A deep learning model can analyze routine chest CTs to predict major adverse cardiac event risk, outperforming traditional methods.
Key Details
- 1Researchers developed a 3D convolutional neural network to assess MACE risk directly from chest CT images.
- 2The AI model leverages 'causal intervention' to improve generalizability across diverse patient populations.
- 3Performance metrics included AUCs of 0.73 (internal test set) and 0.69 (external test set), surpassing existing clinical models.
- 4Traditional risk estimations do not incorporate imaging data and underperform when applied externally.
Why It Matters
Incorporating imaging data with AI could enable more accurate, widely generalizable cardiac risk prediction, improving proactive patient management. Broad deployment of such models in routine chest CTs could significantly expand radiologists' role in population health.
Source
AuntMinnie
Related News
•HealthExec
US Executive Order and HHS Strategy Set AI Policy Directions for Healthcare
The White House executive order and new HHS strategy shift US policy towards unified AI standards and expanded adoption in healthcare.
•AuntMinnie
Study: Patients Prefer AI in Radiology as Assistive, Not Standalone Tool
Survey finds patients support AI-assisted radiology but not AI-only interpretations.
•AuntMinnie
AI-Enhanced MRI Boosts Return-to-Play Predictions for Athlete Muscle Injuries
Adding AI to MRI-based classification systems improves return-to-play predictions for professional athletes with muscle injuries.