Trial of a Voice-Based Generative AI Emotional Support Companion
A Randomised Controlled Trial of a Voice-Based Generative AI Emotional Support Companion
The investigators conduct a randomized controlled trial (RCT) to study an AI-based emotional support app. The focus in this paper is on two main areas: (1) changes in anxiety levels, and (2) safety.
Studie Overzicht
Toestand
Conditie
Interventie / Behandeling
Studietype
Inschrijving (Werkelijk)
Fase
- Niet toepasbaar
Contacten en locaties
Studie Locaties
-
-
California
-
San Francisco, California, Verenigde Staten, 94108
- Office of Sonia Health
-
-
Deelname Criteria
Geschiktheidscriteria
Leeftijden die in aanmerking komen voor studie
Accepteert gezonde vrijwilligers
Geslachten die in aanmerking komen voor studie
Beschrijving
Inclusion Criteria:
- Uses an iPhone running iOS 18 or newer to complete the screening survey
- Reports using an iPhone as their primary phone
- Aged 18 years or older
- Currently located in the United States and a US resident
- Not currently receiving psychological therapy
- Reports less than "very" or "extremely" concerned about privacy when using AI for emotional support
- Reports a comfort level of at least 50 on a 0-100 scale for using AI for emotional support
- Generalised Anxiety Disorder-7 (GAD-7) score between 13 and 20, inclusive
- Successfully completes the phone screening and confirms sufficient time to participate in the study
Exclusion Criteria:*
- Current suicidal ideation identified during the phone screening
- Residence in Nevada or Illinois
- Failure to meet minimum engagement or data-quality criteria during the screening survey, including extremely short completion times inconsistent with attentive participation
- Failure to meet any of the inclusion criteria above
Studie plan
Hoe is de studie opgezet?
Ontwerpdetails
- Primair doel: Ondersteunende zorg
- Toewijzing: Gerandomiseerd
- Interventioneel model: Parallelle opdracht
- Masker: Geen (open label)
Wapens en interventies
Deelnemersgroep / Arm |
Interventie / Behandeling |
|---|---|
|
Experimenteel: Sonia Arm
This arm got access to the AI-based emotional support app |
Sonia is a commercially available voice-based application built around a real-time conversation with a voice generated by artificial intelligence (AI). The application first assesses the user's situation during the initial session and uses this information to create a ''case conceptualisation''. This conceptualisation guides the content and structure of subsequent sessions and is continuously updated as new information emerges. Each of these subsequent sessions is structured around the five standard cognitive behavioural therapy (CBT) building blocks (a mood check, collaborative agenda setting, review of between-session practice, introduction and rehearsal of a specific technique, and a homework assignment). Technically, Sonia uses a reasoning model to create and continuously update the ''case conceptualisation''. The model is prompted to draw on its learned knowledge base when creating or updating the conceptualisation. |
|
Placebo-vergelijker: Control Group
This group received web-based psychoeducational material |
The material consisted of a summary of publicly available information published by the WHO. |
Wat meet het onderzoek?
Primaire uitkomstmaten
Uitkomstmaat |
Maatregel Beschrijving |
Tijdsspanne |
|---|---|---|
|
Generalised Anxiety Disorder-7
Tijdsspanne: 2 weeks
|
The Generalised Anxiety Disorder-7 (GAD-7) is a 7-item self-report measure of generalised anxiety symptoms, with scores ranging from 0 to 21 and higher scores indicating greater symptom severity (see Spitzer et al. 2006). |
2 weeks
|
|
Self-Reported Perception of Safety Relative to Other AI-Chatbots
Tijdsspanne: 2 weeks
|
Answers to a survey question asking wether the intervention feels safe relative to traditional AI-chatbots (yes/no). If participants answer no, they are asked in an open-text box to explain their reasoning. Summaries of these answers will also be reported. |
2 weeks
|
Medewerkers en onderzoekers
Sponsor
Medewerkers
Studie record data
Bestudeer belangrijke data
Studie start (Werkelijk)
Primaire voltooiing (Werkelijk)
Studie voltooiing (Werkelijk)
Studieregistratiedata
Eerst ingediend
Eerst ingediend dat voldeed aan de QC-criteria
Eerst geplaatst (Werkelijk)
Updates van studierecords
Laatste update geplaatst (Werkelijk)
Laatste update ingediend die voldeed aan QC-criteria
Laatst geverifieerd
Meer informatie
Termen gerelateerd aan deze studie
Aanvullende relevante MeSH-voorwaarden
Andere studie-ID-nummers
- 240088NR
Plan Individuele Deelnemersgegevens (IPD)
Bent u van plan om gegevens van individuele deelnemers (IPD) te delen?
Beschrijving IPD-plan
We will share data from our surveys using a github repository. However, this data is sufficiently sparse so that no individual can be identified from it.
IPD-tijdsbestek voor delen
Available from September 2026
IPD-toegangscriteria voor delen
Publicly available
IPD delen Ondersteunend informatietype
- LEERPROTOCOOL
- ANALYTIC_CODE
Informatie over medicijnen en apparaten, studiedocumenten
Bestudeert een door de Amerikaanse FDA gereguleerd geneesmiddel
Bestudeert een door de Amerikaanse FDA gereguleerd apparaatproduct
Deze informatie is zonder wijzigingen rechtstreeks van de website clinicaltrials.gov gehaald. Als u verzoeken heeft om uw onderzoeksgegevens te wijzigen, te verwijderen of bij te werken, neem dan contact op met register@clinicaltrials.gov. Zodra er een wijziging wordt doorgevoerd op clinicaltrials.gov, wordt deze ook automatisch bijgewerkt op onze website .