Sponsored

Leveraging Machine Learning to Maximize Outcomes -- and Savings

By: Eric Hamborg, Co-founder and Chief Commercial Officer, MOBE 

Today, payers and large employers are increasingly sophisticated in their analysis of claims data, and for good reason: assessing members’ health needs is key to addressing and even preempting debilitating health problems. With the capacity to capture and analyze millions of data points efficiently, savvy health technology companies are providing tools for payers, employers, providers and even consumers to put these troves of health data to use.

But in order to move the needle on health outcomes and health system costs, high-tech solutions -- from apps to wellness programs to back-end, in-house claims analysis -- can’t rely on data science alone. Successful programs not only find the right people to pair with the right intervention, they keep them engaged. In other words, the winning formula is machine learning combined with human connection.

Data show this is especially true for people struggling to meet their health goals. Many Americans – 5% of those who receive private insurance – make up a “hidden population” of people who are accessing the healthcare system at a very high rate but aren’t getting better. In fact, these 5% are responsible for approximately 20% of employers’ health care costs – totaling $158 billion each year. 

So, how can payers and employers deploy machine learning technology to connect this population with the care they need most, without adding cost to the system?

Use Data Science to Crack the “Claims Code”

We’ve long known that certain individuals are high-risk and high-utilization when it comes to healthcare, but identifying them isn’t as simple as a single ICD-10 code. Often, the people who can benefit most from additional health support are managing multiple chronic conditions -- and individual life experiences -- making the “hidden population” highly heterogeneous. This is why health and wellness programs focused on a single condition sometimes fail to improve overall health and reduce healthcare utilization -- and where machine learning comes into play.

By using models that assess data across diagnoses, frequency, and treatment effect, payers and employers can better pinpoint the people who most need, and are most likely to engage successfully with, support programs. Additionally, because machine learning grows more fine-tuned with every new data population it’s exposed to, including data on participant experience alongside claims data is crucial. This enables the model to learn, in the most literal sense, why what works for one person may not work for another, and vice versa. And, with 56% of healthcare providers seeing patients with multiple chronic conditions at least bi-monthly -- and 76% of people with multiple chronic conditions reporting their health hasn’t improved in the past year (source: The 2020 Chronic Care Action Index) -- the need to find, reach, and engage these people is urgent.

Combining Data Insights with Personalized Support

Algorithms may be the hero in identifying individuals in need -- but it’s the human connections they enable that drive real impact in health outcomes.

Once this “hidden population” is uncovered, successful health and wellness offerings engage these specific individuals with tailored support. Personalized health guidance on medication, nutrition, fitness, and other factors impacting wellness like sleep and mental health are proven to make a difference in individual outcomes, and in costs.

Though it may sound basic, focusing on the foundational elements of health is precisely where we need to start in order to see change.

The Chronic Care Action Index found that exercise (51%), eating healthier (40%) and getting more sleep (38%) were the changes people most wanted to make in relation to their health -- but found most difficult to make. What’s more, nearly one-third of respondents (29%) cited motivation as a barrier to following their doctor’s guidance. Helping people address and solve these issues is foundational to increasing their health and happiness, while also reducing health system burden and costs.

In order to meet these objectives, health and wellness offerings must achieve high participation rates. This means providing solutions and support where and when individuals need them. Whether via video chat, text, phone or otherwise, programs need to utilize methods that are easy and accessible. Technology makes it simple to create applications that go on phones for on-the-go access and set up an easy line of communication through text, video, or calling. No matter the avenue in which the individuals are engaged, if it is quick and painless, they are more likely to engage. However, it’s critical to pair the ease of access with human connection. When individuals have the personal support and it’s available when they need it, they are more likely to continue towards their health and wellness goals.

Health and wellness offerings are not one size fits all. Data science holds the key to reaching those most in need, but the most successful programs leverage those insights to provide a personalized and accessible experience. Leveraging the power of AI, machine learning, data science, and sophisticated algorithms has unlocked the best-in-class health and wellness offerings that hold the power to not only improve health outcomes but lower costs. That’s a win-win solution.
 

MOBE was founded in 2014 to address a significant unmet need in the health care system: helping people who are frequent users of health care services, but are not experiencing optimal health outcomes. MOBE partners with insurance companies and large employers to provide health solutions to their members and employees at no additional cost to the health plan, the employer, or the individual. Combining data analytics, digital health and a novel one-to-one personalized approach, MOBE helps people live happier, healthier lives.

 

This article was created in collaboration with the sponsoring company and our sales and marketing team. The editorial team does not contribute.