Artificial Intelligence

AI Tool Predicts Disease Risk from Rare Genetic Mutations

Published AUG 29 2025

Story

Above: Three red diagonal DNA chains against dark blue background. Image credit: Unsplash

The Facts

Researchers at the Icahn School of Medicine at Mount Sinai have developed an artificial intelligence framework that uses machine learning to predict disease risk from rare genetic variants, moving beyond traditional binary classifications. The study was published in the journal Science on Thursday.
HIT Consultant
Lab Manager
The AI model analyzes over 1 million electronic health records and routine laboratory data, including cholesterol levels, blood counts, and kidney function tests, to generate ML penetrance scores ranging from 0 to 1. Higher scores indicate a greater likelihood of disease development associated with specific genetic variants.
Technology Networks
www.science.org
Traditional genetic studies classify patients using simple yes/no diagnoses, but this approach inadequately captures diseases like hypertension, diabetes, and cancer that exist on a spectrum of severity. The Mount Sinai team trained AI models to quantify disease expression continuously rather than categorically.
Medical Express
Mount Sinai
The research team calculated ML penetrance scores for more than 1,600 rare genetic variants across 10 common diseases, including familial breast cancer, Lynch syndrome, and polycystic kidney disease. Some variants previously labeled as uncertain showed clear disease signals in the analysis.
EuroNews
UPI
Lead study author Dr. Iain S. Forrest emphasized that the AI model serves only as a clinical decision support tool, particularly when genetic test results are unclear or ambiguous. The tool could help determine appropriate screening schedules and prevent overtreatment.
Lab Manager
Science
The research team reportedly plans to expand the model to include more diseases, broader genetic changes, and more diverse patient populations while conducting longitudinal studies to validate predictions over time. The work was supported by multiple National Institutes of Health grants.
UPI
HIT Consultant

Sources Split

LEFT

CENTER

RIGHT

ESTAB
CRIT

CENTER

PRO
ESTAB

The Spin

Techno-optimist narrative

This breakthrough marks a significant advancement in precision medicine, offering patients and doctors the clear answers they've long been seeking. For too long, genetic testing has left families in limbo with vague "uncertain significance" results that provide no actionable guidance. This AI tool transforms that uncertainty into concrete, data-driven risk scores that can guide real medical decisions — whether that means starting cancer screening early or avoiding unnecessary anxiety about low-risk variants.

Lab Manager

Techno-skeptic narrative

While this technology shows promise, we must proceed with extreme caution before integrating AI-driven risk scores into clinical practice. The model relies heavily on electronic health records that may contain biases and incomplete data, potentially leading to inaccurate risk assessments that could harm patients. Without extensive validation across diverse populations and long-term outcome studies, these AI predictions could create false confidence in medical decision-making where human judgment and established clinical guidelines should prevail.

UPI