AI Maps Brown Fat Genes for Next-Gen Weight Drugs

AI uncovers fat-burning genes that could inspire a new class of obesity drugs targeting metabolism, not appetite

4 Mar 2026

Artificial intelligence is helping researchers identify new biological targets for obesity drugs that could work by increasing the body’s energy use rather than suppressing appetite.

Biotechnology company insitro said it had used AI and computer vision tools to analyse MRI scans and genetic data from nearly 70,000 participants in the UK Biobank. The research focused on brown fat, a specialised tissue that burns calories to regulate body temperature.

Unlike white fat, which stores energy, brown fat consumes it. This has made the tissue an area of interest for scientists seeking metabolic therapies that raise energy expenditure.

Studying brown fat across large populations has historically been difficult because it typically requires specialised imaging methods. Insitro said its approach allows researchers to estimate brown fat levels from widely available MRI scans using AI driven analysis.

The company then linked those measurements to genetic data from the same individuals. The analysis identified several genes associated with brown fat activity.

One gene, known as BAT-01, emerged as a potential target. In diet induced obese mice, reducing BAT-01 activity led to about a 15 per cent decline in body weight over four weeks while preserving lean mass.

If confirmed in further studies, the approach could provide an alternative strategy to current obesity treatments. The leading drugs today, including GLP-1 therapies developed by Eli Lilly and Novo Nordisk, mainly reduce appetite by acting on signals in the brain.

A treatment that increases calorie burning in fat tissue could offer a complementary mechanism for weight loss.

According to insitro founder and chief executive Daphne Koller, artificial intelligence is enabling researchers to identify biological signals that were previously difficult to detect.

“AI driven measurements from routine MRI scans now allow researchers to study brown fat genetics at a scale that was not previously possible,” she said.

The findings were presented at the Keystone Symposia on Obesity Therapeutics. Researchers caution that the work remains early and any drug based on the targets would require years of testing before reaching patients.

The study nonetheless highlights how large biomedical datasets and machine learning are beginning to shape drug discovery, particularly in areas such as metabolic disease where new treatment approaches are being explored.