The digital health market is full of companies promising to transform healthcare, but investors and health systems are finally demanding to see evidence from startups that actually prove their tech’s value proposition in clinical trials.
San Francisco-based KelaHealth is an example of one company that appears to understand the importance of demonstrating its worth through published research — it released a new study last week detailing the efficacy of its surgical intelligence platform. The study found that the platform was associated with a 21% reduction in acute kidney injury and a 24% decrease in costly readmissions following colorectal surgery.
Spun out of Duke University in 2018, KelaHealth is a surgical data analytics company that delivers insights about individual patients’ risks and what their care journeys should look like. The startup’s platform employs a three-pronged approach focused on predicting risks, providing recommendations and tracking outcomes, said CEO Dr. Bora Chang.
For instance, the platform may alert a surgeon that their patient is at risk for a respiratory illness or heart attack. In this case, the platform may recommend the surgeon adjust the patient’s medication before surgery or do another consultation.
“We take a patient’s clinical history and recommend a tailored journey for that patient. We don’t think that each patient is just a unit that goes through the system or that everyone’s the same. We really feel that surgeons need technology to better predict, anticipate and personalize the care that they’re delivering to patients,” she explained.
The study was led by Duke researchers and was not sponsored by KelaHealth. The researchers used the startup’s platform to identify risks among 665 patients undergoing colorectal surgery at Duke in 2020. These patients’ outcomes were then compared to a control group consisting of patients who underwent colorectal surgery at Duke between 2014 and 2018.
The rate of both post-surgical acute kidney injury and hospital readmissions was less among the group whose risks were identified using KelaHealth’s platform, the study showed. The rate of acute kidney injury was 8.8% in the experimental group compared to 11.3% in the control group. As for the rate of readmissions, it was 8.9% in the experimental group compared to 12% in the control group.
For this study, KelaHealth trained its platform on Duke’s own patient data. While many AI models are trained on national or commercially available datasets, doing so hinders their ability to flag local population-based nuances, Dr. Chang pointed out. The platform analyzed 3,617 patient records and amassed more than 200 variables per patient spanning demographics, clinical history and lab values.
“As a colorectal surgeon, it is encouraging that AI may be a useful tool for predicting risk and preventing post-surgical AKI and readmissions. The use of a risk-prediction platform like Kela’s to help surgeons make more informed treatment decisions could be a game changer for our field,” said Dr. Christopher Mantyh, the study’s senior author as well as vice chair of clinical operations at Duke Health.
Acute kidney injury from colorectal surgery is linked to a higher risk of death, shorter survival time and development of chronic kidney disease. The unplanned readmissions that result from this complication costs the U.S. healthcare system $41 billion annually.
Picture: DaevichMikalai, Getty Images
