1 minute read

Building a data driven Mental Health Application that using Node.js on Azure

Lee Janzen; Christopher Roy; Christoph Bendix; and Drew King

Abstract

Current mental health apps fail in several manners:

  • User retention, 80% of users uninstall after first month
  • Large number of participants = good
  • Attrition -> challenging

Goal is to build a Mental health app design that increases retention and has meaningful results:

  • Target audience: users experiencing depression, anxiety, work/school stress
  • Designed using research on app retention
  • Can identify emotional state of user
  • Offers activities proven to reduce stress
  • Issue: varying needs of users Potential solution: use machine learning algorithms to identify emotion states
  • Combine results across procedures
  • Allows enhanced flexibility
  • Issue: no formal guidelines for combining certain procedures Goal: establish guidelines around ensemble procedures Potential impact in mental health research and real world impact on app users.
Methodology of this research will be available upon app publication, see more details at solala.app

Results

Solala is an app designed by students with the intention of improving the mental health of burned out professionals and students.

  • Built with universal design and accessibility principles in mind
  • UX Design process considers the needs of colorblind individuals
  • Additionally the design process considers user retention through gamification Users will care for a plant that grows with them
  • The plant is intended to motivate users to take care of themselves and complete growth tasks
  • Plant encourages users to drink water, meditate, and take breaks This work is backed by research into wellness and mental health for individuals under high stress.
Solala Repository

Next steps: Conduct further research from consenting users including:

  • User retention for Solala based on reward intervals and task intervals
  • Testing other algorithms that process emotion state data.
  • Building this experiment in mobile native languages such as Darwin or Swift for better performance.