Kennedy Brown


2019-11-21 15:38:21 (GMT) Healthgen AI Inc has defined a solution for predicting patient sepsis risk across the care continuum. This solution is unique by identifying early prediction of the highest risk patients in both inpatient and outpatient settings. This allows hospitals, nursing facilities, primary care organizations, accountable care organizations and payers to focus their resources where they are most needed to improve patient outcomes and reduce costs.


The challenge with sepsis is that it’s difficult to diagnose early and treat, with multiple conditions and events leading to this infection along with poor follow-up. Our solution is multi-pronged using patient claims data and machine learning models to bucket patients into sepsis risk levels: High / Medium / Low. It will present alternative and optimal time-dependent care paths for patient outcomes and associated costs. Local EHR data and social population data can be added to improve accuracy levels and optimize for facility and regional dependencies.


Healthgen AI is looking for partners with datasets to run experiments, train and test their models. We will use industry standard methodologies to analyze historic claims data collections focusing primarily on sepsis outcomes and least-expensive routine checkups as control. Explainable AI techniques will determine and utilize the most important features and their interactions to be fed into time-series and deep learning models. Typical events leading to sepsis include Pneumonia, Dehydration, CHF, UTI, Kidney, Skin Ulcers, COPD, Diabetes, Previous Sepsis, Infections, Hospital Stays, and Antibiotic Use. However, the models may find other claims features in the data that are relevant including facility or regional-based dependencies. This solution can include a unique identifier for sepsis risk levels enabling continuity of care across Hospitals, Nursing Facilities and Primary Care Providers, and allow tracking/reporting on sepsis frequency, time to onset and resolution, readmission rates and best performing treatment plans.