SAN FRANCISCO--(BUSINESS WIRE)--Dascena, a developer of machine learning diagnostic algorithms to enable early disease intervention and improved care outcomes for patients, has received a grant from the National Institute on Minority Health and Health Disparities (part of the National Institute of Health) to develop an unbiased Acute Coronary Syndrome (ACS) prediction tool. This tool will leverage Electronic Health Record (EHR) data to develop a machine learning algorithm that helps diagnose ACS without sex- or race-based bias – ultimately promoting more accurate and equitable disease recognition.
ACS is a major cause of morbidity and mortality in the U.S.; according to the American Heart Association, an estimated 805,000 Americans will experience an ACS event annually. Alarmingly, a growing body of evidence has revealed sex- and race-related disparities in the diagnosis, treatment, and outcomes of ACS patients. Because ACS testing and diagnosis has historically focused on white men, less is known about how the condition affects different populations. For example, while common ACS indicators for men include chest or jaw pain, women might experience back or stomach pain. Because doctors have not been trained to correlate these lesser-known symptoms with ACS, patients are often undiagnosed or misdiagnosed – resulting in significant disadvantages for both women and minority racial groups.
Harnessing the power of big data in EHRs, machine learning algorithms have emerged as promising clinical decision support (CDS) tools for the detection and management of ACS patients. While machine learning algorithms have the potential to play a powerful role in mitigating healthcare disparities, they may also be inherently biased depending on their design – and can therefore contribute to maintaining systemic inequities.
To combat this problem, Dascena is using this NIH grant to develop a machine learning tool that is intentionally designed to diagnose ACS in the emergency department without sex- or race-based bias. This will be accomplished by:
- Analyzing EHR data for potential sources of bias
- Adjusting for any systemic inequities between different subpopulations, while maintaining the aspects of the data that reflect relevant patient measurements and outcomes
- Training the model on preprocessed data
- Assessing the model’s performance across different subgroups using multiple statistical analyses to demonstrate a reduction of bias between subgroups
- Comparing the model against established clinical risk scoring systems to enable a comparison of bias between the machine learning algorithms and current standard of care
“Machine learning presents an incredible opportunity to reduce bias and disparities in how patients are diagnosed and treated,” said Jana Hoffman, Vice President of Science at Dascena. “Through this grant, we aim to harness the power of machine learning to reduce bias in ACS detection – and ultimately apply our learnings to other conditions to create even greater opportunities for equitable care.”
About Dascena
Dascena develops and implements clinically proven machine learning algorithms to improve patient outcomes. These innovations help predict or identify disease, improve the delivery of care, and help save lives. Please visit us to learn more at dascena.com and on Twitter and LinkedIn.
