Zero-Shot Medical Event Prediction Using Generative Pretrained Transformer on Electronic Health Records

Strategic Insights -

Researchers from the University of California have made a breakthrough in medical prediction using generative pretrained transformers (GPT) on electronic health records (EHRs). Their study, published in the Journal of the American Medical Informatics Association, demonstrated that GPT-based foundational models can predict future medical events without requiring extensive labeled data or task-specific training. The model's performance was evaluated across multiple time horizons and clinical categories, achieving high true positive rates while maintaining low false positives.

Key Takeaways:

  • The study presented the first comprehensive analysis of zero-shot forecasting with GPT-based foundational models in EHRs, introducing a novel pipeline that formulates medical concept prediction as a generative modeling task.
  • The GPT-based model was able to capture latent temporal dependencies and complex patient trajectories without task supervision, achieving an average top-1 precision of 0.614 and recall of 0.524.
  • The model demonstrated strong zero-shot performance in predicting the next medical concept for 12 major diagnostic conditions, achieving high true positive rates while maintaining low false positives.
  • The study highlighted the versatility of GPT-based models in capturing diverse phenotypes and enabling robust, zero-shot forecasting of clinical outcomes, making them more scalable applications in clinical settings.
  • The research has the potential to enhance the versatility of predictive healthcare models and reduce the need for task-specific training, enabling more efficient and effective medical predictions.

Statistics:

  • Average top-1 precision of 0.614 and recall of 0.524 achieved by the GPT-based model.
  • The model demonstrated strong zero-shot performance in predicting the next medical concept for 12 major diagnostic conditions.
  • High true positive rates (up to 90%) and low false positives (less than 5%) achieved by the model in evaluating performance across multiple time horizons and clinical categories.

Sources:

NewsRx. Findings from University of California Has Provided New Information about Electronic Medical Records (Zero-shot medical event prediction using a generative pretrained transformer on electronic health records). Information Technology Newsweekly. October 21, 2025; p 236.

Journal of the American Medical Informatics Association. Zero-shot medical event prediction using a generative pretrained transformer on electronic health records.