Improving Digital Mental Health Help-Seeking with Mobile Apps and Machine Learning

Nature of the Problem

In 2018, four years after the first Mental Health America report, youths experiencing major depressive episodes increased from 8.66% to 12.63% and 61.5% did not receive treatment. Awareness of mental health is increasing, and stigma has been declining as of recent, but despite increases in the availability and supply of digital and in-person interventions, young people are particularly resistant to seeking professional help due to stigma and embarrassment. Researchers have identified that young people have a strong desire for autonomy, believing they should solve problems for themselves which contributes to a complicated blend of stigma, insurance, awareness, cultural, and social barriers prevent people from seeking care. Mental health help-seeking is different from physical health because it is largely influenced by psychosocial factors like stigma, problem recognition, and preferences for self-reliance and needs improved engagement models to help young adults navigate help-seeking behaviors.

Program Description

We are expanding mental health web and app software as a service that is a front-end digital engagement tool to make it easier for people considering mental health help to find anonymous support 24/7 and get connected to services quicker and easier. By enhancing problem recognition, decisions to seek care, and service selection that is informed by machine learning in the application, the apps will deliver highly-personalized engagement strategies and recommendations to users to reduce incidences of un- or under-treated symptoms of depression. This stand-alone plug-and-play software platform can be implemented at a school, organization, or college and overnight increase their populations access to anonymous and supportive peers and identifies potentially at-risk triggers. The goal is to use machine learning and app user behavior to encourage help-seeking behaviors to customized and preferred resources unique to each organization.

Disruption

The proposed program not only impacts supply of available support services for a client, but also impacts demand for services by influencing help-seeking behaviors. That allows us to partner with many other digital and in-person services and interventions to find the most optimal resource for users. Marblesapp has captured over 22,000 messages shared on the apps and over 120,000 unique data points gathered on our users. Using this unique data set, we will train the patented neural network and incorporate it into the application. This neural network software allows for rapid retraining and test without having to rewrite any code and establishes new practices in data security by keeping valuable user data local to a users device. This will create neural model for identification of at-risk symptoms for depression and can quickly be scaled to other symptoms of other disease states like anxiety, PTSD, and bi-polar disorder.

Leadership

This technology focuses on improvements to the digital mental health help-seeking. It can provide leadership into the broader behavioral health community by gaining greater understandings of how to identify triggers and signals of at-risk behavior from naturalistic data sets like messaging apps or posts on social media. It can also lead to understandings of how to engage an individual with text-based responses or push notifications to influence the likelihood of participation with a recommended resource. Ultimately, it will lead to engagement efforts with much higher referral rates that might be a model for how to engage traditionally underserved populations that may have barriers to accessing care.

Scalability

We are building this product to be a stand-alone software as a service that a client could purchase, customize, and implement overnight with their population. Since it is a digital technology consisting of a web app Administrative portal and the iOS and android apps, scaling can rapid and expansive. We also have an extensive network of potential pilot sites at Universities, colleges, employers, and practitioners/interventions that are interested in collaborating and participating. Founder Adam is speaking at several conferences, high schools, and organizations this year and their research team at the University of Minnesota is currently working on 2 publications about the potential disparity-reducing impacts of the application. These efforts will add additional visibility, credibility, and leads for pilot sites.

Results/Outcomes

Marbles Analytics can help close treatment gaps for hard to reach populations and reduce medical expenditures and adverse event associated with depression such as decreased productivity and turnover or dropouts. Behavioral health interventions have been show to lead to decreases in medical expenditures so by getting the right people to the right care providers quicker, we hope to decrease the adverse effects of underserved depression. Broader goals involve reducing young adult suicide, school shootings, and feelings of isolation and loneliness.