At Waymark, “rising risk” is not a single, static category of patient –– rather, it’s a descriptor we use to indicate how close a patient is to an acute event, and at what pace they’re speeding towards that event.
Waymark defines “rising risk” as a group of patients who have an increasing chance of having an emergency department (ED) visit or hospitalization for a reason that could have been prevented or treated through adequate, timely and quality primary care.
For many rising risk patients, there is often a critical window of time to reach and engage them before their health changes for the worse — and they end up in the ED or hospital as a result. This window of time could be when someone is at risk for relapse to substance use, or has run out of chronic disease medication, or is at risk of an asthma attack due to wildfire smoke traveling over their neighborhood. In all of these instances, proactive data science can help care teams identify risk factors and intervene early to prevent avoidable disease complications, ER visits and hospitalizations that negatively impact both patient health and costs.
This approach is designed to be more upstream than the high-risk care management programs used by many Medicaid health plans. Rather than focusing solely on addressing the needs of high-cost claimants, we prioritize reaching and engaging rising risk patients before they cross into that category. We do this by deploying community-based care teams who are partnered with primary care providers (PCPs) and leverage our proactive data science technologies to reach and engage rising risk patients. By combining local care teams with proactive data science, we can intervene early to help ensure patients stay healthy and don’t become high-cost claimants.
In contrast, one standard approach to serving patients with complex clinical and social needs is to perform a rapid-screening checklist. While well-intentioned, this approach misses patients who are unable to attend primary care appointments and lacks a nuanced understanding of their deeper story, including their goals and priorities (which may not align with screening checklists). We address this concern in several key ways:
- Leveraging data to ensure we’re meeting patients in the critical window of opportunity to address their rising risk, even if it’s outside of typical business hours, at a local library, or in other non-traditional times and settings.
- Analyzing and understanding the ever-changing population dynamics and health determinants that contribute to whole-person health, well beyond standard healthcare data and screening checklists.
- Prioritizing human and social connections alongside medical interventions by using artificial intelligence (AI) and machine learning (ML) to reduce rote work (and in turn, enabling our team to spend more face-time with patients and less time on forms and faxes).
Setting a new standard for data-powered healthcare
Proactive data science can enable us to better target and engage rising risk populations. Waymark SignalTM, our rising risk ML tool, is a great example of this. Signal combines data on social risk factors and patient risk trajectories with healthcare utilization to identify patients at risk for preventable ER and hospital visits with >90% accuracy in a validation sample of over 10 million patients receiving Medicaid nationwide.
There are two components to our utilization of Waymark Signal that allow us to deliver more effective and equitable interventions for rising risk populations:
- Integrated data streams. We built Signal’s algorithm to intake a variety of data that encompass factors beyond traditional healthcare data, including social determinants of health and metrics of risk trajectories that are often unavailable to most Medicaid health plans. Additionally, we found that only when using novel ML algorithms that allow us to capture complex combinations of social determinants and medical needs can we improve risk prediction. For example, information on air pollution and respiratory disease, or information on food insecurity and diabetes, can help us direct people to the right intervention at the right time more reliably than just having simple indicators of a social or medical factor in isolation. Based on our algorithm’s predictive measures, we’re able to reach as many patients as possible before their acute events. Additionally, we’re continually refining our engagement strategies to help ensure we’re reaching patients both virtually and in-person at the critical moments to keep them as healthy as possible.
- Reducing disparities in risk prediction. Historically, risk prediction models in Medicaid are biased toward patients who have better access to more expensive tertiary hospitals. Because Black individuals typically have less access to higher-cost tertiary care centers, traditional cost-based models often under-predict their future costs and assume the lower costs reflect lower health needs. Waymark Signal neutralizes this bias, demonstrating higher sensitivity for Black patients’ needs and offering one approach for a more equitable application of machine learning to providing services to patients receiving Medicaid.
Shifting patient experiences with SDOH in mind
For many patients, it takes a human, social, and medical connection to address the root cause of their most prominent challenges. That’s why we prioritize finding patients who may be at risk for ER visits over the weekend every Thursday – that narrow window of opportunity enables us to prevent acute care visits because we’re reaching them at the critical moment when they need us most.
To many patients, that “critical moment” may not seem like it has much to do with healthcare at first. In Washington state, Waymark CHWs sometimes connect with patients receiving Medicaid following a period of incarceration. Sure, healthcare may be the last thing on their minds at the moment, but we know that they may only have 10-14 days’ worth of prescriptions in their bags, meaning that their risk is rising by the moment (indeed, a classic article in The New England Journal of Medicine shows a spike in their risk of death two weeks after release from incarceration). For the Waymark team, this is the opportunity to meet people where they are and offer connections to resources before their care needs become acute.
Our approach also ensures we can get ahead of patient concerns that prevent them from accessing care. For example, our pharmacy team proactively contacted patients to find an effective substitute for their asthma medication after a key inhaler was discontinued from the market. In Virginia, CHWs connect patients with toll relief resources if toll costs represent a financial barrier to receiving care. Because our teams live in the same communities as our patients, they understand the structural and systemic barriers to care in their communities which many organizations aren’t designed to address. This hyperlocal knowledge is critical in helping us engage rising risk patients before they become “high cost claimants” or “high utilizers,” after which – according to several randomized trials – interventions have unfortunately had more limited success.
Redefining AI’s role in patient experience
When many professionals in the healthcare space consider the implications of AI, they often point to instances that could replace human interaction with automation. While that’s certainly one way to augment healthcare with AI, Waymark’s approach posits a different perspective: AI is a tool to automate repetitive tasks to leave as much time and bandwidth as possible for our care teams to engage meaningfully with patients.
Our approach to AI and ML is focused on reducing administrative work so that our care teams can spend more time listening and responding to patients and their concerns, problem-solving with them and helping them achieve their goals. Our AI technology integrates a human-in-the-loop approach to ensure our care workers are able to provide recommendations for modifying our AI algorithms to reflect the rapidly changing needs of our patients. Waymark’s partnerships with PCPs underscore this focus.
Ultimately, our model is centered around a deep human understanding of our patients’ stories and the challenges they’re facing — and this focus informs how we reach, engage, and interact with them at critical moments in their lives.
