Explainable models for interpreting data from biological and medical sensors based on 2D materials (XAI4BioSense)

Innovative sensors made from atomically thin materials (so-called 2D materials such as graphene) can detect pathogens and biomarkers with extreme sensitivity and speed. These sensors function like tiny electrical switches that react to specific molecules and generate measurable signals in the process. For practical use in medicine, these sensors must operate reliably with complex bodily fluids.

The XAI4BioSense project is developing AI models that can precisely determine the type and quantity of detected pathogens or biomarkers based on the sensor signals from complex samples. The reliability and interpretability of the analysis are crucial for medical applications. The project team therefore relies on explainable AI models that not only deliver results but also show how they arrive at their conclusions. These models take into account the physical and chemical processes within the sensor and can identify which factors influence the signal. This creates a transparent system that can also detect and compensate for unwanted interference.

Knee joint infections serve as the model system. However, the methods can be applied to many other applications. The goal is rapid diagnosis directly at the bedside or in the operating room, enabling doctors to adjust treatment immediately.