AI Can Now Read Body Composition to Estimate Big Health Risks
Researchers are using AI to map fat and muscle distribution from imaging data in order to predict major health risks. The work reinforces a larger trend in healthcare AI: extracting clinically relevant signals from scans that were not originally ordered for that purpose.
Body composition has long been clinically important, but it has often been assessed only indirectly. AI is now making it possible to quantify fat and muscle distribution more precisely and at scale, turning ordinary imaging into a richer source of metabolic and cardiovascular insight.
The appeal is obvious. If a model can identify risk patterns before overt disease emerges, clinicians may be able to intervene earlier with prevention, counseling, or follow-up testing. This is especially valuable in settings where patients already undergo imaging for unrelated reasons.
At the same time, these models sit on a difficult boundary between prediction and action. A result that meaningfully predicts risk is only useful if it leads to care pathways that are understood, reimbursable, and actually followed. Otherwise, the technology may generate better labels without changing outcomes.
The most interesting implication is that imaging AI is becoming phenotype AI. Instead of asking only what organ is affected, these tools ask what the scan reveals about the person’s overall health trajectory. That is a much bigger ambition — and potentially a much bigger market.