Overview
Researchers at the Icahn School of Medicine at Mount Sinai have developed an AI system named InfEHR that improves diagnostic capabilities by analyzing electronic health records (EHRs). This innovative system connects disparate medical events over time, revealing hidden patterns that can assist healthcare professionals in making informed decisions.
Key Findings
- InfEHR creates a diagnostic web that links unconnected medical events, providing actionable insights.
- The system tailors its analysis to individual patients, enhancing the personalization of diagnostics.
- InfEHR was tested using deidentified EHRs from two hospital systems, demonstrating its ability to identify conditions that are often overlooked.
System Functionality
Unlike traditional AI models that apply a uniform diagnostic process, InfEHR adapts its analysis based on a patient’s unique medical history. This allows the system to:
- Build a network of a patient’s medical events and their interconnections.
- Provide personalized diagnostic insights and questions.
Performance Metrics
In a study, InfEHR was able to:
- Identify newborns at risk for sepsis 12-16 times more effectively than current methods.
- Flag patients at risk for postoperative kidney injury 4-7 times more efficiently.
Additionally, the system can indicate when it lacks sufficient information, enhancing safety in clinical settings.
Future Directions
The research team plans to make InfEHR’s coding available to other researchers and explore its potential in personalizing treatment decisions based on clinical trial data. This could help bridge the gap between clinical research and real-world patient care.
Conclusion
InfEHR represents a significant advancement in the use of AI for diagnostics, particularly in identifying rare diseases and improving patient outcomes. By leveraging vast amounts of EHR data, this system has the potential to revolutionize how healthcare providers approach diagnosis and treatment.