Trial Outcomes & Findings for Study of a PST-Trained Voice-Enabled Artificial Intelligence Counselor (SPEAC) for Adults With Emotional Distress (Phase 1) (NCT NCT04524104)

Study of a PST-Trained Voice-Enabled Artificial Intelligence Counselor (SPEAC) for Adults With Emotional Distress (Phase 1)

Two neural targets defined a priori, specifically activation of the amygdala for nonconscious threat-related emotional reactivity and activation of the DLPFC for cognitive control will be assessed using fMRI. In the facial emotion viewing paradigm, facial expression stimuli are standardized black and white photographs of 8 identities (4 female, 4 male) with evoked expressions of threat-related emotions (fear, anger), sad-related emotions (sadness) and reward-related emotions (happiness), along with neutral. To assess amygdala activation for the negative affect circuit, our analysis focused on threatening faces only. Threat stimuli included a combination of fear and anger stimuli relative to neutral blocks. For the Go-NoGo paradigm, the 'Go' and 'NoGo' stimuli are presented for 500 ms each with an inter-stimulus interval of 750 ms. Higher score indicates higher activation of the amygdala and DLPFC.

Two neural targets defined a priori, specifically activation of the amygdala for nonconscious threat-related emotional reactivity and activation of the DLPFC for cognitive control will be assessed using fMRI. In the facial emotion viewing paradigm, facial expression stimuli are standardized black and white photographs of 8 identities (4 female, 4 male) with evoked expressions of threat-related emotions (fear, anger), sad-related emotions (sadness) and reward-related emotions (happiness), along with neutral. To assess amygdala activation for the negative affect circuit, our analysis focused on threatening faces only. Threat stimuli included a combination of fear and anger stimuli relative to neutral blocks. For the Go-NoGo paradigm, the 'Go' and 'NoGo' stimuli are presented for 500 ms each with an inter-stimulus interval of 750 ms. Higher score indicates higher activation of the amygdala and DLPFC.

Two neural targets defined a priori, specifically activation of the amygdala for nonconscious threat-related emotional reactivity and activation of the DLPFC for cognitive control will be assessed using fMRI. In the facial emotion viewing paradigm, facial expression stimuli are standardized black and white photographs of 8 identities (4 female, 4 male) with evoked expressions of threat-related emotions (fear, anger), sad-related emotions (sadness) and reward-related emotions (happiness), along with neutral. To assess amygdala activation for the negative affect circuit, our analysis focused on threatening faces only. Threat stimuli included a combination of fear and anger stimuli relative to neutral blocks. For the Go-NoGo paradigm, the 'Go' and 'NoGo' stimuli are presented for 500 ms each with an inter-stimulus interval of 750 ms. Higher score indicates higher activation of the amygdala and DLPFC.

Two neural targets defined a priori, specifically activation of the amygdala for nonconscious threat-related emotional reactivity and activation of the DLPFC for cognitive control will be assessed using fMRI. In the facial emotion viewing paradigm, facial expression stimuli are standardized black and white photographs of 8 identities (4 female, 4 male) with evoked expressions of threat-related emotions (fear, anger), sad-related emotions (sadness) and reward-related emotions (happiness), along with neutral. To assess amygdala activation for the negative affect circuit, our analysis focused on threatening faces only. Threat stimuli included a combination of fear and anger stimuli relative to neutral blocks. For the Go-NoGo paradigm, the 'Go' and 'NoGo' stimuli are presented for 500 ms each with an inter-stimulus interval of 750 ms. Higher score indicates higher activation of the amygdala and DLPFC.

HADS, including 7 questions for anxiety and 7 questions for depression, measures self-reported anxiety and depression in a general medical population of patients. Each item on the questionnaire is scored from 0-3 and the score for either anxiety or depression ranges between 0 and 21 with the following categories: normal 0-7, mild 8-10, moderate 11-14, and severe 15-21. Total score for the entire scale (emotional distress) ranges from 0 to 42, with higher score indicating more distress. Cronbach's alpha coefficients range from 0.78-0.93 for the anxiety subscale and from 0.82-0.90 for the depression subscale. Test-retest correlations are ≥0.80 after ≤2 weeks. Correlations of the anxiety and depression subscales with commonly used anxiety and depression measures (e.g., Beck Depression Inventory, Patient Health Questionnaire, State-Trait Anxiety Inventory, Symptom Checklist-90-Revised) vary between 0.60 (good) and 0.80 (very good).

HADS, including 7 questions for anxiety and 7 questions for depression, measures self-reported anxiety and depression in a general medical population of patients. Each item on the questionnaire is scored from 0-3 and the score for either anxiety or depression ranges between 0 and 21 with the following categories: normal 0-7, mild 8-10, moderate 11-14, and severe 15-21. Total score for the entire scale (emotional distress) ranges from 0 to 42, with higher score indicating more distress. Cronbach's alpha coefficients range from 0.78-0.93 for the anxiety subscale and from 0.82-0.90 for the depression subscale. Test-retest correlations are ≥0.80 after ≤2 weeks. Correlations of the anxiety and depression subscales with commonly used anxiety and depression measures (e.g., Beck Depression Inventory, Patient Health Questionnaire, State-Trait Anxiety Inventory, Symptom Checklist-90-Revised) vary between 0.60 (good) and 0.80 (very good).

HADS, including 7 questions for anxiety and 7 questions for depression, measures self-reported anxiety and depression in a general medical population of patients. Each item on the questionnaire is scored from 0-3 and the score for either anxiety or depression ranges between 0 and 21 with the following categories: normal 0-7, mild 8-10, moderate 11-14, and severe 15-21. Total score for the entire scale (emotional distress) ranges from 0 to 42, with higher score indicating more distress. Cronbach's alpha coefficients range from 0.78-0.93 for the anxiety subscale and from 0.82-0.90 for the depression subscale. Test-retest correlations are ≥0.80 after ≤2 weeks. Correlations of the anxiety and depression subscales with commonly used anxiety and depression measures (e.g., Beck Depression Inventory, Patient Health Questionnaire, State-Trait Anxiety Inventory, Symptom Checklist-90-Revised) vary between 0.60 (good) and 0.80 (very good).

Participants' problem solving abilities will be assessed using the reliable and valid SPSI-R:S that contains 25 items in the following 5 subscales: positive problem orientation (PPO), negative problem orientation (NPO), rational problem solving (RPS), impulsive/careless style (ICS), and avoidance style (AS). Each item is rated on a 5-point scale ranging from "not at all true of me" (0) to "extremely true of me" (4). The SPSI-R:S assessed total problem-solving ability. The total problem-solving ability score ranged from 0 to 20 by averaging the subscale scores, with the higher score indicating more productive problem-solving skills.

DAS (Form A) is a 40-item self-reported scale that measures the presence and intensity of dysfunctional attitudes. Each item is rated a 7-point Likert scale (7 = fully agree; 1 = fully disagree). DAS score is the sum of the 40 items, with a range of 40-280. Higher score indicates the more dysfunctional attitudes an individual possesses.

PSWQ is a self-reported, 16-item, Likert-type scale that measures the trait of worry. PSWQ score has a range of 16-80, with a higher total score indicating more worry.

PANAS consists of two 10-item self-reported scales to measure positive and negative affect. Each item asks about the extent one has felt a positive or negative feeling on a 5-point scale of 1 (not at all) to 5 (very much). Positive Affect scores can range from 10 - 50, with higher scores representing higher levels of positive affect.

PANAS consists of two 10-item self-reported scales to measure positive and negative affect. Each item asks about the extent one has felt a positive or negative feeling on a 5-point scale of 1 (not at all) to 5 (very much). Negative Affect scores can range from 10 - 50, with lower scores representing lower levels of negative affect.

The Sheehan Disability Scale is a validated questionnaire that measures functional disability and is sensitive to treatment effects in clinical trials. Cronbach's alpha is 0.89. Patients rate the extent to which their symptoms impair work/school, social, and family life on a visual analog scale from 0 (Not at all) to 10 (Extremely) and answer the number of days when their symptoms cause them to miss work/school and be unproductive at work/school. Total score is the sum of the three separate scores on work/school, social, and family life, ranging from 0 to 30.

Work productivity and activity impairment questionnaire (WPAI) consists of 6 questions and measures impairments in work and activities. Four main outcomes can be calculated and expressed in percentages by multiplying the scores by 100: 1) percent work time missed due to health = Q2/(Q2 + Q4) for people who were currently employed; 2) percent impairment while working due to health = Q5/10 for people who were currently employed and actually worked in the past 7 days; 3) percent overall work impairment due to health Q2/(Q2 + Q4) + ((1 - Q2/(Q2 + Q4)) × (Q5/10)) for people who were currently employed; and 4) percent activity impairment due to health Q6/10 for all respondents.

Work productivity and activity impairment questionnaire (WPAI) consists of 6 questions and measures impairments in work and activities. Four main outcomes can be calculated and expressed in percentages by multiplying the scores by 100: 1) percent work time missed due to health = Q2/(Q2 + Q4) for people who were currently employed; 2) percent impairment while working due to health = Q5/10 for people who were currently employed and actually worked in the past 7 days; 3) percent overall work impairment due to health Q2/(Q2 + Q4) + ((1 - Q2/(Q2 + Q4)) × (Q5/10)) for people who were currently employed; and 4) percent activity impairment due to health Q6/10 for all respondents.

Work productivity and activity impairment questionnaire (WPAI) consists of 6 questions and measures impairments in work and activities. Four main outcomes can be calculated and expressed in percentages by multiplying the scores by 100: 1) percent work time missed due to health = Q2/(Q2 + Q4) for people who were currently employed; 2) percent impairment while working due to health = Q5/10 for people who were currently employed and actually worked in the past 7 days; 3) percent overall work impairment due to health Q2/(Q2 + Q4) + ((1 - Q2/(Q2 + Q4)) × (Q5/10)) for people who were currently employed; and 4) percent activity impairment due to health Q6/10 for all respondents.

Work productivity and activity impairment questionnaire (WPAI) consists of 6 questions and measures impairments in work and activities. Four main outcomes can be calculated and expressed in percentages by multiplying the scores by 100: 1) percent work time missed due to health = Q2/(Q2 + Q4) for people who were currently employed; 2) percent impairment while working due to health = Q5/10 for people who were currently employed and actually worked in the past 7 days; 3) percent overall work impairment due to health Q2/(Q2 + Q4) + ((1 - Q2/(Q2 + Q4)) × (Q5/10)) for people who were currently employed; and 4) percent activity impairment due to health Q6/10 for all respondents.

The SF-12 is a 12-item version of the SF-36 that measures overall health-related quality of life. Physical and mental health composite scores are computed using the scores of 12 questions and range from 0 to 100, with 0 indicating the lowest level of health and 100 indicating the highest level of health. Test-retest correlation is 0.89 for the physical health subscale and 0.76 for the mental health subscale.

The SF-12 is a 12-item version of the SF-36 that measures overall health-related quality of life. Physical and mental health composite scores are computed using the scores of 12 questions and range from 0 to 100, with 0 indicating the lowest level of health and 100 indicating the highest level of health. Test-retest correlation is 0.89 for the physical health subscale and 0.76 for the mental health subscale.

Daily end-of-day assessments will be captured over a one-week period at baseline and every 2 weeks (i.e., on weeks 2, 4, 6, 8, 10, 12, and 16). During active treatment in the Lumen arm (Studies 1 \& 2) or the in-person PST arm (Study 2), this will occur for 3 days prior to, the day of, and for 3 days after a scheduled PST session. For waitlist controls, this will occur for 7 days starting on Sunday of each assigned week. Daily mood will be measured using 8 mood items (0 = Not at all, 6 = Extremely), with 4 measuring positive affect and 4 measuring negative affect. Positive affect score is mean of item scores, with higher scores indicating more positive affect. Negative affect score is mean of item scores, with higher scores indicating more negative affect.

Daily end-of-day assessments will be captured over a one-week period at baseline and every 2 weeks (i.e., on weeks 2, 4, 6, 8, 10, 12, and 16). During active treatment in the Lumen arm (Studies 1 \& 2) or the in-person PST arm (Study 2), this will occur for 3 days prior to, the day of, and for 3 days after a scheduled PST session. For waitlist controls, this will occur for 7 days starting on Sunday of each assigned week. Daily mood will be measured using 8 mood items (0 = Not at all, 6 = Extremely), with 4 measuring positive affect and 4 measuring negative affect. Positive affect score is mean of item scores, with higher scores indicating more positive affect. Negative affect score is mean of item scores, with higher scores indicating more negative affect.

Daily end-of-day assessments will be captured over a one-week period at baseline and every 2 weeks (i.e., on weeks 2, 4, 6, 8, 10, 12, and 16). During active treatment in the Lumen arm (Studies 1 \& 2) or the in-person PST arm (Study 2), this will occur for 3 days prior to, the day of, and for 3 days after a scheduled PST session. For waitlist controls, this will occur for 7 days starting on Sunday of each assigned week. Daily stress will be assessed by the question "how stressful was your day today" (0 = Not at all, 6 = Extremely).

Daily end-of-day assessments will be captured over a one-week period at baseline and every 2 weeks (i.e., on weeks 2, 4, 6, 8, 10, 12, and 16). During active treatment in the Lumen arm (Studies 1 \& 2) or the in-person PST arm (Study 2), this will occur for 3 days prior to, the day of, and for 3 days after a scheduled PST session. For waitlist controls, this will occur for 7 days starting on Sunday of each assigned week. The appraisal items will focus on assessing "problem orientation" or the general awareness and appraisals of problems during the day. A main appraisal question asks "To what degree were you able to accept your problem as normal or an expected part of everyday life?" (0 = Not at all, 6 = A great deal).

PHQ-9 is a self-administered instrument for screening, diagnosing, monitoring, and measuring the severity of depression. It rates the frequency of symptoms as "0" (not at all) to "3" (nearly every day) and has been validated for use in primary care. The PHQ-9 total score ranges from 0 to 27 and is categorized as follows: None or minimal depression 0-4, Mild depression 5-9, Moderate depression 10-14, Moderately severe depression 15-19, and Severe depression 20-27. Cronbach's alpha coefficients range from 0.86 to 0.89 and test-retest correlations range from 0.84-0.95 within 48 hours and from 0.81-0.96 at 7-day reassessment. PHQ-9 scores were found to be highly correlated with Beck Depression Inventory scores in the general population (r=0.73).

GAD-7 is a valid and reliable 7-question scale to screen for 4 anxiety disorders: Post Traumatic Stress Disorder, Panic Disorder, Generalized Anxiety Disorder, and Social Phobia. A score of ≥10 indicates a high probability of 1 or more of these disorders. Cronbach's alpha is 0.92 and test-retest correlation is 0.83. GAD-7 scores also correlate highly with scores of 2 commonly-used anxiety scales: the Beck Anxiety Inventory (r=0.72) and the anxiety subscale of the Symptom Checklist-90 (r=0.74).

A modified version of the NASA Task Load Index (TLX) measures workload. The TLX rating sheet was administered assuming similar weights for each of the 5 task load items: mental demand, temporal demand (e.g., being rushed), effort, frustration, and performance. The original TLX includes a physical demand item which was not included herein, as it was not applicable for the task of interacting with Lumen. An overall task load index score was calculated as sum of the 5 task load items, each ranging from 1 to 7. A higher overall score reflected greater (unfavorable) demand.

UEQ-S is a validated instrument containing 8 items from the 26 items of the original UEQ. UEQ-S measures user experience. From the UEQ-S survey, scale values were calculated by rescaling the survey responses to the range -3 to 3 and UEQ-S total score were calculated as mean of survey responses. UEQ-S total scores \<-0.8 represented a negative evaluation, between -0.8 and 0.8 represented a neutral evaluation, and \>0.8 represent a positive evaluation.

WAI-Tech was based closely on the original 36-item WAI and uses a parallel set of 36 items rated on a 7-point scale (1 = "never" to 7 = "always") to measure the level of alliance between patient and digital coach along 3 domains: Task (12 items), Bond (12 items), and Goal (12 items). An overall score was calculated based on item means. A higher overall score reflected a greater treatment alliance.