Sunday, Nov. 22, 2026 A Vik Media Group publication
Mongevity Longevity in Motion
Leer en español

Thermal camera tool for diabetes care tested to predict foot ulcer healing

☕ Includes a question for the family ↓

Tools for living with diabetesPart of a story we are following · 7 articles · See every update
Girl exercising side plank
Girl exercising side plankPhoto: Tyler Read / Wikimedia Commons, CC BY 2.0

Foot sores are one part of diabetes care that doctors watch closely, and they would love to know early which ones will take a long time to heal. A new camera tool is a first step toward that kind of early answer. It is not ready for your clinic yet.

In a study of adults with foot ulcers from two Melbourne clinics, a thermal camera tool trained on 44 healed and 24 unhealed ulcers correctly flagged about 76% of the ones that did not heal by 12 weeks. It was less sure about the ones that did heal. The researchers call the overall accuracy moderate.

How the study worked

Researchers in Australia recruited adults with neuropathic diabetic foot ulcers (sores linked to nerve damage) from two specialist foot clinics in Melbourne. At the first visit, they took thermal images of each ulcer, then checked at 12 weeks whether it had healed. The tool learned from those results, and then got a small first test on a separate set of ulcers. The study was published on November 8, 2026 in the J Diabetes Sci Technol. It was small, and the final test covered only a handful of cases, so it shows early feasibility, not proof that the tool works in everyday care.

What does the tool actually do?

The tool is called TherMUL. It is software that works with a thermal camera, the kind that shows heat as colors. It never touches the wound.

At the first visit, the camera takes a picture of the ulcer. The software then studies the texture of the heat pattern, a bit like reading the grain of a photo. An AI model compares that pattern with patterns from ulcers whose outcomes were already known. Then it guesses: likely to heal by 12 weeks, or not.

What they found

The tool was better at spotting ulcers that would not heal than ulcers that would.

  • For ulcers that did not heal, it got it right 75.8% of the time. That means roughly three out of four of those ulcers were flagged.
  • For ulcers that did heal, it got it right 53% of the time. That is only a little better than a coin flip, so some ulcers that went on to heal were wrongly flagged as unlikely to.

In the small first test on ulcers the tool had never seen, 5 of 7 predictions matched what the foot specialists found at 12 weeks. That is encouraging, but seven is a very small number to lean on.

The researchers describe the overall performance as moderate. They see it as a sign the idea is workable, and as something that could one day help clinicians spot ulcers unlikely to heal sooner and plan care earlier.

Why does an early prediction matter?

Right now, doctors usually find out that a diabetic foot ulcer is stuck by waiting. The current standard looks at whether the wound has shrunk by at least half over 4 weeks. If it has not, it is treated as a chronic ulcer.

That means a month can pass before the picture is clear. A tool that gives a hint at the very first visit could, in theory, let a care team adjust the plan sooner. This study did not test whether that actually happens. It only tested whether the camera could make a useful guess.

What you can do today

Nothing about current advice changes, and this tool is not something you can ask for at your clinic. What this study does point to is how healing is judged today: by whether the wound shrinks by about half in 4 weeks. If you or someone you care for has a foot ulcer, ask the foot care team how the wound is being measured at each visit and whether it is on track to shrink by about half in that first month.

For clinicians and researchers

Design: Development and preliminary validation study of TherMUL, an AI thermal-imaging device, in adults with neuropathic diabetes-related foot ulcers from 2 specialist foot clinics in Melbourne, Australia. An SVM classifier was trained on images from 44 healed and 24 unhealed DFUs; 12-week healing status was the label.

Main result: The classifier achieved a mean AUC of 0.635, with sensitivity of 75.8% for unhealed DFUs and specificity of 53% for healed DFUs. In the preliminary evaluation, 5 of 7 predictions matched 12-week podiatry outcomes.

Main limitation: Small training set (44 healed, 24 unhealed ulcers) and a very small unseen evaluation set (7 predictions); discriminative performance was moderate, and the study was limited to 2 clinics in one city.

How it fits: Complements earlier coverage of a Danish trial of an add-on that healed more diabetic foot ulcers, by addressing early identification of ulcers unlikely to heal rather than treatment. Quitting smoking pays off in diabetes care: fewer eye, kidney and nerve problems · A simple add-on to diabetes care healed more foot ulcers in a Danish trial

Open question: Whether performance holds in larger, independent, multicentre cohorts, and whether early prediction with TherMUL changes wound management and healing outcomes, remain to be tested.

Talk about it at home

Researchers used a thermal camera, the kind that turns heat into colors, a bit like a weather map for warm and cool spots. They took pictures of foot sores, waited to see which ones healed, and taught a computer program to notice patterns in the colors. Then they tried it on pictures it had never seen, the way a teacher gives a surprise quiz after practice questions.

Question for the table

What if you could walk through your home with a camera that paints heat in colors? Where would you expect the warmest colors, and where the coolest? What would a glass of cold water look like next to your hand?

Keep exploring

Copies this article and its questions so you can paste them into ChatGPT, Claude, Gemini or the AI you use, and keep the conversation going.

Got a question? Ask us ›

Get the best of longevity science by emailThe most useful new studies, in plain language and with the source. Free.

We'll email you once to confirm. Occasional messages from our sponsors, clearly labeled. Unsubscribe anytime. We never ask for health data.
Cite this article
AMA
Mongevity Desk. Thermal camera tool for diabetes care tested to predict foot ulcer healing. Mongevity. Published November 22, 2026. Accessed November 22, 2026. https://mongevity.science/metabolism/2026/11/thermal-camera-tool-predict-diabetic-foot-ulcer-healing/
Vancouver
Mongevity Desk. Thermal camera tool for diabetes care tested to predict foot ulcer healing [Internet]. Mongevity; 2026 Nov 22 cited 2026 Nov 22]. Available from: https://mongevity.science/metabolism/2026/11/thermal-camera-tool-predict-diabetic-foot-ulcer-healing/
APA
Mongevity Desk. (2026, November 22). Thermal camera tool for diabetes care tested to predict foot ulcer healing. Mongevity. https://mongevity.science/metabolism/2026/11/thermal-camera-tool-predict-diabetic-foot-ulcer-healing/
For your website
<a href="https://mongevity.science/metabolism/2026/11/thermal-camera-tool-predict-diabetic-foot-ulcer-healing/">Thermal camera tool for diabetes care tested to predict foot ulcer healing</a>

Free to share or quote with credit and a link.

Questions and answers

What is TherMUL?
TherMUL is software that works with a thermal camera. It studies the texture of the heat pattern on a diabetic foot ulcer at the first visit and predicts whether it will heal by 12 weeks.
Can I get this camera test at my clinic?
Not yet. The study shows early feasibility only, and it was tested at two clinics in Melbourne, Australia.
How accurate was it?
The researchers call it moderate. It flagged 75.8% of the ulcers that did not heal, but correctly identified only 53% of the ones that did heal.
How are foot ulcers judged today?
The current standard looks at whether the wound area shrinks by at least half over 4 weeks. If it does not, the ulcer is treated as chronic.

Sources

  1. J Diabetes Sci Technol: Automated Prediction of Diabetes-Related Foot Ulcer Healing at Initial Assessment (PMID 43000711 · DOI 10.1177/19322968261501979) · 2026-11-22

More on this story

Report an error