AI system doesn’t require extensive training data, instead learning directly from patient records. This adaptability allows it to work effectively across different hospital systems and populations.
For instance, InfEHR was significantly more effective in identifying newborns with sepsis, a life-threatening condition, compared to conventional methods. It was 12-16 times more likely to flag affected infants.
Similarly, the system identified patients at risk for postoperative kidney injury 4-7 times more effectively than current practices.
One of the standout features of InfEHR is its ability to indicate uncertainty. It can respond “not sure” when there is insufficient data, a key safety feature that addresses a major drawback of traditional AI, which might provide incorrect answers with unwarranted confidence.
“Traditional AI asks, ‘Does this patient resemble others with the disease?’ InfEHR takes a different approach: ‘Could this patient’s unique medical trajectory result from an underlying disease process?’ It’s the difference between simply matching patterns and uncovering causation,” added lead author Justin Kauffman, a senior data scientist in the Windreich Department of Artificial Intelligence and Human Health at the Icahn School of Medicine.
Looking ahead, the research team plans to extend InfEHR’s applications, such as personalizing treatment decisions by integrating clinical trial data. This could bridge gaps between trial-based research and patient care in diverse clinical settings.
“Clinical trials often focus on specific populations, while doctors care for every patient,” Kauffman added. “Our probabilistic approach helps bridge that gap, making it easier for clinicians to see which research findings truly apply to the patient in front of them.”