Skip to main content
Home News Cohere-Med Spins Out Suite of AI-based Predictive Analytics

Cohere-Med Spins Out Suite of AI-based Predictive Analytics

Cohere-Med Inc., a San Francisco based health data science startup, has spun out a suite of artificial intelligence-based predictive analytics and algorithms from Duke University.

Following last year’s announcement of licensing the Sepsis Watch algorithm from Duke University, Cohere Med has further strengthened its commitment towards early detection of disease onset in the areas of cardiac failure and mortality in addition to sepsis through predictive analytics by becoming a Duke co-owned startup.

According to the National Center for Health Statistics, approximately 2% of patients admitted to US hospitals die during the inpatient admission. Nationally, patients diagnosed with sepsis have a 28% mortality rate with an in-patient hospital death-rate of 16.3%. 1 in 3 patients who die in a hospital has sepsis. Apart from Sepsis, in-hospital cardiac arrest is also very common and one of the largest contributors to hospital mortality rates. An estimated 290,000 in-hospital cardiac arrests occur each year in the United States with a survival rate of only 25%.

Efforts to reduce preventable in-hospital mortality have focused on improving interventions and care delivery. Also, improving the quality of life in terminally ill patients by integrating patient preferences and extending care outside of traditional facilities show a positive patient experience, and lowers hospital costs. Early identification of patients at high risk of in-hospital mortality may improve clinical and operational decision-making and improve outcomes for patients.

To test this possibility, a machine learning model that predicts in-hospital mortality at the time of hospital admission for adult patients was prospectively and externally validated The model uses readily available EHR data and accessible computational methods, can be applied to all adult patients and was designed to be integrated into clinical care to support workflows that address the needs of patients at high risk of mortality. Duke Health evaluated the model on retrospective data from 3 hospitals and prospectively on a technology platform integrated with the production EHR system. This machine learning model, designed to be implementable at a system-level demonstrates a good discrimination in identifying patients at high risk of in-hospital mortality and can be used to improve clinical and operational decision-making.

Assessing patient risk with standard medical practices is not new. Although they have evolved over a period of time, their accuracy has always been a question. Moreover, an increased number of false positives has resulted in silencing alerts which eventually leads to alarm fatigue. The Sepsis Watch machine learning model has significantly reduced the percentage of false positives, which results in increased confidence on alerts for the caregivers. “Machine learning today has reached a phase where it’s no more a theoretical mathematical exercise, but pragmatically put in practice widely at Duke to improve patient outcomes”, says Suresh Balu, Associate Dean and Program Director, Duke Institute for Health Innovation.

“The overall approach for assessing and predicting risk for any patient in a hospital needs to be thought through three phases. One, in general, the 24 to 48-hour mortality, two, whether a patient will be septic in the next 4 to 5 hours, and three, will these acute conditions lead to cardiac decompensation”, says Dr. Manesh Patel, Chief of Cardiology, at Duke and co-director of the Duke Heart Center

Cohere Med Inc., a clinical analytics company headquartered in the US strives to provide predictive solutions to clinicians to help improve clinical outcomes and reduce the cost of healthcare. Its patent-pending algorithm to detect arrhythmia across a diversified set of sources adds well to the portfolio of predictive analytics. “Cohere-Med’s agile methodologies and Software-as-a-Service (SaaS) based application integration design helps hospitals to minimize change management and adapt to machine learning-based early detection systems quickly”, according to Renjith S. Nair, Chief Technology Officer at Cohere Med.

“We are excited to be part of this journey with Duke University and look forward to scaling these high-impact solutions globally. Given the ongoing pandemic, it becomes even more important for us to re-imagine application of data science”, says Srikanth Muthya, President and CEO, Cohere-Med Inc.

 

References:

 

 

Have Questions?

Please contact us or subscribe for more opportunities

Stay in Touch with Us