Not provided
| ID | Type | Description | Link |
|---|---|---|---|
| 10.17605/OSF.IO/DCP4M | Other Identifier | Open Science Framework (OSF) |
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
| Name | Class |
|---|---|
| Harvard University | OTHER |
| Mclean Hospital | OTHER |
| TheraLoop AI, Inc. | UNKNOWN |
Not provided
Not provided
Not provided
Not provided
This study is testing a digital tool called BraveBot for young people who are receiving cognitive behavioral therapy (CBT) for anxiety, OCD, or related problems. BraveBot is a computer program, not a person. It talks with youth through their phone or computer while they do "face-your-fears"-style exposure therapy homework that their therapist has assigned. Sometimes an exposure is done with BraveBot's real-time coaching, and sometimes on their own; after each exposure, youth answer a few short questions about how it went.
Rather than dividing participants into separate groups, the study randomizes each individual exposure homework assignment. Every time a young person opens an eligible exposure, the system makes a 1:1 random assignment deciding whether that exposure is completed with BraveBot's support or independently (self-guided).
The main goal is to learn whether using BraveBot helps youth understand their exposure assignments better, put in more effort, stick with exposures when they are hard, feel more capable, and find exposures more helpful in "fighting back" against anxiety. The study also examines whether BraveBot increases the likelihood that assigned exposures are completed, and explores effects on anxiety symptoms and how safe, easy to use, and useful BraveBot feels for youth, their therapists, and parents.
BraveBot does not replace the therapist, diagnose, or design exposures; it only supports the homework the clinician has assigned, and is used under clinician oversight. A built-in safety system can detect possible risk-related language, pause the session, show crisis resources (such as 988), and notify the treating clinician. The study is conducted within routine outpatient psychology clinics at Mass General Brigham. Up to 40 youth ages 12-22 will take part.
Design. Open-label, within-participant, exposure-instance-level randomized trial embedded in routine outpatient exposure-based CBT at Mass General Brigham clinics. The unit of randomization is the individual clinician-assigned, "BraveBot-eligible" exposure homework instance. Each time a youth opens the link for a BraveBot-eligible exposure delivered via SMS reminder, the backend system immediately randomizes that single instance 1:1 to BraveBot-assisted vs self-guided completion.
Intervention. BraveBot is a scope-limited, clinician-supervised, large language model (GPT)-powered audio-to-audio conversational assistant (provided by TheraLoop AI, Inc.) that supports completion of clinician-assigned exposure homework. For BraveBot-assisted instances it clarifies the assignment and rationale, collects Subjective Units of Distress (SUDS) ratings and expectations, encourages persistence when anxiety rises, prompts about safety behaviors and "white-knuckling" at mid-exposure check-ins, and guides brief end-of-exposure reflection. It does not diagnose, generate exposure hierarchies, or make treatment decisions. For self-guided instances, the youth completes the same assigned exposure independently and then completes the same post-exposure survey.
Aims.
Aim 1 (primary): whether BraveBot improves exposure-homework QUALITY, assessed by five youth-reported 0-100 ratings per completed instance (Effort, Perceived Capability, Understanding, Ability to Continue, Perceived Helpfulness).
Aim 2 (secondary): whether BraveBot increases exposure-homework ADHERENCE (probability that a randomized instance is completed).
Aims 3-5 (exploratory): preliminary effects on clinical outcomes (RCADS anxiety, CALIS-C life interference, IUSC intolerance of uncertainty) and associations with exposure-related processes; acceptability/usability/burden among youth, clinicians, and parents; and the performance and burden of BraveBot's automated risk-detection and clinician-notification (safety) workflow.
Participants and schedule. Up to 40 youth ages 12-22 with a clinical or subclinical anxiety or related disorder for which exposure-based CBT is indicated. BraveBot access remains active for up to 45 homework-reminder days (about six weeks), with a 12-week backstop (i.e., 84 calendar days). Standardized self-report measures are collected at baseline (pre-BraveBot), at the end of the active BraveBot period, and at a ~3-week follow-up. A separate population of up to 10 clinicians may voluntarily enroll to use the BraveBot system with their patients and complete surveys about their experience.
Analysis. Primary outcomes are analyzed with linear mixed-effects models (fixed effect of condition; random intercepts for youth, and clinician/site as needed), with Benjamini-Hochberg false-discovery-rate correction across the five co-primary outcomes. Adherence is analyzed with a binomial-logit generalized linear mixed-effects model.
Not provided
Not provided
Not provided
Not provided
| Label | Type | Description | Intervention Names |
|---|---|---|---|
| BraveBot-Assisted Exposure | Experimental | Exposure instances randomized to this condition are completed with BraveBot, an LLM-powered audio-to-audio conversational assistant that provides real-time, structured coaching: clarifying the assigned exposure and its rationale, collecting SUDS ratings, encouraging persistence, prompting about safety behaviors and 'white-knuckling' at mid-exposure check-ins, and guiding brief post-exposure reflection. |
|
| Self-Guided Exposure | Active Comparator | Exposure instances randomized to this condition are completed independently by the youth (standard self-guided homework), without BraveBot coaching; the youth then completes the same brief post-exposure survey. |
|
| Name | Type | Description | Arm Group Labels | Other Names |
|---|---|---|---|---|
| BraveBot | Behavioral | Scope-limited, clinician-supervised, GPT-powered audio-to-audio conversational assistant delivered via a secure web link in SMS reminders. Supports clinician-assigned exposure homework by clarifying instructions and rationale, collecting SUDS, encouraging persistence, prompting about safety behaviors and white-knuckling, and guiding post-exposure reflection. Includes an automated risk-detection workflow that pauses the session, shows crisis resources, and notifies the treating clinician. |
| Measure | Description | Time Frame |
|---|---|---|
| Exposure Quality -- Effort (0-100) | Youth-reported effort put into the exposure, rated on a 0-100 slider after each completed randomized exposure instance. Minimum 0, maximum 100; higher scores indicate greater effort (better). | Immediately after each completed, randomized exposure instance during the active BraveBot period (up to 45 homework-reminder days, or the 12-week backstop). |
| Exposure Quality -- Perceived Capability (0-100) | Youth-reported sense of being capable to approach or handle anxiety-provoking situations, rated on a 0-100 slider after each completed randomized exposure instance. Minimum 0, maximum 100; higher scores indicate greater perceived capability (better). | Immediately after each completed, randomized exposure instance during the active BraveBot period (up to 45 homework-reminder days, or the 12-week backstop). |
| Exposure Quality -- Understanding (0-100) | Youth-reported understanding of what to do and why during the exposure, rated on a 0-100 slider after each completed randomized exposure instance. Minimum 0, maximum 100; higher scores indicate greater understanding (better). | Immediately after each completed, randomized exposure instance during the active BraveBot period (up to 45 homework-reminder days, or the 12-week backstop). |
| Exposure Quality -- Ability to Continue When Distressed (0-100) | Youth-reported ability to stay with the exposure when it was difficult or distressing, rated on a 0-100 slider after each completed randomized exposure instance. Minimum 0, maximum 100; higher scores indicate greater ability to continue (better). | Immediately after each completed, randomized exposure instance during the active BraveBot period (up to 45 homework-reminder days, or the 12-week backstop). |
| Exposure Quality -- Perceived Helpfulness (0-100) |
| Measure | Description | Time Frame |
|---|---|---|
| Exposure-homework Adherence | Proportion of randomized exposure instances classified as Complete (vs. Incomplete), compared between BraveBot-assisted and self-guided conditions. Completion is determined automatically from condition-specific workflow flags. Reported as a proportion (0-1); a higher completion proportion for BraveBot-assisted vs. self-guided is better. | For each randomized exposure instance during the active BraveBot period (up to 45 homework-reminder days, or the 12-week backstop). |
| Measure | Description | Time Frame |
|---|---|---|
| Revised Children's Anxiety and Depression Scale (RCADS) - Total Anxiety (youth-report) | Anxiety symptom measure; each item scored 0 (Never) to 3 (Always). The Total Anxiety score is the sum of the five anxiety subscales (separation anxiety, social phobia, generalized anxiety, panic, obsessive-compulsive). Total Anxiety score range 0-111; higher scores indicate greater anxiety symptoms (worse). | Baseline, end of active BraveBot period (average of ~6 weeks, up to max of 12 weeks), and follow-up (3 weeks following end of active BraveBot period). |
INCLUSION CRITERIA (youth participants):
EXCLUSION CRITERIA (present at enrollment):
The investigators will also recruit up to 10 clinicians at participating programs who may use the BraveBot system with their patients (who have enrolled in the study independently) and complete brief baseline and end-of-study surveys.
Not provided
Not provided
Not provided
Not provided
| Name | Role | Phone | Extension | |
|---|---|---|---|---|
| Joshua S Steinberg, MA | Contact | 6179453115 | jssteinberg@mgb.org |
| Name | Affiliation | Role |
|---|---|---|
| Madelaine R Abel, PhD | Massachusetts General Hospital | Principal Investigator |
| Facility | Status | City | State | ZIP | Country | Contacts |
|---|---|---|---|---|---|---|
| McLean Hospital | Belmont | Massachusetts | 02478 | United States |
| PubMed Identifier | Type | Citation | Retractions |
|---|---|---|---|
| Background | Zieve, G. G., Dong, L., & Harvey, A. G. (2019). Patient Memory for Psychological Treatment Contents: Assessment, Intervention, and Future Directions for a Novel Transdiagnostic Mechanism of Change. Behaviour Change, 36(1), 1-11. https://doi.org/10.1017/bec.2019.1 | ||
| 24135256 | Background | Lyneham HJ, Sburlati ES, Abbott MJ, Rapee RM, Hudson JL, Tolin DF, Carlson SE. Psychometric properties of the Child Anxiety Life Interference Scale (CALIS). J Anxiety Disord. 2013 Oct;27(7):711-9. doi: 10.1016/j.janxdis.2013.09.008. Epub 2013 Sep 26. | |
| 29137655 |
Not provided
| ID | Type | URL | Comment |
|---|---|---|---|
| Study Protocol | View IPD |
No routine sharing of individual participant data is planned at this time. De-identified, coded analytic datasets may be shared with named collaborators (Harvard University and TheraLoop AI, Inc.) under appropriate agreements and IRB approval. No deposit of individual-level data in public repositories is planned.
Not provided
Not provided
Not provided
Not provided
Not provided
| ID | Term |
|---|---|
| D001008 | Anxiety Disorders |
| ID | Term |
|---|---|
| D001523 | Mental Disorders |
Not provided
Not provided
Within-participant design. Each clinician-assigned, BraveBot-eligible exposure homework instance is individually randomized 1:1 at the moment of initiation to BraveBot-assisted vs self-guided completion; each youth therefore experiences both conditions and, theoretically, serves as their own control. That is, this is an instance-level, within-person randomized design rather than a fixed two-period crossover with washout.
Not provided
Not provided
Open label. Of note, treating clinicians cannot view the numeric 0-100 post-exposure quality ratings (Effort, Capability, Understanding, Ability to Continue, Helpfulness), which are stored separately and are accessible to study staff only; this is an information barrier for the primary outcomes rather than formal masking.
Not provided
|
|
| Self-Guided Exposure Therapy Homework | Behavioral | The youth completes the same clinician-assigned exposure independently, without BraveBot coaching (standard between-session homework). |
|
Youth-reported helpfulness of the exposure in reducing anxiety, rated on a 0-100 slider after each completed randomized exposure instance. Minimum 0, maximum 100; higher scores indicate greater perceived helpfulness (better).
| Immediately after each completed, randomized exposure instance during the active BraveBot period (up to 45 homework-reminder days, or the 12-week backstop). |
| Child Anxiety Life Interference Scale (CALIS) (youth-report) | Self-report measure of anxiety-related interference in the child's daily life. Each of the 8 items are rated on a 0-4 scale. Total score range 0-32; higher scores indicate greater anxiety-related life interference (worse). | Baseline, end of active BraveBot period (average of ~6 weeks, up to max of 12 weeks), and follow-up (3 weeks following end of active BraveBot period). |
| Intolerance of Uncertainty Scale-Child (IUSC) (youth-report) | Self-report measure of intolerance of uncertainty; items rated 1-5. Total score range 27-135; higher scores indicate greater intolerance of uncertainty (worse). | Baseline, end of active BraveBot period (average of ~6 weeks, up to max of 12 weeks), and follow-up (3 weeks following end of active BraveBot period). |
| Memory for Treatment Content (youth-report) | Vignette-based, multiple-choice assessment of memory for exposure-CBT principles, scored as proportion correct (0-100%); higher scores indicate better memory for treatment content (better). | End of active BraveBot period (average of ~6 weeks, max of 12 weeks) and follow-up (3 weeks later). |
| Developmentally Adapted or Original Version of the System Usability Scale (SUS) (youth- and clinician-report) | Standardized usability measure of the BraveBot system (developmentally adapted for youth; standard version for clinicians). Score range 0-100; higher scores indicate better perceived usability (better). | End of active BraveBot period for youth (i.e., ~6 weeks on average, max of 12 weeks); end of study for clinicians (anticipated to be one year after recruitment begins). |
| Global Post-BraveBot Peference/Acceptability ratings (youth-report) | Youth-reported 0-100 slider ratings comparing BraveBot-assisted vs. self-guided exposure (e.g., effort, persistence, understanding, helpfulness, enjoyment, preference, future use, safety, willingness to recommend). Each item: minimum 0, maximum 100; higher scores indicate a more favorable rating of BraveBot. | End of active BraveBot period (i.e., ~6 weeks on average, max of 12 weeks). |
| Clinician Attitudes Toward AI | Clinician-reported AI-attitudes scale at orientation (items rated on 1-5 scale; total scores range from 7 to 35); higher scores indicate more favorable attitudes. | At baseline (orientation) and end of study (anticipated to be one year after recruitment begins) |
| Clinician Questionnaire - Global Feedback on BraveBot | 0-100 global ratings of BraveBot's effects on exposure homework, workflow fit, acceptability, and clinical alignment, plus open-text feedback; higher scores indicate more favorable ratings. | End of study (anticipated to be one year after recruitment begins). |
| Novel Parent/Caregiver Perceptions of BraveBot Questionnaire | For parents/caregivers of youth participants under 18 years of age, 0-100 sliders comparing BraveBot-supported vs self-guided exposure homework (e.g., child willingness/engagement, persistence during distress, independence from parent, parent-child conflict/stress, structure/guidance, comfort/appropriateness, future preference, and concerns about AI). | At the end of the youth participant's active BraveBot period (i.e., ~6 weeks on average, max of 12 weeks). |
| Number of Automated Safety Alerts | Count of automated safety alerts generated by BraveBot's risk-detection workflow during BraveBot-assisted exposures. Reported as a count. | Throughout the active BraveBot period (i.e., average of 6 weeks; up to 12 weeks at most). |
| Clinician Response Time to Safety Alerts | Time from generation of an automated safety alert to documented clinician review, in hours. Reported as hours. | Throughout the active BraveBot period (i.e., average of 6 weeks; up to 12 weeks at most). |
| Proportion of Safety Alerts Classified as True Positive vs. False Positive | Among automated safety alerts reviewed by clinicians, the proportion classified as genuine concern (true positive) vs. false positive. Reported as a percentage. | Throughout the active BraveBot period (i.e., average of 6 weeks; up to 12 weeks at most). |
| Number of Clinician-Initiated Manual Safety Flags | Count of safety concerns flagged by clinicians that were not captured by the automated workflow (potential false negatives). Reported as a count. | Throughout the active BraveBot period (i.e., average of 6 weeks; up to 12 weeks at most). |
| Number of Youth-Reported "I Felt Unsafe" Responses | Count of post-exposure surveys on which a youth endorsed the "I felt unsafe" item. Reported as a count. | Throughout the active BraveBot period (i.e., average of 6 weeks; up to 12 weeks at most). |
| Massachusetts General Hospital | Boston | Massachusetts | 02114 | United States |
|
| Harvard University | Cambridge | Massachusetts | 02138 | United States |
|
| Background |
| Harvey AG, Dong L, Lee JY, Gumport NB, Hollon SD, Rabe-Hesketh S, Hein K, Haman K, McNamara ME, Weaver C, Martinez A, Notsu H, Zieve G, Armstrong CC. Can integrating the Memory Support Intervention into cognitive therapy improve depression outcome? Study protocol for a randomized controlled trial. Trials. 2017 Nov 14;18(1):539. doi: 10.1186/s13063-017-2276-x. |
| 28221056 | Background | Dong L, Lee JY, Harvey AG. Memory support strategies and bundles: A pathway to improving cognitive therapy for depression? J Consult Clin Psychol. 2017 Mar;85(3):187-199. doi: 10.1037/ccp0000167. |
| 19719351 | Background | Comer JS, Roy AK, Furr JM, Gotimer K, Beidas RS, Dugas MJ, Kendall PC. The intolerance of uncertainty scale for children: a psychometric evaluation. Psychol Assess. 2009 Sep;21(3):402-11. doi: 10.1037/a0016719. |
| 10937431 | Background | Chorpita BF, Yim L, Moffitt C, Umemoto LA, Francis SE. Assessment of symptoms of DSM-IV anxiety and depression in children: a revised child anxiety and depression scale. Behav Res Ther. 2000 Aug;38(8):835-55. doi: 10.1016/s0005-7967(99)00130-8. |
| Background | Brooke, J. (1995). SUS: A quick and dirty usability scale. Usability Eval. Ind., 189. |
| 41578705 | Background | Bekes V, Aafjes-van Doorn K. Who Wants to Have an AI Therapist? Acceptance of Using Artificial Intelligence for Mental Health Interventions Among Clinicians, Patients and the General Community. Clin Psychol Psychother. 2026 Jan-Feb;33(1):e70220. doi: 10.1002/cpp.70220. |