Analytics Can Make Behavioral Health Work

In medicine, value-based pricing has gained a lot of adherents who want to put some risk on the pharma companies and tamp down on irresponsible marketing or prescribing. But behavioral health is usually licensed by health plans on a flat monthly basis, not linked either to the number of members treated or how well they do. According to Peter Shalek, Chief Product Officer of AbleTo, this unfortunately allows a lot of pseudoscience to pass as behavioral health.

AbleTo distinguishes itself by resting its service on quality. They support value-based care by promising outcomes determined by the patient, provider, and coach. It’s up to AbleTo to engage the member of the plan, keep them engaged through the eight-week program, and produce a desired result. This article describes how they do it.

AbleTo’s program falls under the category of cognitive behavioral therapy (CBT). The program is data-driven and evidence-based from the very start of the process, which consists of choosing health plan members who might benefit from the program. AbleTo’s tools check claims data to find good candidates. It then contacts them through phone calls or–increasingly–digital media, and signs them up for a targeted program based on their needs. Some patients refer themselves to AbleTo, after they hear about it from their health care plan.

The initial consultation determines the problems a patient would like to work on, which might be something pervasive like depression or anxiety, or something specific like difficulties getting along with their coworkers.

The initial consultation also assigns a diagnosis if a clinician decides that is appropriate. Finally, the platform finds out what kinds of treatment the person prefers and determines some mix of automated interventions and personal contact. AbleTo recognizes that physical health and mental health are linked; a problem in one domain inevitably leads to problems in the other.

Computer programmers can think of AbleTo’s interventions as a decision tree. The first decision the program must make is whether to contact a member of the health care plan. Further decisions are taken to set up the initial treatment plan and to decide on the interventions to offer at each stage. Perhaps an anxious patient just needs a yoga session or breathing exercises. Another might need medication or therapy. Computer analytics, often in conjunction with AbleTo’s human experts, make each decision. The process is pretty transparent, so even the patient can look at the plan and track their progress.

AbleTo combines human and automated processing, but in different combinations for different people. Their original offering, now called Therapy 360, relies heavily on human contact (although online). It’s appropriate for very vulnerable patients with complex mental health needs and comorbidities such as diabetes or heart disease. A new and expanded program, Therapy+, shifts more of the emphasis to digital support to help those without medical comorbidities. Another Digital+ program, for mild-to-moderate cases, is almost all automated. However, a trained coach is always available to personalize the program and boost motivation. All these programs employ advances in data analytics and algorithms to adjust their approaches.

I spent some time discussing data and analytics with Shalek and with CEO Trip Hofer. The kinds of issues that behavioral health deals with are subjective and highly personal, so the relevant information tends to be tucked away in unstructured data. The questions AbleTo asks each patient provide somewhat more quantitative data. Still, it’s hard to quantify something like one’s interpersonal performance at work.

I asked Shalek whether AbleTo collects passive data, also called digital phenotyping. Many other programs, for instance, use cell phone data to determine whether a depressed person has been spending more time lying down, from which the programs may infer that the depression has worsened. Shalek said that there’s a lot of promise in this passive data, but that research shows it’s not as accurate as questions collected directly from a patient by common tests such as the Patient Health Questionnaire-9 (PHQ-9) and Generalized Anxiety Disorder Assessment (GAD-7). Thus, AbleTo asks the patient for the information the analytics need, and just tries to make information collection easy.

What informs the platform’s decisions? Like many data-driven healthcare firms, AbleTo combines standard, published clinical research with insights derived from their own analytics. They employ a research team to initialize their recommendation engine from the clinical literature. They update their recommendations continuously based on results.

The eight-week programs include assessments at the beginning, middle, and end of treatment. The 10 percent of patients who don’t achieve their goal after eight weeks can get a follow-up or “booster” program lasting another three or four sessions. At the 3-month and 12-month points after the program ends, AbleTo contacts the patient for follow-up evaluations. All the data collected by the assessments and follow-up interviews feed into the analytics.

The analytics help to overcome one of the major contributions to disparities in health care treatment. Much of the clinical literature draws on subjects who are white, male, middle-class, heterosexual, and younger than most patients. People who don’t fit the demographic aren’t represented as often in the research, so the resulting treatments might not fit them well. Analytics collected from AbleTo’s patient data help to make the recommendations work better across a range of geographies and demographics.

The analytics also help AbleTo tailor the care they deliver based on a patient’s specific needs. This process helps them distinguish between cases that appear similar but actually require different treatments. AbleTo sees these analytics as the future vision of data-driven care.

Take Fred, for example: a patient with Type 2 diabetes who is not managing his chronic illness well and has been identified through his claims data as being at-risk for depression. Patients like Fred can find it harder to manage their chronic health condition when also dealing with a behavioral health concern. They often need the support that comes from Therapy 360, which is specifically tailored to help patients with comorbidities. So AbleTo onboards Fred into the eight-week program, where he learns about depression, how it often manifests in patients with diabetes such as himself, and how to improve his mental and physical health at the same time.

Karen is another patient with a similar profile to Fred. Her profile may look similar when analyzing claims data, but during her initial consultation with an AbleTo therapist, it becomes clear she has a strong diabetic self-management plan in place with an active formal and informal support team. She and her therapist agree that her goals and treatment preference is most likely to be met by a standard course of therapy through the Therapy+ program, focused on depression management with the use of digital tools alongside sessions with a therapist.

Behavioral health is a crucial ingredient of modern health care, which deals with chronic conditions highly influenced by individuals’ behavior. Behavioral health is also a great way to implement value-based care, which all observers say is needed to save the health care system, but which is slow to take off. AbleTo illustrates how a vendor can profit while demonstrating value and taking on risk.

This article is part of the #HealthIT100in100

Listen and subscribe to the Healthcare IT Today Interviews Podcast to hear all the latest insights from experts in healthcare IT.

And for an exclusive look at our top storiessubscribe to our newsletter.

Tell us what you think. Contact us here or on Twitter at @hcitoday. And if you’re interested in advertising with us, check out our various advertising packages and request our media Kit.

About the author

Andy Oram

Andy Oram

Andy Oram writes and edits documents about many aspects of computing, ranging in size from blog postings to full-length books. Topics cover a wide range of computer technologies: data science and machine learning, programming languages, Web performance, Internet of Things, databases, free and open source software, and more. My editorial output at O'Reilly Media included the first books ever published commercially in the United States on Linux, the 2001 title Peer-to-Peer (frequently cited in connection with those technologies), and the 2007 title Beautiful Code. He is a regular correspondent on health IT and health policy for He also contributes to other publications about policy issues related to the Internet and about trends affecting technical innovation and its effects on society. Print publications where his work has appeared include The Economist, Communications of the ACM, Copyright World, the Journal of Information Technology & Politics, Vanguardia Dossier, and Internet Law and Business.