Collaborative Research: SCH: Using Multi-Stage Learning to Prioritize Mental Health Risk Using Evidence from Speech and Text

点击翻译按钮获取中文摘要

According to the World Health Organization and the Global Burden of Disease 2010 studies, mentalhealth issues are a top contributor to global disease and a leading cause of disability worldwide. It is anenormous personal and societal toll. Mental illness is a common precursor to suicide, and suicidality isthe second leading cause of death in youth and young adults between 10 and 34 years of age. Ineconomic terms, mental illness exceeds cardiovascular diseases in the projected 2011-2030 cost ofnoncommunicable diseases (USD16.3T worldwide). Complicating this picture further is the fact thatmental healthcare is desperately resource-limited, and clinicians treating people for mental healthproblems operate in a vacuum between visits. This project proposes a fundamental shift in how machinelearning is used to approach the problem of mental health detection and monitoring, with a technologicalinvestigation that brings together speech analysis, language analysis, and machine learning research,informed by deep clinical experience and expertise and fueled by ethically collected data. A tiered multiarmedbandit framework will be used to provide a highly flexible way to evaluate multiple kinds ofevidence in settings where there can be diverse methods for assessment that vary in cost and the value ofthe information they provide. As such, it is an excellent fit for the real-world problem of mental healthassessment in resource-limited settings. Investigations will include simulations of patient monitoringbetween clinical visits that will be informed by realistic, real-world assumptions and team members'clinical experience treating patients with schizophrenia, depression, and risk of suicide.At the core of this project's technical approach is the recognition that the “multi-armed bandit” problem inmachine learning is a good fit for the real-world scenario that mental health providers face whenmonitoring a population of patients in treatment: what is the best way to allocate limited resources amongcompeting choices, given only limited information? This project develops a tiered multi-armed banditformulation, where a succession of stages is applied to a population of patients in order to best allocatedifferent types of resources, each with different per-patient impact but also cost. Conceptually, tieredapproaches are familiar in current medical practice. For example, patient contact typically progressesfrom a receptionist, to a nurse or intake coordinator, perhaps to a certified nurse practitioner, to a primarycare doctor, ultimately to a specialist---each step involving corresponding increases in both the cost of theprofessional involved and their degree of expertise. The tiered multi-armed bandit model developed bythis award includes concerns of stochastic and adverse selection, where patients at one tier do not proceeddeterministically to the next, even when explicitly selected. It also incorporates complex (e.g., non-linearsuch as monotone submodular) objective functions that better capture within-cohort interactions. Onecore strength of the tiered model is that it provides a flexible way to incorporate multiple kinds ofevaluative evidence in settings where there can be diverse methods for assessment that vary in cost andthe value of the information they provide. Toward that end, this project also includes both text analysisand speech analysis components that make use of ethically collected language and speech data andclinically validated assessments of mental condition. Techniques developed under this award, whiledirectly motivated by and tested in the mental health setting, will be useful in other settings in bothhealthcare as well as other settings where a "prioritization funnel" is in play, including talent sourcing andcustomer acquisition.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.