A new report by Milliman concluded that, after it tested the impact of various modifications to Together Senior Health’s Dementia AlgoRithm, new claims-sourced models demonstrated positive predictive value to support quality care.

WHY IT MATTERS

Together, part of Linus Health, developed its proprietary cognitive-impairment identification and stratification algorithm – validated using electronic health record data – to help find patients with elevated risks for undiagnosed Alzheimer's disease, dementia and cognitive decline.

“RADAR offers payers and providers a quick, highly efficient method for identifying their patients with the highest risk of having undiagnosed dementia,” Deborah Barnes, cofounder of Together, said in the report announcement. “Most importantly, RADAR is highly accurate, according to the results of this thorough analysis from Milliman.”

With new RADAR models that use claims data, overall predictive performance might be twice as high as predictive mammography algorithms and three times that of predictive colonoscopy algorithms, according to Dr. John Showalter, an internal medicine physician specializing in dementia and chief strategy officer at Linus Health.

“The evidence clearly demonstrates that the gold standard for breast and colorectal cancer screenings yield predictive results on par with the refit RADAR model,” he said in a statement Tuesday. 

“RADAR, however, is far superior in that it does not require any patient time, discomfort or involvement at all, nor is it anywhere near as costly as these common, guideline-directed procedures.”

Actuarial firm Milliman evaluated RADAR using administrative claims data to test the impact of various modifications, including recalculating the model weights and evaluating the marginal efficacy of including new prediction variables or predictors. According to the new report, the process required creating cohort datasets, running the models, calculating summary statistics and evaluating the new predictors.

The new claims-based models could help drive proactive care and diagnoses. When combined with clinical insights, they could “enable payers and providers to secure more accurate and appropriate risk-adjustment results that support quality care,” Mike Butler, independent board director of Linus Health and former president of Providence Health, added. 

“Based on my decades of experience overseeing value-based care and risk-based contracts, RADAR’s ability to rapidly stratify populations for undiagnosed cognitive impairment is transformative.”

“When using claim-based predictors, accuracy for predicting the risk of undiagnosed dementia is significantly improved by refitting the model to reweight the predictors,” Milliman said in the report. 

Together is also developing and testing versions of RADAR that include claims, EHR and other data sources that are not included in this report.

THE LARGER TREND

While healthcare artificial intelligence and machine learning models are often pursued to replace manual tasks, that’s not how brain health technology companies like Linus view AI, according to David Bates, the company’s CEO and cofounder.

“In brain health, AI is unlocking a whole world of possibility for earlier detection – and intervention – at a time when the urgency has never been greater to find different approaches that can change the global outlook on brain health and aging,” he previously told Healthcare IT News.

Surfacing often imperceptible signs of cognitive impairment by leveraging AI to augment clinical expertise and analyze cues and patterns of performance during cognitive tests are only part of the picture, he said.

When clinical cognitive-test metrics are combined with other inputs like health history, life habits, symptoms and social determinants of health, AI can help empower providers “with new visibility into their patients' brain health so they can, in turn, empower their patients with education and next steps as appropriate,” he said.

ON THE RECORD

“Health systems want to do the right thing to care for their aging patients, but struggle with targeting cognitive assessments to the right people,” Butler said in a statement. “The RADAR algorithm is a game-changer, especially for Medicare Advantage plans and value-based care providers.”

“Due to the historic care demands facing primary care practices today, providers and payers need to carefully triage which patients should first receive more in-depth cognitive impairment screenings, so those with higher risk can begin receiving preventive interventions as soon as possible,” Barnes added.

Andrea Fox is senior editor of Healthcare IT News.
Email: afox@himss.org
Healthcare IT News is a HIMSS Media publication.