Foot check & AI: better photos and support for triage

With the Diabetes Foot Check app, people with diabetes can check their feet at home and share photos with their podiatrist. Practical experience and a previous evaluation study have shown that these photos are not always easy to assess: for example, they may be out of focus, too dark, taken from too far away or incomplete. This makes remote assessment difficult, can lead to additional contact appointments and may reduce confidence in digital monitoring. As part of the app’s ongoing development, we therefore investigated how artificial intelligence (AI) can help users take better photos and support the podiatrist’s assessment.

Together with Fontys, we investigated which characteristics make a foot photograph suitable for assessment by a podiatrist. We looked at sharpness, exposure, distance, framing and the visibility of the relevant area of the foot. We then explored how image recognition during photography can flag that a photograph is likely to be of insufficient quality for assessment. We investigated how this can be translated into clear instructions, such as moving closer, using more light or keeping the camera steady. In addition, we explored how AI can support triage in the longer term: by recognising potential risk indicators and highlighting cases that require priority assessment. Within these lines of research, we developed insights and prototypes, based on the principle that the final assessment and decision regarding follow-up care remain with the healthcare professional.

In the short term, the greatest practical value of AI lies in guiding patients to take a usable foot photograph. Immediate and clear feedback can help to rectify problems whilst taking the photograph, rather than having to retake it afterwards. In this way, AI offers opportunities to improve ease of use and provide the podiatrist with images that are easier to assess. Possibilities have also been explored for recognising risk indicators and supporting triage, but the prototypes developed do not yet constitute a final product for use in healthcare. For this further support, larger and representative datasets, reliable clinical assessments of the images and independent validation are necessary. AI is therefore a tool for patients and healthcare professionals, not a replacement for the podiatrist or a system that makes diagnoses independently.

The insights and prototypes from this project form the basis for follow-up research and the safe integration of AI support into the future Rondom app. The initial focus is on supporting patients in taking high-quality photographs of their feet. Applications for triage and risk assessment require further development and validation before they can be used responsibly in day-to-day care. This also requires clear agreements on privacy, responsibilities and the requirements applicable to medical devices. In this way, we are working towards digital foot care in which patients can carry out their foot checks more easily and podiatrists receive more useful information, whilst professional assessment remains central.

Artificial intelligence, diabetes, image quality, image recognition, triage, digital foot care

Partner organisations: Rondom Podotherapeuten and Fontys.
Funding: Centre of Expertise Health.

Diabetes Foot Check website