Body fat is often associated with weight gain and metabolic problems. But not all fat behaves the same way. Brown adipose tissue, commonly called brown fat, burns energy to produce heat rather than simply storing it. Now, an artificial intelligence study involving more than 101,000 patients has linked detectable brown fat to lower rates of several conditions, including heart disease, type 2 diabetes, depression and anxiety. Published in Nature on October 14, 2026, the study used medical records and imaging data to investigate how brown fat relates to health across multiple organ systems. The findings broaden scientists’ understanding of this specialised tissue, although they do not establish that brown fat directly prevents disease.
What makes brown fat different?
Unlike white adipose tissue, which primarily stores excess energy, brown fat contains abundant mitochondria, the structures that generate energy inside cells. These mitochondria contain a protein called uncoupling protein 1 (UCP1), which helps convert stored chemical energy into heat. Brown fat plays a role in maintaining body temperature, particularly during cold exposure. Scientists have also found that adults retain some active brown fat, often around the neck, collarbone and armpits. Its ability to consume glucose and other fuels has made it an important target for metabolic research. However, studying brown fat across large populations has been difficult because identifying and measuring its activity requires specialised imaging.
Researchers analysed clinical records and PET/CT scans from 101,243 patients at a major cancer centre. These scans use a radioactive glucose tracer to highlight metabolically active tissues, including brown fat. Using machine learning, the team developed a method to estimate brown fat activity in approximately 8,700 patients. They then compared health records to identify conditions associated with the presence of brown fat. The differences were notable. Patients with detectable brown fat had approximately 58% lower odds of type 2 diabetes and 47% lower odds of chronic ischaemic heart disease than those without detectable brown fat. They also had lower odds of depression and anxiety disorders. These figures describe differences in odds within the study population, not reductions in an individual’s future risk. They also found that people with brown fat tended to have more favourable blood glucose measurements, lipid profiles and inflammatory markers. Higher activity in certain brown fat depots, particularly around the armpits, was associated with healthier metabolic indicators and a more favourable distribution of white fat.
Does brown fat protect the heart and brain? That remains an open question. The study was retrospective, meaning researchers examined existing medical records rather than assigning participants to an intervention and tracking future outcomes. Brown fat could contribute to better health, but other factors may explain some of the associations. The participants were also patients receiving cancer-related imaging, so the results may not apply equally to the general population. In addition, glucose uptake on a scan is an indirect measure of brown fat activity, not a direct measurement of heat production. The team also investigated medications that might stimulate brown fat, identifying losartan, a commonly prescribed blood-pressure drug, for further study. Experiments in cells and mice showed increased thermogenic activity, but this has not established a brown-fat-related benefit in humans. The larger message is that body fat is biologically diverse. By combining AI with large medical datasets, researchers can identify patterns that might otherwise remain hidden—and generate new questions about how metabolism, cardiovascular health and mental well-being are connected.
Sources:
Gómez-Banoy, N., Barrero, J. A., Pinheiro de Carvalho, L. R., [remaining authors]. (2026). Artificial intelligence-enabled clinical phenotyping of human brown fat. Nature. https://doi.org/10.1038/s41586-026-11162-7
