OR WAIT null SECS
Machine learning technologies offer providers opportunities to address rising risk and deliver better care.
By design, healthcare is largely reactive in nature. After experiencing symptoms, a patient sees a provider, receives a diagnosis, and is treated. That provider is then paid for services delivered and measured on reactive performance (i.e., how quickly a heart attack was treated appropriately or how well post-op pain was controlled), which means there is no incentive to promote healthy behaviors and avoid high-cost treatments.
But the shift to value-based care, with its promise of lower costs and better population health, has opened the door to new models that engage with people proactively-when they are at “rising risk” for worsening health and skyrocketing costs, rather than waiting until they are already high-cost, high-acuity patients. And luckily, today’s advanced technology is here to help.
Leveraging technology to assess future risk
Traditional prospective risk tools are driven by current risk and grounded in the heuristic, rule-based model that high-risk members today will be high-risk tomorrow. While adequately identifying persistently high-risk members, they fail to identify individuals who are low cost today and may seem “healthy,” but will become high cost tomorrow.
It no longer has to be this way. Advances in technology-including predictive analytics, artificial intelligence (AI), and machine learning-are a making the dream of proactive care management a reality. Using sophisticated algorithms to analyze numerous data streams, these systems can accurately predict future risk by pairing pattern recognition with known outcomes. Providers are able to match machine learning predictions that identify rising-risk patients with suggested actions that can be taken, then engage with and support them in proactively changing behaviors before they experience an acute episode.
Consider the ability to proactively engage with individuals at rising risk of worsening diabetes. In the past, a provider would need to identify and reach out to every patient diagnosed with pre-diabetes, diabetes, or metabolic syndrome. Such a practice would have identified many diabetic members who are either well-controlled or unwilling to engage, and would have offered little opportunity to engage with the small subset who were both in need of and willing to change their clinical trajectories. AI-enabled technology now makes finding and focusing on that small group of rising-risk members open to engagement and change a reality.
Benefits of a proactive care management model
Beyond improving patient health and lowering costs, this proactive model enables care managers to take an integrated approach aligned with the shift to value-based holistic care. By training the data sets to understand the characteristics of patient who are likely to engage, and those who are not, care managers will be able to spend more effective time with the members they can help the most, and less time with those they cannot help. Consider how fundamentally an entire organizational structure and culture could be redesigned around proactive engagement.
Imagine a patient in her 50s with heart disease, chronic pain treated with opiates, tobacco use, anxiety, and depression. She hasn’t been hospitalized, but she’s had more than 20 outpatient visits over 12 months and her risk of experiencing an acute episode is rising every day. A reactive care management model wouldn’t find her until she was hospitalized. But what if an AI-enabled partnership identified her unique, individual risks and suggested that a provider engage with her? The provider might then reach out, discover that she needed a much deeper connection with counseling, and help her transition off opiates and quit smoking.
This isn’t an imaginary scenario – it’s a real-world example of a patient who would have fallen through the cracks of our reactive healthcare system. Instead, using AI-enabled predictive analytics, the care manager got to the patient at the right time, in the right way to both dramatically improve her health trajectory and reduce her total cost of care.
Ready or not, the AI-enabled future of healthcare is here. By finding the most engageable and clinically impactable members before they experience acute issues, machine learning technologies offer providers opportunities to address rising risk and deliver better care sooner.
Chris DeRienzo is chief medical officer for Cardinal Analytx Solutions, leading the company’s clinical portfolio and helping connect Cardinal’s world-class data science to better patient outcomes. Formerly chief quality officer for Mission Health, Dr. DeRienzo is Board Certified in both Pediatrics and Neonatology, and completed his MD, Masters in Public Policy and post-graduate medical training at Duke.
Erica Kaitz is care management lead for Cardinal Analytx Solutions, contributing to the company’s clinical portfolio and helping connect Cardinal’s world-class data science to better patient outcomes. Formerly the behavioral health director for Cigna Health-Spring of Illinois’ government programs, Erica Kaitz is a licensed clinical social worker, and completed her Masters of Social Work at Columbia University and post-graduate training in psychiatric crisis and community mental health settings.