SepINav, which stands for Sepsis ICU Navigator, is a medical informatics project developed to assist ICU practitioners and researchers in the efficient and interactive monitoring and intervention of current sepsis patients, as well as conducting retrospective studies. The software is data-driven and leverages advanced machine learning techniques, specifically Bayesian Online Change Point Detection (BOCPD), to recognize structural alterations in a patient's vital sign regimes that may signal the harbinger of septic shock. This capability is crucial for providing an early warning system in the critical care environment.
The tool is designed to plug into a hospital's Electronic Medical Record (EMR) system to extract comprehensive patient data, including vital signs, lab tests, interventions, and prescription records. This EMR integration allows SepINav to provide a holistic view of the patient's trajectory and the interventions made by practitioners. By identifying changepoints in vital signs, SepINav aims to bridge the perceptual gap between explainability and computational intricacy in machine learning-based medical informatics solutions, offering a custom-tailored approach to a disease shaped by complex host and pathogen factors.
Key Capabilities:
- Real-Time Monitoring: Provides efficient and interactive monitoring of existing sepsis patients.
- Early Warning System: Uses BOCPD to detect structural changes in vital signs that may precede septic shock.
- Retrospective Analysis: Enables researchers to conduct studies to find rationales for various sepsis scenarios.
- Data Integration: Extracts and processes data from hospital EMR systems (patient ID, vital signs, lab tests, prescriptions, interventions).
As an academic/research tool, SepINav is open-source and primarily implemented using Jupyter Notebook and R, requiring internal deployment within a clinical or research IT environment.