A Graph-Based Model for Tag Recommendations in Clinical Decision Support System.

The healthcare providers use clinical decision support systems to manage the patients’ electronic health records. In this paper, we aim to enhance the computer-aided diagnosis in medical imaging. We developed a graph-based tag recommendations approach that suggests relevant diseases and pathologies by analysing the tagged medical images. Healthcare providers can rapidly get an improved diagnostic value of radiographs using the graph-based tag recommendations that enable discovering common and relevant diseases used within the patient’s community, his related images and semantically tied tags. The dataset ChestX-Ray14 has been conducted to evaluate the accuracy and effectiveness of our proposal. Futures works will address the online evaluation of the suggested tags by exploiting the healthcare providers’ feedback.