XZEVN: Mental Health and Machine Learning

Xzevn is a startup founded by Robert Williams and Neil Norris.  Robert is a former combat engineer, U.S. Army Drill Instructor, and a veteran with 12 years of military experience. They came to us with a vision to create a web and mobile application that supports mental health.

For this project, the Topflight team built a personalized content recommendation app that helps individuals maintain emotional health and improve their time management, decision making, and goal setting. Users input their daily emotions (journaling) via chatbot conversations, respond to daily questions, and like or dislike recommended articles and famous quotes.

• Summary:
  Total Cost: 2,000 hours | Length: 1 year 8 months | Type: Prototype, Dev | MVP v1: 600 hours, 5 months
• Platform:
  Web, iOS, Android
• Technology Stack
  MySQL, Laravel, ReactJS, PWA, NLTK, TensorFlow, RESTful services with Flask

An App To Help Users Create Better Personal Experiences

The XZEVN concept is an application that gives users the type of support that comes from confession, therapy, or the guidance of a school counselor. They use voice or text inputs to express their concerns, frustrations, problems, and issues.

In return, the system applies machine learning and artificial intelligence (AI) tools to understand and respond to their needs. To succeed, it has to react like a human listener, provide an environment that is safe and non-judgmental, and deliver useful guidance to the user.

Design and Validation

We had to design an engaging user experience that provided users immediate feedback in order to incentivize them to respond to daily questions.  We also needed to design a chatbot that doubles as a stream of suggested content and an outlet for live conversation. See screenshots of our design at https://www.behance.net/gallery/56644913/AVA-Screenshots

Customer Discovery, Product Market Fit and Validation were conducted through participation in NSF I-Corps (George Washington University) and IGNITE Pilot Program with NSA and The Maryland Center at Bowie State University with additional research on behavior at University of North Carolina School of Social Work.

Related: The Complete Guide to Designing a Mobile App

Development

The solution that we developed matches a progressive web app with a backend that uses artificial intelligence to learn user preferences and suggest helpful reading materials. The app delivers articles and famous quotes, based on its predictions about what will produce positive changes.

The users journal about their daily mental state as honestly as they can, answer questions, read and rate the suggested content, and express their responses as likes and dislikes.

The recommendation system we built integrates two types of machine learning platforms (a K Nearest Neighbors algorithm and Restricted Boltzmann Machine). It learns about users and content in real time to improve the accuracy of recommendations, classifications, and predictions.

Related: How to Create a Mental Health App

Future Improvements

Learning goes both ways. As the AI system learns more from users, it makes better recommendations, which helps the user make better personal choices and experience better outcomes.

To make the system faster and classify content better, we are currently evaluating DoD technology through a Patent License Agreement and a Cooperative Research and Development Agreement with the National Security Agency Office of Technology Transfer to further enhance the capabilities of XZEVN.

We also plan to ask even better questions, including after users read an article, which could also serve as a feature to measure engagement and comprehension.