AI-based wearable for early detection and prevention of deep vein thrombosis

The VDE is participating in the EU project ThrombUS+, which aims to improve risk assessment and prevention of deep vein thrombosis with a portable vein analyzer and an AI-based diagnostic device. In an interview with MEDICA-tradefair.com, Dr. Thorsten Prinz describes the three-and-a-half-year project.

Dr. Prinz, how should we envision this device?

Dr. Thorsten Prinz: Currently, deep vein thrombosis is diagnosed primarily using ultrasound. With simultaneous compression, we observe whether the vessels can be compressed by external pressure. If the ultrasound shows that they cannot be fully compressed, this is an indication of deep vein thrombosis.

The wearable developed in this project is based on a stocking for the entire leg or a pair of trousers. It will integrate sensors for ultrasound as well as for performing electrical impedance plethysmography and light reflection rheography (LRR) in order to ensure continuous and highly sensitive monitoring.

The intention is to evaluate the sensor data using AI-based software. The healthcare professional can then view the results on a monitor, tablet or smartphone, for example.

For which type of applications is this appropriate?

Prinz: The main application is likely to be in the clinical environment post-operatively, for example in the recovery room. There, monitoring can be carried out by clinical staff. If a thrombosis forms, an alarm is triggered.

In addition to monitoring, a second aspect is prevention using a network of sensors for physiological signals and activity measurements to assess the activity patterns of the lower limb. Shortly after an operation, the exercise program usually begins with physiotherapy and appropriate medication to prevent thrombosis.

The activity sensors check whether patients are moving sufficiently and whether they are moving correctly. The wearable could potentially also be used for this in the home care environment.

What is the timetable for the project?

Prinz: The ThrombUS+ project, which was launched in January, consists of nine work sections that are being worked on by 18 interdisciplinary and international project partners in eight countries. It is being funded for a period of three and a half years with 9.5 million euros in the EU's Horizon Europe program. This not only includes technical development but also includes clinical studies with which we want to generate data for training the AI models and ensure their safety and performance during use in patients.

The VDE is also providing regulatory support for the project from the outset. This medical device in development is not only subject to high regulatory hurdles from the European Medical Device Regulation (MDR), but also the recently adopted Artificial Intelligence Act.

What are the development challenges?

Prinz: There are technical hurdles, for example in terms of miniaturization, the positioning of sensors or pressure control for compression. The second technical challenge is that the AI models need to be able to be evaluated in real time.

On the regulatory side, there are currently no specific standards or official guidelines for AI-based medical devices. This is a kind of regulatory vacuum. At VDE, we developed processes and documents early on to describe the development of medical AI models, data management and the corresponding technical evaluation of these AI models. We are now incorporating these and using them to support our partner institutions in the project. Another task for us is to identify and analyze the regulatory requirements for this very complex medical product and to efficiently coordinate the resulting tasks.