Particle.news
Download on the App Store

AI Tool Links Insulin Resistance to Elevated Risk of 12 Cancers in UK Biobank Study

A University of Tokyo model infers insulin resistance from nine routine tests, enabling population-scale risk analysis.

Overview

  • The Nature Communications study from a University of Tokyo–led team reports population-scale evidence connecting insulin resistance to increased incidence across 12 cancer types.
  • Applying the AI-IR model to roughly 500,000 UK Biobank participants, researchers found a 25% higher overall cancer risk for those flagged as insulin resistant, with uterine cancer showing the strongest association at 134% increased risk.
  • AI-IR was trained and validated in independent U.S. and Taiwanese cohorts and showed strong agreement with directly measured insulin resistance.
  • The model outperformed BMI, metabolic syndrome criteria, and other standard markers, capturing risk in some normal-weight individuals whom BMI-based screening would miss.
  • Researchers say the approach could help identify high-risk people for targeted screening, and they are pursuing genetic and molecular follow-ups to probe mechanisms and guide implementation.