Despite the ever-increasing number of theoretical models of suicide and treatments developed for suicidal behavior, rates of suicide attempts and death by suicide in the United States continue to climb. Research has identified multiple risk factors for suicidal behavior; however, the predictive value of these risk factors is limited, and the majority of individuals with these risk factors do not attempt suicide. This inadequate understanding limits our ability to predict who will engage in suicidal behavior, and when this behavior is likely to occur. One important part of the risk assessment process focuses on both the content and qualitative nature of individuals' speech. However, this process is subjective and particularly problematic as the majority of patients deny suicidal ideation and intent in their last communication before their death by suicide. Recent efforts have been made to objectively monitor and assess speech patterns in clinical populations using computational speech analysis, but this research is limited in important ways. First, studies have primarily focused on the assessment of speech at one time point, limiting predictive utility. Second, speech is typically collected in highly structured and controlled environments, making generalizability difficult. Identifying speech patterns as they occur in the real world, over time, will allow us to identify changes in affect and suicidality that may necessitate intervention. The current study proposes an in-depth examination of speech patterns as novel prospective predictors of suicidal ideation and behavior in a sample of psychiatric inpatients at high risk for future suicidality and healthy controls. Using unobtrusive and secure cell phone based technology, we will record individuals' speech during their day-to-day phone calls. In addition, we will simultaneously collect real-world affect and cognition using ecological momentary assessment technology in order to assess how experiential risk factors temporally relate to speech patterns in this population. The overarching goal of this proposal is to use computational speech analysis to determine the predictive utility of patterns of risk variation derived from unstructured speech on suicide risk during the 6-month period following psychiatric hospitalization. This proposal lays the groundwork for future efforts to monitor large numbers of patients simultaneously, discreetly, and continuously, and to detect changes in their baseline speech patterns that might indicate the need for intervention. The ?Predicting Suicide? study proposes the first investigation of the predictive utility of speech patterns in a high-risk sample of patients admitted to a psychiatric hospital due to significant suicide risk.
Suicide is a major public health problem, with approximately 41,000 deaths by suicide per year. Although risk factors for suicide have been identified, the majority of people who have these characteristics do not go on to commit suicide; therefore, it is important to identify novel processes that underlie suicidal behavior. The present study proposes an in-depth investigation of speech patterns and associated emotional responses as predictors of suicide in a sample of psychiatric inpatients at high risk for future suicidal ideation and behavior and healthy controls.
|Allen, Kenneth J D; Fox, Kathryn R; Schatten, Heather T et al. (2018) Frequency of nonsuicidal self-injury is associated with impulsive decision-making during criticism. Psychiatry Res 271:68-75|