An Evidence-Based App to Help MS Patients with Depression
Pear Therapeutics is known for producing the first smartphone app to be FDA-approved for the treating of medical conditions. They received FDA approval for doctor-prescribed apps to treat substance abuse disorder and opioid use disorder, and are on track to producing clinical applications that can help treat a number of other conditions.
Echobind leveraged several of their senior engineers to help the Pear Therapeutics team research, develop and publish their app, simply named “Pear 006”, whose goal is to treat depression in Multiple Sclerosis patients.
The App’s Purpose
Depression is a common symptom for multiple sclerosis patients, who have their own set of concerns, worries and biological symptoms. The depression symptoms associated with MS are known to be very manageable with the help of Cognitive Behavioral Therapy (CBT), a widely-used treatment methodology in which patients make note of their automatic negative thoughts, and replace them with more constructive ones.
The Pear 006 app took the form of a chatbot that would help a patient form the habit of assessing negative thoughts and turning them into more constructive ones. Each iteration of the app was validated with patients to ensure it was going to be maximally helpful to them.
Our process with the Pear team involved a regular assessment of goals for the company end user, combined with an assessment of what features could be developed to satisfy those needs and how. For the CBT chatbot, we did a comparative analysis of existing chatbot frameworks including Microsoft Bot Framework, Rasa, Google’s DialogFlow and others, carefully weighing out the pro and cons of each through the lens of client concerns, and making recommendations.
Ultimately we discovered that the AI portions of this future chatbot were less immediately important than the content management; at least initially, the Pear team’s mental health experts would be diligently building and refining content that was perfectly suited to the needs of MS patients with depression. The client’s responses to the chatbot’s messages didn’t need to be open-ended for the initial phases, but rather multiple-choice responses. In addition, the Pear team needed to reduce DevOps overhead (which it would incur by having to engage in new ways to host chatbot resources), so using an external tool (on a new platform like Microsoft Azure, for example) would add additional overhead.
In the end, the right solution was different than expected: the Echobind team built a custom chatbot “brain” that was contained within the app, which would direct the patient through different chat threads based on their input. And every conversational thread would be completely administered by the Pear 006 expert team via a very complex but stable spreadsheet interface. It was built in such a way that the brain could be easily replaced with a different one when the need arose.
In short, Pear was able to continually refine and improve the way the “bot” would help patients while requiring minimal developer interaction.