AI-supported early signal detection for prevention of chronic disease in Well@Home (research prototype)
- Project
- 22022 TREAT
- Type
- Enhancement
- Description
This result is a research prototype that adds predictive AI to the Well@Home remote patient monitoring platform. Its aim is to help care professionals, such as general practitioners, recognise patients at risk of developing a chronic condition earlier, so that prevention and follow-up can start before the condition is diagnosed.
An AI insights dashboard brings together three types of signals from a patient’s measurement history: patients with an elevated predicted risk of a future diagnosis, measurements that are projected to cross a clinical limit, and patients whose evolution differs from comparable patients. The care professional reviews each signal and decides on the next step, for example starting a preventive care path with remote monitoring.
Human oversight is part of the design: every prediction can be acknowledged or dismissed by the care professional, and this feedback is intended to improve the models over time.
In TREAT, the prototype was shown in the TREAT Complete demonstrator, where a GP identifies a persona at risk of type 2 diabetes and enrols her in a telemonitoring care path. The result is still in the research phase. The models are being evaluated and improved, and the prototype is not intended for clinical use yet.
- Contact
- Seppe Neyens
- seppe.neyens@bewellinnovations.com
- Research area(s)
- Artificial intelligence in healthcare; predictive analytics; clinical decision support; prevention of non-communicable diseases; remote patient monitoring; human-centred AI
- Technical features
- AI insights dashboard for care professionals
- Diagnosis prediction from a patient’s history over time
- Signals shown with an indicative confidence level. The confidence reflects how much the models agree; it is not a clinical risk score.
- Human-in-the-loop: the care professional acknowledges or dismisses each prediction. This supports transparency and alignment with the AI Act, and gives the basis for improving the models later.
- Integrated in Well@Home, so a flagged patient can be enrolled directly in a deterministic care path
- Integration constraints
The prototype is part of the Well@Home platform and requires a Well@Home environment. It is in the research phase and not available for clinical use. Clinical deployment will require further validation of model performance on representative data, clinical evaluation, and compliance with the EU AI Act, the Medical Device Regulation where applicable, and GDPR. The models need sufficient historical measurement data per patient to generate meaningful signals.
- Targeted customer(s)
General practitioners and primary-care networks, hospitals and outpatient clinics, home-care organisations, and prevention and chronic disease management programmes. In the current phase: clinical and research partners interested in co-validating the models.
- Conditions for reuse
The result is owned by BeWell Innovations and is in the research phase. Collaboration for further research and clinical validation is possible under a written agreement with BeWell Innovations. Future commercial use will be offered as part of the Well@Home platform under a commercial licence.
- Confidentiality
- Public
- Publication date
- 31-05-2027
- Involved partners
- BeWell Innovations (BEL)