New AI Model Targets Surgical Kidney Injury Risks
The $4 million project will utilize federated learning to improve surgical outcomes for patients across 15 U.S. hospitals.
Updated on Oct. 27, 2026 in Artificial Intelligence
Live Poll
Do you believe using artificial intelligence to analyze patient health records will improve surgical outcomes?
The National Institutes of Health has awarded a $4 million grant to the Icahn School of Medicine at Mount Sinai to develop an AI model for predicting postoperative acute kidney injury. This research-stage project aims to improve diagnostic reliability across 15 hospitals.
Why it matters
Current prediction methods for kidney injury rely on small datasets and lack clinical reliability, potentially impacting a portion of the 50 million major noncardiac surgeries performed annually in the U.S. This initiative seeks to standardize risk assessment using large-scale, multi-institutional data.
The research team will analyze clinical data from 320,000 patients to improve upon existing methods that struggle with low reliability. The study uses federated learning to develop a shared prediction model while maintaining local data privacy across 15 participating hospitals.
The players
Icahn School of Medicine at Mount Sinai
A leading medical research institution focused on clinical innovation and data-driven healthcare.
National Institutes of Health
The primary federal agency responsible for biomedical and public health research funding.
Vanderbilt University
A research-intensive university participating in the multi-center study of postoperative outcomes.
University of California, Los Angeles
A major public research university contributing institutional data to the AI model development.
The details
The project uses federated learning, a decentralized machine learning technique that trains algorithms across multiple servers holding local data samples without exchanging the actual records. By training on a larger, more diverse patient cohort than current individual hospital models, the researchers aim to increase prediction accuracy for acute kidney injury. The model will be benchmarked against current approaches and alternative models built using standard pooled institutional data.
Timeline
October 27, 2026: The research grant award was formally announced.
The project spans a five-year period to complete the model development and analysis.
The Tech Race
This project aligns with the National Institutes of Health's broader efforts to advance predictive analytics in critical care. It follows a competitive trajectory to solve the data silo problems that have previously limited the effectiveness of hospital-based AI models.
While the technology is currently in the research-stage, findings are intended to inform future clinical trials for personalized prevention strategies. New York City patients at Mount Sinai may eventually benefit from these more precise risk assessments during surgical recovery.
The takeaway
The success of this study will be measured by whether the federated learning model outperforms traditional pooled data approaches in predictive accuracy. Researchers and clinicians should track the publication of results related to award 1R01DK144051-01A1 over the next five years.
Further reading
For broader trends in clinical diagnostics, visit the Artificial Intelligence section.
Source note: This article includes information reported by News-Medical.
Live Poll
Do you believe using artificial intelligence to analyze patient health records will improve surgical outcomes?
