Diagnosing and predicting risk in children with SARS-CoV-2- related illness

In the wake of COVID-19 pandemic, Multisystem Inflammatory Syndrome in Children (MIS-C) has evolved as a new threat to children exposed to SARS-CoV-2. Currently there are no diagnostic tests to readily identify these patients nor are there tools to predict disease progression. Through established, funded, multi-center consortia in the U.S. (CHARMS: Characterization of MIS-C and its Relationship to Kawasaki Disease funded by PCORI) and the UK (DIAMONDS), we have collected clinical data and samples to support development and testing of diagnostic and prognostic tests for MIS-C. In phase 1 of this project (R61HD105590, years 1-2), we used these data to devise machine learning-based clinical decision support tool (KIDMATCH) to assist clinicians in diagnosis of MIS-C and other causes of fever in children. This proposal focuses on the development of a Software as a Medical Device (SaMD) and clinical deployment of this SaMD in a clinical setting. Additionally, we aim to develop a quality management system (QMS) to minimize and manage unintentional outcomes related to patient safety as required for FDA regulatory approval. To this end, this proposal aims to further develop a cross-platform SMART- on-FHIR app for display of the KIDMATCH outputs for integration within electronic health records. In close collaboration between our clinical and data science teams, we will evaluate the KIDMATCH app set-up, user workflow and physician preferences through structured interviews, simulations, and direct feedback. Finally, we aim to complete an ongoing application for Emergency Use Authorization (EUA) to enable wider deployment and commercialization of the KIDMATCH SaMD. The synergistic expertise of the investigative team in this proposal provides a unique opportunity to create diagnostic tools for children suffering from the spectrum of SARS- CoV-2 illnesses. 1