The Most Important Feature for Software-Based Digital Health Products
(Hint: It's not "AI")
An Important Question
“What’s the most important feature in a software digital health product?” someone asked me when I first joined Verily Life Sciences.
What would you say?
At the time, I gave some answer along the lines of “passive sensing + continuous data streams + machine learning.” All good things, no doubt, and areas where I think it’s important to continue to make progress.
But now, I’d agree with Doge:
What’s the Deal with Behavior Change?
Most digital health products are primarily software, and therefore have limited means to directly modify health (which is mostly the result of human biology interacting with the environment).
But what software can do, and has been shown to be very effective at doing, is modifying human behavior by helping people perceive and think in new ways.
For example, take Twitter - Twitter allows anyone with an internet connection to communicate news and interesting thoughts directly with anyone else in the world, at any time. In doing so, it has changed (for good or bad) the way that many humans interact with each other, resulting in the average American spending 144 minutes a day on social media.
However, despite the proliferation of digital health apps and lots of investment, we don’t see nearly that level of engagement with digital health apps. Why is that?
I’m going to argue that it is because we need to go beyond the “shallow” design patterns that mostly come from the world of “attention” and “entertainment” app economies and move towards deeper design patterns focused on understanding upstream psychological drivers of behavior.
It means that if you want to increase metrics like “user engagement” with your application, you’re not going to be able to A/B test your way to success — instead, you need to do the much, much harder work of trying to understand each individual user and what they are hoping to gain from your app.
Models of Behavior Change
There are many academic models of behavior change that exist. Out of the ones I’ve seen, the model that I find most personally compelling is the Transtheoretical Model (also known as the Stages of Change Model).
Specifically, 10 processes have been identified that are key to moving a person through the cycle (re-posted below from the original article):
Consciousness Raising - Increasing awareness about the healthy behavior.
Dramatic Relief - Emotional arousal about the health behavior, whether positive or negative arousal.
Self-Reevaluation - Self reappraisal to realize the healthy behavior is part of who they want to be.
Environmental Reevaluation - Social reappraisal to realize how their unhealthy behavior affects others.
Social Liberation - Environmental opportunities that exist to show society is supportive of the healthy behavior.
Self-Liberation - Commitment to change behavior based on the belief that achievement of the healthy behavior is possible.
Helping Relationships - Finding supportive relationships that encourage the desired change.
Counter-Conditioning - Substituting healthy behaviors and thoughts for unhealthy behaviors and thoughts.
Reinforcement Management - Rewarding the positive behavior and reducing the rewards that come from negative behavior.
Stimulus Control - Re-engineering the environment to have reminders and cues that support and encourage the healthy behavior and remove those that encourage the unhealthy behavior.
Most digital health apps I’ve seen have traditionally focused on the bolded items above (1, 8, 9, and 10), but don’t address the remaining items that particularly focus on a person’s personal goals/values and interaction with social norms. This is a missed opportunity.
CBT as a Guide
I think it’s especially helpful to look to Cognitive-Based Therapy (CBT) as a model for how to enact behavior change. CBT is a psychotherapy approach that was developed in the 1960s for the treatment of anxiety and depression, and has been one of the most validated approaches for changing behavior in clinical psychology.
At the core of CBT is a relationship with a therapist who helps the patient identify the goals that are important to them, and supports them in removing obstacles that get in the way. Notice that the therapy starts with the person’s goals, and is tailored to those goals.
Second, the therapist in CBT focuses on using Socratic Dialogue to help the patient deeply engage in “Self-reevaluation” and feeling like they have a “Helping” relationship.
This is an existing gap in digital health solutions - they often try too much to be one-size-fits-all and focus on making everyone align to a pre-defined template, instead of tailoring the treatment to the patient’s goals. Digital health companies that are able to measure and effect change upstream in the process are going to have a bigger impact on behavior change, and ultimately, health and business outcomes.
One key place in most digital health product flows where this principle can apply is during onboarding. Most digital health applications would benefit from expanding the onboarding process to include questions about a patient’s goals, and then tailoring a custom treatment plan to those goals (either manually or, in the ideal future world, automatically).
What’s Next?
In order for digital health to arrive at the next frontier of impact, the conversation needs to focus less on applications of technology like machine learning and sensors and more on a fundamental understanding of human behavior and what drives changes in behavior.
I’m still learning and practicing in this area, and I’d encourage you to consider it more as well. Here are a list of resources that I’ve found particularly helpful in my journey:
What do you think is the most important feature of digital health products?
As always, if you’ve enjoyed the content, you can subscribe below - I publish ~1x weekly, and no spam - ever: