IMPROVE HEALTH THROUGH AUGMENTED CLINICAL DECISION ANALYTICS

Using data science to optimize clinical outcomes

data science enabled Clinical DECISION support analytics

Achieve the best clinical outcome through augmented care decisions (1) from the time a patient is admitted for acute care, (2) through her stay and (3) eventual discharge.

Our growing list of solutions (as it stands today):

Sepsis Watch: Reimagine how data and technology quickly & accurately detect sepsis allowing timely delivery of care while reducing variability of outcome
Mortality Model: Accurately predict in-hospital mortality. Improve early clinical decisions at time of admission or during patient stay to affect outcomes.
Cardiac Decompensation: Early Identification of high-risk cardiac decompensation phenotypes via real-time EHR Data

Why Cohere Med?

“My current protocols work. Why should I change anything?” or “I just installed a new system.”

Whether you use rule-based alerts built into EHRs (Epic etc.) or other analytics, the critical questions are:

“Is there room to improve clinical performance?”

AND

“Do financial PLUS clinical benefits substantially outweigh new system/training costs?”

Built by Duke Institute for Health Innovation as part of their ongoing drive to improve population health, our analytics are successfully delivering care since late 2019 in Duke (multiple hospitals in their system) and other installations.

Observed benefits are:

What differentiates patient care analytics?

Traditional analytics

Cohere Med analytics

Generic rule-based alerts; use limited set of clinical parameters available at any given time

Known to generate false positives that lead to “alarm fatigue” which is leading cause of mis-diagnosis or delayed treatment

Designed with inputs from clinical teams for specific clinical settings that each model applies to
Risk, at any given time, calculated using time-series of clinical data available since patient’s admission; this takes patient’s progression into account

Actionable insights created from a database of millions of clinical parameters of over 42,000 patients at Duke resulting in true evidence based medicine

Operational for over two years at Duke with accuracy consistently greater than 90% while reducing variation in clinical outcomes

Full technical discussion of why our analytics stand out is here.

(Sneak peek of the technical discussion: Apart from high-quality patient data, the robust computationally intensive machine learning model development and validation is one of a kind. It would be difficult to achieve without the significant resources of Duke University.)