AI Tool Reveals Insulin Resistance as Hidden Risk Factor for Twelve Cancers

A New Way to Spot a Hidden Danger

A machine learning tool developed by researchers at the University of Tokyo has uncovered compelling evidence that insulin resistance, a condition where the body's cells stop responding properly to insulin, is a significant risk factor for twelve different types of cancer.

The study, published in Nature Communications on the fifteenth of February twenty twenty-six, analysed data from approximately half a million participants enrolled in the UK Biobank, making it the first population-scale investigation of the link between insulin resistance and cancer.

How AI-IR Works

The tool, called AI-IR, uses nine standard clinical parameters collected during routine health checkups to predict insulin resistance. Directly measuring insulin resistance typically requires specialised testing only available in advanced diabetes clinics, making large-scale screening impractical. AI-IR sidesteps this limitation entirely by relying on data that doctors already collect.

Critically, AI-IR outperformed body mass index, metabolic syndrome criteria, and other established markers in predicting diabetes incidence. BMI has long been used as a rough proxy for metabolic health, but it produces both false positives and false negatives. Some people with obesity remain metabolically healthy, while others with a normal BMI quietly develop insulin resistance.

Twelve Cancers Identified

The research found that AI-IR is significantly associated with increased risk of six cancers: uterine, kidney, oesophagus, pancreas, colon, and breast. It also showed notable associations with six additional cancers: renal pelvis, small intestine, stomach, liver and gallbladder, leukaemia, and bronchial and lung cancers. Participants classified as insulin resistant who did not have diabetes still faced elevated cancer risk compared to those who tested negative.

What Comes Next

Because the tool uses only routine health data, it could enable widespread screening for diabetes, cardiovascular disease, and cancer without requiring specialist clinic visits. The research team validated AI-IR using independent cohorts from the United States and Taiwan before applying it to the UK Biobank population, demonstrating its robustness across diverse groups. The team is now investigating how genetic differences influence insulin-resistance-related cancer risk.