Unmet Need
Children hospitalized with SARS-Cov-2 infection are at risk for two similar inflammatory syndromes: Kawasaki disease (KD), and multi-system inflammatory syndrome in children (MIS-C). However, current clinical diagnostic standards for these two conditions have considerable overlap, which can lead to misdiagnosis and suboptimal treatment and response. Therefore, there is a strong need for diagnostic tools that can more confidently diagnose children with KD vs MIS-C and guide an appropriate treatment regimen to improvement patient response and recovery.
Value Proposition
· Algorithm to differentially diagnose KD from MIS-C more accurately
· Prediction of treatment response and potential adverse advents to design patient-specific treatments and management strategies
· Ability to predict treatment and treatment response using only data obtained from the first day of hospitalization
Technology Description
Researchers at Johns Hopkins have developed a clinical decision support system to aid in diagnosis and treatment of pediatric patients with suspected MIS-C and/or KD. This system includes predictive models to help clinicians with the diagnosis, treatment, and management of these patients. To predict a patient’s diagnosis, an unsupervised machine learning model was developed based on clinical parameters obtained on the first day of hospitalization to predict a MIS-C or KD diagnosis. In addition, two machine learning models were developed using these parameters to predict how patients will respond to different therapies, and to predict the likelihood of severe complications. Overall, this system provides clinicians with the ability to tailor their treatment of an individual patient to maximize the chance of a successful response and management of illness.
Stage of Development
The model has been validated on patient data, and a prototype user interface has been designed.
Data Availability
Data available upon request.
Publication
WO2025059270 - Machine learning differentiation of kawasaki disease
