Background: General awareness and exposure to generative artificial intelligence (AI) have increased recently. This transformative technology has the potential to create a more dynamic and engaging user experience in digital mental health interventions (DMHIs). However, if not appropriately used and controlled, it can introduce risks to users that may result in harm and erode trust. At the time of conducting this trial, there had not been a rigorous evaluation of an approach to safely implementing generative AI in a DMHI. Objective: This study aims to explore the user relationship, experience, safety, and technical guardrails of a DMHI using generative AI compared with a rules-based intervention. Methods: We conducted a 2-week exploratory randomized controlled trial (RCT) with 160 adult participants randomized to receive a generative AI (n=81) or rules-based (n=79) version of a conversation-based DMHI. Self-report measures of the user relationship (client satisfaction, working alliance bond, and accuracy of empathic listening and reflection) and experience (engagement metrics, adverse events, and technical guardrail success) were collected. Descriptions and validation of technical guardrails for handling user inputs (eg, detecting potentially concerning language and off-topic responses) and model outputs (eg, not providing medical advice and not providing a diagnosis) are provided, along with examples to illustrate how they worked. Safety monitoring was conducted throughout the trial for adverse events, and the success of technical guardrails created for the generative arm was assessed post trial. Results: In general, the majority of measures of user relationship and experience appeared to be similar in both the generative and rules-based arms. The generative arm appeared to be more accurate at detecting and responding to user statements with empathy (98% accuracy vs 69%). There were no serious or device-related adverse events, and technical guardrails were shown to be 100% successful in posttrial review of generated statements. A majority of participants in both groups reported an increase in positive sentiment (62% and 66%) about AI at the end of the trial. Conclusions: This trial provides initial evidence that, with the right guardrails and process, generative AI can be successfully used in a digital mental health intervention (DMHI) while maintaining the user experience and relationship. It also provides an initial blueprint for approaches to technical and conversational guardrails that can be replicated to build a safe DMHI. Trial Registration: ClinicalTrials.gov NCT05948670; https://clinicaltrials.gov/study/NCT05948670

Using a single case design, a parent-mediated spoken-language intervention was delivered to three mothers and their school-aged sons with fragile X syndrome, the leading inherited cause of intellectual disability. The intervention was embedded in the context of shared storytelling using wordless picture books and targeted three empirically derived language-support strategies. All sessions were implemented through distance videoteleconferencing. Parent education sessions were followed by 12 weekly clinician coaching and feedback sessions. Data were collected weekly during independent homework and clinician observation sessions. Relative to baseline, mothers increased their use of targeted strategies, and dyads increased the frequency and duration of story-related talking. Generalized effects of the intervention on lexical diversity and grammatical complexity were observed. Implications for practice are discussed.

BACKGROUND. FXLEARN, the first-ever large multisite trial of effects of disease-targeted pharmacotherapy on learning, was designed to explore a paradigm for measuring effects of mechanism-targeted treatment in fragile X syndrome (FXS). In FXLEARN, the effects of metabotropic glutamate receptor type 5 (mGluR5) negative allosteric modulator (NAM) AFQ056 on language learning were evaluated in 3- to 6-year-old children with FXS, expected to have more learning plasticity than adults, for whom prior trials of mGluR5 NAMs have failed. METHODS. After a 4-month single-blind placebo lead-in, participants were randomized 1:1 to AFQ056 or placebo, with 2 months of dose optimization to the maximum tolerated dose, then 6 months of treatment during which a language-learning intervention was implemented for both groups. The primary outcome was a centrally scored videotaped communication measure, the Weighted Communication Scale (WCS). Secondary outcomes were objective performance-based and parent-reported cognitive and language measures. RESULTS. FXLEARN enrolled 110 participants, randomized 99, and had 91 who completed the placebo-controlled period. Although both groups made language progress and there were no safety issues, the change in WCS score during the placebo-controlled period was not significantly different between the AFQ056 and placebo-treated groups, nor were there any significant between-group differences in change in any secondary measures. CONCLUSION. Despite the large body of evidence supporting use of mGluR5 NAMs in animal models of FXS, this study suggests that this mechanism of action does not translate into benefit for the human FXS population and that better strategies are needed to determine which mechanisms will translate from preclinical models to humans in genetic neurodevelopmental disorders. TRIAL REGISTRATION. ClincalTrials.gov NCT02920892. FUNDING SOURCES. NeuroNEXT network NIH grants U01NS096767, U24NS107200, U24NS107209, U01NS077323, U24NS107183, U24NS107168, U24NS107128, U24NS107199, U24NS107198, U24NS107166, U10NS077368, U01NS077366, U24NS107205, U01NS077179, and U01NS077352; NIH grant P50HD103526; and Novartis IIT grant AFQ056X2201T for provision of AFQ056.

Background: In addition to significant cognitive delays, boys with fragile X syndrome display phenotypic characteristics that include delays in language, inattention, social anxiety, and escape-maintained challenging behaviors. Despite these challenges, families affected by fragile X syndrome often have limited access to center-based intervention programs. Methods: The present study utilized a multiple baseline design across participants to examine the preliminary effectiveness of a 12-week, parent-implemented spoken language intervention for three 5- to 7-year-old boys with fragile X syndrome. The goal of the intervention was to teach the biological mothers of each boy to use a set of verbally responsive language support strategies while participating in shared book reading activities with her child. All aspects of the intervention, including pre- and post-treatment measures, were delivered into the family’s home via distance video teleconferencing. Results: Results from this study provide preliminary support for the efficacy of this parent-implemented language intervention approach by demonstrating generalized improvements in both targeted maternal strategy use and measures of child spoken language. Conclusions: The present study expands upon prior language intervention research utilizing distance delivery of services for families affected by fragile X syndrome. Additionally, the study provides early support for the feasibility of collecting pre- and post-treatment assessments at a distance.

Although previous research supports the notion that characteristics of both the child and the mother impact maternal well-being and parenting stress in mothers of children with FXS, more work is needed in which self-report measures are supplemented by physiological measures of stress and well-being. The inclusion of physiological measures, such as heart rate variability (HRV), may provide a window into the biological origins and consequences of maternal perceptions of their experiences, including the challenges of raising a child with developmental challenges. The proposed project, therefore, involved the collection of multimodal assessment data from mothers and their school-aged children with FXS. Further, given the importance of understanding how mothers of youth with FXS are faring in their everyday environment, the present study collected all data using telehealth procedures and wearable technology. Participants were 20 biological mothers and their children with FXS between the ages of 6 and 11 years. We measured maternal mental health and parenting stress through self-report as well as through HRV as a more objective measure of psychological well-being. We also examined the associations between these variables and child characteristics such as externalizing and internalizing behaviors as well as autism symptomatology. We found significant support for an elevated rate of depressive symptoms in the sample of mothers (35%) and some potential indicators for heightened rates of anxiety (15%) when compared to normed samples and rates in the general population. We also found that the mothers presented with an atypical HRV profile akin to those experiencing depression or chronic stress, although limitations of the present measure suggest the need for additional confirmatory research. Further, we found that child externalizing behaviors were the primary correlates of maternal well-being. These findings contribute to the growing body of literature regarding the unique challenges faced by these mother-child dyads and supports the importance of increasing the availability of services available to these mothers, not only for meeting the needs of their children's development and behavior, but in supporting their own well-being as well.

Background: Decentralized clinical trials (DCT), digital survey methodologies, and health monitoring technologies create the potential to reduce study participant burden as well as enhance sample diversity and enrollment pace. However, fraudulent participant activity poses a significant threat to study validity and data integrity. Purpose: This study quantifies fraudulent attempts at participation in a DCT of a mobile mental health intervention for problematic substance use and discusses methods to prevent and detect fraudulent activity. Methods: Adults residing in the US reporting problematic substance use were recruited for a fully remote DCT via 2 primary channels: social media advertisements and survey panels. The DCT offered incentives totaling up to $100 for completing assessments over the 12-week study. To prevent and detect fraudulent activity, the research team utilized VPN and proxy detection, checked for duplicate identifiers (e.g., emails, phone numbers, IP Addresses), and compared age and date of birth (DOB) responses across timepoints. Descriptive statistics and group comparisons across the 2 sources of recruitment methodology were utilized to quantify and characterize fraudulent activity. Results: Of the 2,781 eligible screeners completed, 1,725 (62%) were determined to be fraudulent prior to randomization, detected most commonly by duplicate identifiers (65%) and/or VPN and proxy detection (47%). Of the 258 randomized participants, 51 (20%) were later determined to be fraudulent based upon age and/or DOB mismatch. Notable patterns in fraudulent activity (e.g., 42% of fraudulent screening respondents reported the exact age of 30 years; stylistic formatting of email address accounts) were identified. The fraudulent recruitment rate was higher for social media advertising (85%) than survey panels (26%). Conclusions: Both social media and survey panel recruitment resulted in high levels of fraudulent activity in a DCT of a mobile mental health intervention. Researchers conducting DCTs and/or online surveys are urged to take several precautions and preventative measures to insulate against fraudulent activity including embedding identity verification procedures in consent processes. Researchers should consider making personal contact with a participant to verify identity as well as remain vigilant for fraudulent activity and its real-time dynamic potential. Trial Registration: NCT04925570.