Mass General Brigham researchers developed AI models that use EMR, including radiology data, to predict risk for intimate partner violence (IPV) years before patients seek care.
Key Details
- 1AI pipeline trained on 673 IPV cases and 4,169 controls using EMR data from 2017-2022.
- 2Models included structured data, unstructured clinical notes, and radiology/emergency department reports.
- 3The fusion 'Holistic AI in Medicine (HAIM)' model achieved the highest accuracy (88%).
- 4Fusion model could predict 80.5% of IPV cases on average 3.7 years before patients sought care.
- 5Validated on two external datasets with similarly high accuracy.
- 6Research led by Mass General Brigham radiologist; funded by NIBIB and NIH.
Why It Matters
Early detection of IPV could allow healthcare providers to intervene and prevent serious harm. Demonstrates the unique value of including radiology and unstructured EMR data in risk prediction models for vulnerable patient groups.
Source
EurekAlert
Related News
•EurekAlert
AI-Powered OCT Enables Rapid 'Optical Biopsy' for Early Endometrial Cancer Detection
A team at Washington University has developed a catheter-based 3D OCT system with AI to quickly and noninvasively detect early endometrial cancers.
•EurekAlert
AI Clinical Reasoning in Diagnostics and Digital Fatigue in Healthcare
Recent JMIR features explore large language models in clinical diagnostics and digital fatigue among healthcare professionals.
•EurekAlert
KAIST, MIT, Microsoft Develop Efficient AI Image Upsampling for Robotics
KAIST, MIT, and Microsoft have created 'Upsample Anything,' a training-free AI method to restore high-resolution visual data from compressed images with up to 16x improved GPU memory efficiency.