This study introduces MDscan, a cutting-edge tool for screening ten mental disorders, addressing the global challenge where nearly 94% of the one billion affected individuals remain undiagnosed due to a clinician shortage. Utilizing the SCL-90-R screening instrument, an explainable AI method, and our ShapRadiation algorithm, MDscan transforms 90 mental health indicators into diagnostic images for efficient interpretation. This innovation significantly eases clinicians' cognitive workload, enhancing their capacity to screen more patients rapidly. Field evaluation with clinical data reveals MDscan's impressive classification accuracy, with F1 scores ranging from 0.77 to 0.94, validated against clinician-assessed data. Unlike traditional AI systems, MDscan offers transparency and explainability, fostering trust in its clinical application. This tool represents a significant stride in utilizing AI for mental health diagnostics, potentially revolutionizing patient screening processes in clinical settings.

In the United States, Advanced Practice Providers (APPs) such as physician assistants (PAs) and nurse practitioners (NPs), are often utilized to address both the shortage of physicians and to reduce costs. These medical professionals have significant medical training, but are not medical school graduates/physicians, and are engaged to treat patients and reduce the load on physicians. They are typically less expensive as employees. However, there are ongoing questions whether APPs actually provide the same quality of care as that provided by physicians. Using data from the US Centers for Medicare and Medicaid Services (CMS), we compare the quality performance of primary care groups that utilize APPs to those that do not. Our results show that APP utilization has an interesting and somewhat unexpected relationship to primary care quality. This study has implications for policymakers, healthcare managers, patients, and scholars

This paper proposes an exploratory-explanatory decision support framework for identifying and explaining risk factors associated with graft survival during organ transplantation procedures. Our framework makes use of the Elastic Net for effective feature selection, as well as Bayesian Belief Networks and Shapley Additive Explanation (SHAP) approaches for assessing the global and local importance of graft survival risk factors. We demonstrated our proposed framework using a large sample of patients who had undergone kidney transplants. Our findings show that the importance of risk factors varies greatly depending on the individual patient's unique characteristics, despite their global significance. This insight transforms our understanding of graft failure risk and optimizes kidney transplant operations, demonstrating the value of explainable analytics in healthcare decision-making.

This study aims to understand the impact of the rarely co-occurring combination of diseases on the mortality rate using network science methods. To model the rare combination of diseases, structural holes in the comorbidity network are identified, and the variables measuring the distance between diseases in the network are computed. The results show that the patients diagnosed with the rare combinations of diseases have a higher mortality risk than the patients with diseases that co-occur more often among patients.