iPonder: Tackling Teen Mental Health Stigma via Deep Learning-Based Diagnosis and Anonymous Therapy

AJAS · 2022 Behavioral Science

Vivek Kogilathota

Thumbnail supplied by the source for iPonder: Tackling Teen Mental Health Stigma via Deep Learning-Based Diagnosis and Anonymous Therapy

Overview

Suicide is the second leading cause of death for teenagers. Depressed teens feel obligated to deal with these serious issues by themselves to avoid burdening those around them and keep quiet due to shame and stigma. This, a solution for teens to anonymously be evaluated and given peer-based therapy from similar individuals was proposed. The researcher hypothesized that if text-based and audiovisual-based deep learning neural networks were trained on substantial mental health datasets over various modes of data, then these neural networks would be able to remotely predict mental health disorder tendencies in teens with accuracy, precision, and recall of over 95% and 80% per network type respectively. Thus, iPonder, a cross-platform accessible mobile application for teenagers to privately overcome their issues, was created. Using a user-submitted video and responses to behavioral questionnaires, iPonder constructs a hyper-personalized user profile, which is used to pair teenagers anonymously online with similar profiles to discuss their issues. Three artificial neural networks were constructed to predict user risk of depression, anxiety, and suicide watch respectively based on their free-text responses, using transfer learning from a token-based text embedding and Reddit’s Self-Reported Mental Health Diagnosis Corpus. Labeled time-based video, audio, and text data from the Distress Analysis Interview Corpus was used to train three long short-term memory neural networks. These models predict the risk of mental health issues from a user-submitted video via encoded facial action units and speech analysis. iPonder achieves approximately 98% accuracy, recall, and precision per text-based neural network and over 82% for long short-term memory networks. It successfully integrates these models into the mobile application. Anonymous peer-based therapy and discussion is guided by application chats with cognitive behavioral therapy prompts. iPonder provides the solution deeply desired by teenagers, pointing to novel, accessible, and effective mental health therapy for teens.

My Story

In August 2020, feeling an unprecedented tension on social media due to the sharp increase in negative teen sentiment, I embarked on a journey to create iPonder, a multimodal approach to anonymous teen mental health diagnosis and therapy via machine learning. Teens with mental health issues face the intense social stigma of seeing a counselor, as their feelings are often downplayed by those who don't understand them. So, after seven months of diligent research and programming, I finished the iPonder mobile application, which uses user-submitted textual, visual, and aural data to predict the user's risk of depression, anxiety, and suicide via neural networks. After continuously optimizing these networks, I obtained an accuracy of over 98% for each. Using these risk predictions, iPonder matches similar users online to anonymously discuss their mental health issues with someone who sympathizes with them, gradually easing them into comfort with professional therapy.

In March 2021, I won a national honorable mention in the ExploraVision Research Competition for my potential to create significant social change through a new innovative technique. Simultaneously, intending to eventually implement iPonder in the real world, I studied the entrepreneurial facets of my research. Submitting a five-minute business proposal pitch in February 2021, I became an international top ten finalist in the Blue Ocean Entrepreneurship Competition in May 2021 out of approximately 500 pitches worldwide. Achieving competitive validation, I deeply desired to contribute my work back to the community of dedicated researchers exploring the intersection of mental health and machine learning. So, I submitted my research paper to the Institute of Electrical and Electronics Engineers (IEEE) International Conference on Artificial Intelligence, Robotics, and Communication. In June 2021, I was invited to orally present my research and network with other researchers at the virtual conference, becoming the youngest presenter in the conference's history. Recently, in September 2021, my paper was published under conference proceedings, allowing others to explore ways to improve my research.

After conducting additional research with iPonder this previous summer, experimenting with modern variations of neural networks, and exploring more conversational therapy techniques, I submitted my revised abstract to the Texas Junior Academy of Science Competition. In October 2021, I was invited to present my research with other brilliant student researchers at Texas A&M University. I won first prize and was inducted as an AJAS Fellow to present my research at the American Junior Academy of Science Conference held in conjunction with the American Association for the Advancement of Science Conference. In January 2022, I was named a Regeneron Science Talent Search Top 300 Scholar for my research.