A University of Tokyo team deployed a machine‑learning predictor called AI‑IR to estimate insulin resistance from nine routine clinical measurements across 500,000 UK Biobank participants and found insulin resistance associated with increased incidence in 12 cancer types. The research, published in Nature Communications, provides population‑scale evidence linking metabolic dysfunction to broad cancer risk. AI‑IR infers insulin resistance where direct measures are impractical, enabling large‑scale epidemiologic analysis. Authors report that AI‑predicted insulin resistance remained a significant risk factor after adjusting for conventional confounders, suggesting metabolic interventions might affect cancer incidence patterns. Why it matters: the findings could reshape screening and prevention strategies by integrating metabolic risk profiling into oncology risk models and by prioritizing metabolic-targeted trials for cancer prevention.