Detect Sepsis early and improve patient care outcomes using ZASTI Early Sepsis prediction AI
Sepsis is one of the deadliest conditions in the United States, with about 1.7 million adults developing sepsis resulting in 250,000 fatalities annually. Studies have shown that mortality rate increases by 7.6% for every hour the treatment for sepsis is delayed.
Symptoms of sepsis are very similar to those of many other illnesses, making it very difficult to diagnose. There is no single diagnostic test that can tell if someone has sepsis or not.
Rule based scores such as SOFA, qSOFA, MEWS, SIRS are inefficient and time consuming when diagnosing Sepsis.
Traditional automated systems are generally inaccurate and give a lot of false alarms, causing healthcare staff to not respond to alerts quickly enough, as doctor’s time is very precious.
The ZASTI Sepsis AI model has the capability to ingest large volumes of structured and unstructured data such as those found in EHR systems. The model is trained to classify patients, at individual level into one of the four risk buckets (Low, Medium, High and Ultra High) based on the patient’s risk probability score of contracting Sepsis. The explainable AI (XAI) module provides insights into the primary reasons for the predicted results and gives relevant recommendations to manage the risk.