No systematic approach to detect invasive GBS infections in adults
Manual review of medical records delayed the process of reporting the invasive status of the disease in patients
We built a descriptive and predictive analytical framework to identify patients with invasive GBS long before symptoms appear, avoiding hospitalization and future complexities. We leveraged real-world data to mine patient-level insights from hospitals, including clinical and pathologic findings. Using machine learning models, we automated the identification of key drivers contributing to the invasive properties of GBS, resulting in a faster diagnosis.
Through our predictive model, the customer identified invasive patterns in patients with GBS before the onset of symptoms. High vitals, such as the patient temperature at the time of hospitalization, and lab values, such as aspartate aminotransferase, albumin, and more, were identified as key predictors that helped in determining the invasive status of the infection.
Accuracy in early detection of invasive GBS
Clinical features supporting the prediction
Machine learning modules tested