The virtual doctor prescribing the future: Diagnostics with interactive clinical decision support

The growing shortage of healthcare professionals, particularly in rural areas, poses a significant challenge to timely and effective medical care. To address this issue, we present the Virtual Doctor, a walk-in cabin designed to collect and analyze medical data without direct medical supervision. This system integrates non-invasive biomedical sensors for continuous data acquisition and leverages artificial intelligence based decision support to facilitate early risk detection and enhance healthcare accessibility. The Virtual Doctor employs a distributed medical data acquisition system composed of multiple components that are detailed in this study. The collected biomedical data is processed by two pre-trained deep learning models developed on external datasets to predict the risk of type 2 diabetes mellitus and skin cancer. The diabetes model achieved an ROC-AUC of 0.729 on the test set, while the skin cancer model attained a mean ROC-AUC of 0.81. Beyond technical development we evaluated the socio-psychological acceptance of the Virtual Doctor through a survey of 1217 participants focused on two use cases, routine skin cancer screenings and antibiotic prescriptions. The developed model explaining users' intention to use the Virtual Doctor showed meaningful results for both scenarios (R² ≥ 0.604). The results suggest broad general acceptance of the system while also highlighting the need for comprehensive explanations during serious disease screenings to reduce anxiety and potential resistance.