TEO: Assessing the Impact of Conversational AI in the Treatment of Stress and Anxiety in Aging Persons: A Randomized Controlled Trial Study (Preprint)

Abstract

BACKGROUND While mental health applications are increasingly available for large populations of users, there is a lack of controlled trials on the impact of such applications. An important population segment includes the senior adults. AI-empowered agents are evaluated when assisting adults with cognitive impairments, while few applications are available for aging adults who are still in their active working life. Those persons often suffer from high levels of stress related to changes in their working places, and the stress related symptoms eventually affect their quality of life. OBJECTIVE Evaluate the contribution of TEO (Therapy Empowerment Opportunity), an m-health agent based on conversational AI technology. TEO promotes mental health and wellbeing by reducing work related stress symptoms and anxiety levels in aging subjects. METHODS The study is based on a protocolized intervention for stress and anxiety management where patients with stress symptoms and mild-to-moderate levels of anxiety received eight-weeks CBT treatment delivered remotely. A group of participants also interacted with an m-health agent. Based on deep learning methodology, the agent assisted them in completing the therapy assignments, and engaged them in conversations where they were helped to explain the details of events that increased their anxiety. The participants were active workers aged over 55. Four experimental groups were selected. Group1 received traditional therapy while Group2 also conversed with the m-health agent. Group3 received support only by the m-health application, and Group4 did not receive any treatment. The symptoms related to stress (levels of anxiety, physical disease, and depression) were assessed prior to the treatment (T1), at the end (T2), and three months after (T3) by submitting standardized psychological questionnaires. At T2 participants in Groups 1, 2 and 3 filled in a satisfaction questionnaire. RESULTS Despite the randomization of the sample, at T1 statistically significant differences between groups were present: G4 showed lower levels of anxiety and depression compared to G1, and lower levels of stress compared to G2. Comparisons between groups at T2 and T3 did not show significant differences in outcomes. Analyses conducted within groups showed that there are significant differences between times in Group 2, for which greater improvements were found on the levels of stress and scales related to overall well-being. A general worsening trend between T2 and T3 can be detected in all groups, showing a significant increase in the levels of stress in G2. As for the satisfaction questionnaire, Group 2 reported higher levels of perceived usefulness and satisfaction. CONCLUSIONS In general, the study results showed that participants who received traditional CBT treatment with the support of the m-health application were likely to report better satisfaction and more stable trend of improvement limited to the individual perception of stress related symptoms. CLINICALTRIAL ClinicalTrials.gov Identifier: NCT04809090 INTERNATIONAL REGISTERED REPORT RR2-10.2196/30053

Examining the role of AI technology in online mental healthcare: opportunities, challenges, and implications, a mixed-methods review

“…The identified studies (n=29) were mostly conducted in the USA (n=10), Sweden (n=6), and the UK (n=6). The studies included a diverse sample of adults (n=22), college students and young adults (n=4), anonymous users (n=2), and adolescents (n=1) ( 28 – 33 , 41 , 73 – 94 ). Eight studies reported on the ethnicity of the participants ( 30 , 32 , 41 , 74 , 76 , 77 , 80 , 83 ), and in most of the studies between 31% and 91% of the participants identified as White.…”

“…The meta-analysis assessing the impact of AI-augmented interventions on mental health symptoms considered n=10 studies ( 41 , 73 – 78 , 80 , 84 , 90 ) presenting pre- and post-intervention outcomes. The results of this meta-analysis showed that these interventions had a significantly large effect size in the reduction of mental health symptoms.…”

“…The sub-group analysis showed that AI-self-guided eCBT had a significant medium effect size, AI-guided eCBT effect had a non-significant effect, and AI-modified eCBT, AI-self-guided ACT and Combined had a significant large effect size in the reduction of mental health symptoms ( Figure 3A ). The NMA considered n=6 RCTs ( 41 , 73 , 74 , 76 – 78 ). The head-to-head comparisons against AI-self-guided eCBT determined that this intervention was significantly more effective than AI chatbot with a large effect and psychoeducation with a medium effect size for the reduction of mental health symptoms.…”

“…The sub-group analysis showed that AI-self-guided eCBT had a significant small effect size, AI-guided eCBT effect had a non-significant effect, and AI-modified eCBT had a significant medium to large effect size in the reduction of anxiety symptoms ( Figure 3C ). The NMA considered n=4 RCTs ( 73 , 74 , 76 , 78 ). The head-to-head comparisons against AI-self-guided eCBT determined that the effects of this intervention were not significantly different than AI-guided eCBT, combined, no intervention, psychoeducation, and remote CBT ( Figure 4C ).…”

“…No specific trends were identified when comparing the attrition rates with intervention length or whether the online intervention was therapist guided or not ( Table 1 ). The NMA considered n=6 RCTs ( 41 , 73 , 74 , 76 – 78 ). The head-to-head comparisons against AI-self-guided eCBT determined that this intervention had a 14% lesser risk of dropping out compared to psychoeducation, this comparison was statistically significant.…”