AI-IR model provides first population-scale evidence for cancer risk link
University of Tokyo researchers have demonstrated that insulin resistance is a risk factor for 12 types of cancer. Publishing their findings in Nature Communications on February 15, the team applied a machine learning tool named AI-IR to data from approximately 500,000 participants in the UK Biobank. The study offers large-scale evidence linking the metabolic condition to cancer risk and introduces a screening method utilizing data from routine health checkups.
The AI-IR model utilizes nine standard clinical parameters typically collected during medical examinations to predict insulin resistance, a condition where the body’s cells respond poorly to insulin. This approach addresses a significant logistical barrier in medical research, as direct measurement of insulin resistance generally requires specialized testing available only in advanced diabetes clinics. By leveraging accessible data, the tool makes population-level screening feasible without the need for advanced clinical resources.
According to Yuta Hiraike, a researcher at the University of Tokyo Hospital who led the study, the method fills a gap in epidemiological evidence. “While a possible link between insulin resistance and cancer has been suggested, large-scale evidence has been limited due to the difficulty of evaluating insulin resistance in the clinic,” Hiraike stated. He emphasized the study’s contribution to the field, noting, “But with AI-IR, we have provided the first population-scale evidence that insulin resistance is a risk factor for cancer.”
The model demonstrated superior performance compared to established diagnostic markers. In assessments, AI-IR outperformed body mass index (BMI), metabolic syndrome criteria, and other standard markers in predicting diabetes incidence. This predictive capability provided the necessary foundation for the researchers to examine associations with cancer. The limitations of using BMI as a proxy for insulin resistance are well-documented; the metric produces both false positives and false negatives. This inconsistency is evident in individuals with obesity who remain metabolically healthy and in individuals with normal BMI who develop insulin resistance.
Hiraike explained the technical advantage of the new metric. “By combining nine clinical parameters into a single metric, AI-IR can detect insulin resistance that BMI alone cannot explain,” he said. The model’s accuracy was confirmed through rigorous testing: “When compared with directly measured insulin resistance in validation datasets, AI-IR achieved strong predictive performance.” The researchers validated the tool using independent cohorts from the United States and Taiwan prior to its application to the UK Biobank data.
Analysis of the UK Biobank data revealed a distinct correlation between the AI-IR classification and cancer risk. Participants who were classified as AI-IR positive but did not have diabetes still faced an elevated risk of cancer compared to those who tested negative. The team is currently investigating how genetic differences influence insulin-resistance-related cancer risk. Their objective is to connect large-scale human data with molecular biology to develop strategies for overcoming insulin resistance.