Innovaccer, which develops data tools to help health systems with accountable care and population health management, has published a new research paper focused on the ways advanced machine learning algorithms can improve risk models to better predict the future cost of care.
WHY IT MATTERS
The company’s model analyzes data from multiple data sources, including electronic health records, social determinants of health and claims data, officials say. With the estimation of future risk scores, healthcare providers can better gauge the likelihood of outcomes such as the number of hospital admissions or emergency department visits.
Innovaccer suggests its AI-based model could be an improvement over other claims-based risk models, such as the Centers for Medicare and Medicaid Services’ Hierarchical Condition Category, the ACG System developed by Johns Hopkins and DxCG from Verisk Health. Those are good at forecasting the risk for populations, rather than focusing on an individual level, according to the company.
By observing key clinical and lifestyle metrics, and using analytics to determine how these metrics impact overall outcomes, health systems could reap huge benefits by developing and assigning risk scores by surveying expansive groups of patients with comparable features.
THE LARGER TREND
The research paper notes that as the healthcare industry moves to a value-based care model, this will require providers to improve the quality of care being provided to patients, and improve the overall status of population health – something Innovacer says AI could help facilitate.
The company’s algorithm uses an ensemble of six different regression models, including Lasso and Elastic regression models and 62 independent features to predict the future cost of a patient.
In addition, the use of varied regression models means Innovaccer’s risk model is able to account for outliers present in the data.
The study also highlights Innovaccer’s approach in estimating the future cost of care based on past medical history, clinical and socioeconomic data, as well as additional factors.
“Every patient is different and just because the technology in healthcare is still stuck in the pre-internet era,” Abhinav Shashank, CEO and co-founder of Innovacccer, said in a statement. “Healthcare needs an information superhighway that can open up the space for innovation. With this new research, we take a step ahead and move one step closer to preventive care.”
Using AI as part of the transformation to value-based care could have wide-ranging impacts, from helping to eliminate variations in care quality to ensuring accurate reimbursements.
ON THE RECORD
“The transformation from volume to value requires innovative strategies for assessing risk and predicting outcomes,” said former Geisinger CEO Dr. Glenn Steele, now vice chair of the Health Transformation Alliance and an advisor to Innovaccer, in a statement.
“This innovation must be based on a solid data foundation and it’s encouraging to see Innovaccer’s data-driven approach being applied to an AI-based risk scoring model – something that can address one of the most pressing needs in healthcare today,” said Steele.
“Simply identifying why a patient is sick is not enough,” said Dr. David Nace, chief medical officer at Innovaccer and author of the paper. “We need to ensure that when that stage arrives, everyone involved in the patient’s care journey is equipped with the best insights.
“With this research, we hope to open new ways of delivering care,” he added. “This journey does not just stop with identifying the high-risk patients but by providing every possible assistance they need. It’s how we come closer to a patient-centered healthcare.”