Automated Mental Health (AMH) to Predict Inward and Outward Destructive Behaviors along with Mental Health
Abstract
Approved for Public ReleaseThis proposal will research and engineer the components needed for an automated mental health (AMH) system that can predict outwardly destructive behavior (oDB), using advances in computational behavior and artificial intelligence/machine learning (AI/ML) that are based on validated neuroscience. This research will test how well variables that quantify individual reward/aversion judgments can be used to predict past and current oDB, along with inwardly destructive behavior (iDB) and mental health (MH) problems. A recently discovered engineering-based framework for reward/aversion judgment, relative preference theory (RPT), will be used to predict oDB/iDB/MH. Behaviors classified as oDB include harassment, disruptive behavior, emotional, verbal & physicalviolence/aggression, whereas iDB includes addiction and suicidal thought and behavior. MH will focus on depression, anxiety and post-traumatic stress disorder; these disorders can be co-morbid with, and worsen the risk for, oDB/iDB. oDB, iDB, and MH can disrupt mission readiness and execution, and post-mission assessment, which are now recognized as significant challenges for the DoD and US Navy. Currently no system exists that can rapidly and automatically assess the likelihood of oDB/iDB and discriminate them from the many potential co-morbid MH conditions. Tools to assess oDB/iDB and MH are needed for one-time assessments, for longitudinal risk assessments, and for assessing the rapidity for which intervention is needed. Such tools are needed in response to the increasing incidence of oDB/iDB and MH problems within the Navy and other military sectors over the past decades, and reports of increasing first responder oDB/iDB and MH concerns. The technical approach proposed for this problem follows a novel path suggested by the AI/ML literature that has not been implemented by existing frameworks using: (i) text mining, (ii) big data scale surveys (including ecological momentary assessments), or (iii) smartphone/wearable sensors. This new approach will implement a small set of variables, derived from RPT, as inputs for data-driven AI/ML prediction of oDB/iDB/MH. This approach of using a small variable set for AI/ML is distinct from traditional big data approaches that use many uninterpretable variables. This new approach uses cutting-edge research from researchers at University of Cincinnati, Northwestern, and Mass General Hospital/Harvard who found the mathematics within relative preference theory (1) meets Feynman criteria for lawfulness and (2) can predict iDB and MH as one of the first demonstrations of explainable artificial intelligence (xAI). RPT integrates reward/aversion mathematics from two neuroscience frameworks and two Nobel Prizes in Economics. This work will study how well RPT variables predict the full set of oDB, iDB and MH problems. It will also test which natural language processing (NLP) variables accurately produce RPT preference curves and similar RPT variables. It will further quantify the accuracy of these NLP-based RPT variables to predict oDB vs. iDB and MH. Lastly, this work will test the integration and scalability of these functions as the first instantiation of AMH, one that could run in real time as a form of xAI. Upon completion, this work will produce both (1) the necessary research for an AMH and (2) a working prototype for follow-up study. Necessary proof of concept work has been done to show viability for RPT-based prediction of iDB and MH giving the research proposed herein a high probability of success. Furthermore, the lead investigators for this team are two internationally recognized researchers, one a pioneer in AI, and the other a pioneer in reward/aversion neuroscience who have a long history of success with government funded research.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 15, 2023
- Source ID
- N000142312396
Entities
People
- Hans Breiter
Organizations
- Office of Naval Research
- United States Navy
- University of Cincinnati