Self-Injurious Thoughts/Behavior

Two measures of SITB were used in this study. The first reflected measurement at a single moment and was procured at the same time as the vocal sample (recorded during the dMSE session). Hence, they were “matched” in temporal resolution. This involved a digital slider to indicate how much participants felt like “harming themselves” on a scale from one to 100. Data were available for 811 of 875 potential data points (i.e., missing for 64 samples). A lower temporal resolution measure of SITB, based on the “suicidality” item from the Brief Psychiatric Rating Scale (65) was also used; this measure of SITB was not “matched” in temporal resolution with the single moment, ambulatory vocal sample. This item involves an ordinal rating of the most severe SITB over the last 2 weeks. Information was based on self-report, medical record and staff. All clinical assessments were conducted by a licensed clinical psychologist (Alex Cohen) and his research team. Ratings and diagnoses reflect a consensus from this team. BPRS scores were collected for all 25 outpatients, and for 47 inpatients. Due to limited interviewing resources and time, we prioritized BPRS ratings for those inpatients that carried SMI diagnoses (i.e., psychosis, mania or depression-spectrum) or were actively symptomatic per medical records or staff report. BPRS data were not collected for 52 inpatients (K = 805 samples). For prediction purposes, the momentary (using a cut-score of 51/100 and above) and 2-week (using a cut-score of 2.1, indicating at least “mild” symptoms) measures were dichotomized. Using these criteria, 201 of 1,958 samples and 105 of 1,698 samples, respectively, met threshold for SITB. The SITB measures did not correlate highly with each other (r = 0.15).