1
|
Ding H, Simmich J, Vaezipour A, Andrews N, Russell T. Evaluation framework for conversational agents with artificial intelligence in health interventions: a systematic scoping review. J Am Med Inform Assoc 2024; 31:746-761. [PMID: 38070173 PMCID: PMC10873847 DOI: 10.1093/jamia/ocad222] [Citation(s) in RCA: 1] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/23/2023] [Revised: 11/04/2023] [Accepted: 11/24/2023] [Indexed: 02/18/2024] Open
Abstract
OBJECTIVES Conversational agents (CAs) with emerging artificial intelligence present new opportunities to assist in health interventions but are difficult to evaluate, deterring their applications in the real world. We aimed to synthesize existing evidence and knowledge and outline an evaluation framework for CA interventions. MATERIALS AND METHODS We conducted a systematic scoping review to investigate designs and outcome measures used in the studies that evaluated CAs for health interventions. We then nested the results into an overarching digital health framework proposed by the World Health Organization (WHO). RESULTS The review included 81 studies evaluating CAs in experimental (n = 59), observational (n = 15) trials, and other research designs (n = 7). Most studies (n = 72, 89%) were published in the past 5 years. The proposed CA-evaluation framework includes 4 evaluation stages: (1) feasibility/usability, (2) efficacy, (3) effectiveness, and (4) implementation, aligning with WHO's stepwise evaluation strategy. Across these stages, this article presents the essential evidence of different study designs (n = 8), sample sizes, and main evaluation categories (n = 7) with subcategories (n = 40). The main evaluation categories included (1) functionality, (2) safety and information quality, (3) user experience, (4) clinical and health outcomes, (5) costs and cost benefits, (6) usage, adherence, and uptake, and (7) user characteristics for implementation research. Furthermore, the framework highlighted the essential evaluation areas (potential primary outcomes) and gaps across the evaluation stages. DISCUSSION AND CONCLUSION This review presents a new framework with practical design details to support the evaluation of CA interventions in healthcare research. PROTOCOL REGISTRATION The Open Science Framework (https://osf.io/9hq2v) on March 22, 2021.
Collapse
Affiliation(s)
- Hang Ding
- RECOVER Injury Research Centre, Faculty of Health and Behavioural Sciences, The University of Queensland, Brisbane, QLD, Australia
- STARS Education and Research Alliance, Surgical Treatment and Rehabilitation Service (STARS), The University of Queensland and Metro North Health, Brisbane, QLD, Australia
| | - Joshua Simmich
- RECOVER Injury Research Centre, Faculty of Health and Behavioural Sciences, The University of Queensland, Brisbane, QLD, Australia
- STARS Education and Research Alliance, Surgical Treatment and Rehabilitation Service (STARS), The University of Queensland and Metro North Health, Brisbane, QLD, Australia
| | - Atiyeh Vaezipour
- RECOVER Injury Research Centre, Faculty of Health and Behavioural Sciences, The University of Queensland, Brisbane, QLD, Australia
- STARS Education and Research Alliance, Surgical Treatment and Rehabilitation Service (STARS), The University of Queensland and Metro North Health, Brisbane, QLD, Australia
| | - Nicole Andrews
- RECOVER Injury Research Centre, Faculty of Health and Behavioural Sciences, The University of Queensland, Brisbane, QLD, Australia
- STARS Education and Research Alliance, Surgical Treatment and Rehabilitation Service (STARS), The University of Queensland and Metro North Health, Brisbane, QLD, Australia
- The Tess Cramond Pain and Research Centre, Metro North Hospital and Health Service, Brisbane, QLD, Australia
- The Occupational Therapy Department, The Royal Brisbane and Women’s Hospital, Metro North Hospital and Health Service, Brisbane, QLD, Australia
| | - Trevor Russell
- RECOVER Injury Research Centre, Faculty of Health and Behavioural Sciences, The University of Queensland, Brisbane, QLD, Australia
- STARS Education and Research Alliance, Surgical Treatment and Rehabilitation Service (STARS), The University of Queensland and Metro North Health, Brisbane, QLD, Australia
| |
Collapse
|
2
|
Li H, Zhang R, Lee YC, Kraut RE, Mohr DC. Systematic review and meta-analysis of AI-based conversational agents for promoting mental health and well-being. NPJ Digit Med 2023; 6:236. [PMID: 38114588 PMCID: PMC10730549 DOI: 10.1038/s41746-023-00979-5] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/20/2023] [Accepted: 11/29/2023] [Indexed: 12/21/2023] Open
Abstract
Conversational artificial intelligence (AI), particularly AI-based conversational agents (CAs), is gaining traction in mental health care. Despite their growing usage, there is a scarcity of comprehensive evaluations of their impact on mental health and well-being. This systematic review and meta-analysis aims to fill this gap by synthesizing evidence on the effectiveness of AI-based CAs in improving mental health and factors influencing their effectiveness and user experience. Twelve databases were searched for experimental studies of AI-based CAs' effects on mental illnesses and psychological well-being published before May 26, 2023. Out of 7834 records, 35 eligible studies were identified for systematic review, out of which 15 randomized controlled trials were included for meta-analysis. The meta-analysis revealed that AI-based CAs significantly reduce symptoms of depression (Hedge's g 0.64 [95% CI 0.17-1.12]) and distress (Hedge's g 0.7 [95% CI 0.18-1.22]). These effects were more pronounced in CAs that are multimodal, generative AI-based, integrated with mobile/instant messaging apps, and targeting clinical/subclinical and elderly populations. However, CA-based interventions showed no significant improvement in overall psychological well-being (Hedge's g 0.32 [95% CI -0.13 to 0.78]). User experience with AI-based CAs was largely shaped by the quality of human-AI therapeutic relationships, content engagement, and effective communication. These findings underscore the potential of AI-based CAs in addressing mental health issues. Future research should investigate the underlying mechanisms of their effectiveness, assess long-term effects across various mental health outcomes, and evaluate the safe integration of large language models (LLMs) in mental health care.
Collapse
Affiliation(s)
- Han Li
- Department of Communications and New Media, National University of Singapore, Singapore, 117416, Singapore
| | - Renwen Zhang
- Department of Communications and New Media, National University of Singapore, Singapore, 117416, Singapore.
| | - Yi-Chieh Lee
- Department of Computer Science, National University of Singapore, Singapore, 117416, Singapore
| | - Robert E Kraut
- Human-Computer Interaction Institute Carnegie Mellon University, Pittsburgh, PA, 15213, USA
| | - David C Mohr
- Center for Behavioral Intervention Technologies, Department of Preventive Medicine, Northwestern University, Chicago, IL, 60611, USA
| |
Collapse
|
3
|
Motevalli S, Salahshour HM, Bailey RP. The mediating role of cognitive flexibility in the relationship between cognitive emotion regulation strategies and mindfulness in patients with type 2 diabetes. J Affect Disord 2023; 339:676-682. [PMID: 37479040 DOI: 10.1016/j.jad.2023.07.043] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 03/09/2023] [Revised: 06/19/2023] [Accepted: 07/08/2023] [Indexed: 07/23/2023]
Abstract
INTRODUCTION This study aimed to examine the mediating role of cognitive flexibility in the relationship between cognitive emotion regulation strategies and mindfulness in patients with type 2 diabetes. METHODS The research was conducted by correlation method) using Structural Equation Modeling). The statistical population consisted of all women and men with type 2 diabetes. Two hundred fifty-three samples were selected by convenience sampling method. The participants responded to the Cognitive Emotion Regulation Questionnaire, the Kentucky inventory of mindfulness skills, and the Cognitive Flexibility Inventory. RESULTS The results showed that the total path coefficient between the adaptive cognitive emotion regulation strategies and mindfulness (β = 0.243, P = 0.005) was positive and significant, and the total path coefficient between the maladaptive cognitive emotion regulation strategies and mindfulness (β = -0.453, P = 0.001) was negative and significant. The path coefficient between cognitive flexibility and mindfulness (β = 0.273, P = 0.009) was positive and significant. The indirect path coefficient between the adaptive cognitive emotion regulation strategies and mindfulness (β = 0.094, P = 0.007) was positive and significant, and the indirect path coefficient between the maladaptive cognitive emotion regulation strategies and mindfulness (β = -0.117, P = 0.009) was negative and significant. CONCLUSION Improving emotion regulation skills increases cognitive flexibility and mindfulness in patients with type 2 diabetes.
Collapse
Affiliation(s)
- Saeid Motevalli
- Department of Psychology, Faculty of Social Sciences & Liberal Arts, UCSI University, Kuala Lumpur, Malaysia.
| | | | - Richard Peter Bailey
- Department of Education, Faculty of Social Sciences & Liberal Arts, UCSI University, Kuala Lumpur, Malaysia
| |
Collapse
|
4
|
Bassi G, Donadello I, Gabrielli S, Salcuni S, Giuliano C, Forti S. Early Development of a Virtual Coach for Healthy Coping Interventions in Type 2 Diabetes Mellitus: Validation Study. JMIR Form Res 2022; 6:e27500. [PMID: 35147505 PMCID: PMC8881774 DOI: 10.2196/27500] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/27/2021] [Revised: 04/27/2021] [Accepted: 12/20/2021] [Indexed: 11/15/2022] Open
Abstract
Background Mobile health solutions aimed at monitoring tasks among people with diabetes mellitus (DM) have been broadly applied. However, virtual coaches (VCs), embedded or not in mobile health, are considered valuable means of improving patients’ health-related quality of life and ensuring adherence to self-care recommendations in diabetes management. Despite the growing need for effective, healthy coping digital interventions to support patients’ self-care and self-management, the design of psychological digital interventions that are acceptable, usable, and engaging for the target users still represents the main challenge, especially from a psychosocial perspective. Objective This study primarily aims to test VC interventions based on psychoeducational and counseling approaches to support and promote healthy coping behaviors in adults with DM. As a preliminary study, university students have participated in it and have played the standardized patients’ (SPs) role with the aim of improving the quality of the intervention protocol in terms of user acceptability, experience, and engagement. The accuracy of users’ role-playing is further analyzed. Methods This preliminary study is based on the Obesity-Related Behavioral Intervention Trial model, with a specific focus on its early phases. The healthy coping intervention protocol was initially designed together with a team of psychologists following the main guidelines and recommendations for psychoeducational interventions for healthy coping in the context of DM. The protocol was refined with the support of 3 experts in the design of behavioral intervention technologies for mental health and well-being, who role-played 3 SPs’ profiles receiving the virtual coaching intervention in a Wizard of Oz setting via WhatsApp. A refined version of the healthy coping protocol was then iteratively tested with a sample of 18 university students (mean age 23.61, SD 1.975 years) in a slightly different Wizard of Oz evaluation setting. Participants provided quantitative and qualitative postintervention feedback by reporting their experiences with the VC. Clustering techniques on the logged interactions and dialogs between the VC and users were collected and analyzed to identify additional refinements for future VC development. Results Both quantitative and qualitative analyses showed that the digital healthy coping intervention was perceived as supportive, motivating, and able to trigger self-reflection on coping strategies. Analyses of the logged dialogs showed that most of the participants accurately played the SPs’ profile assigned, confirming the validity and usefulness of this testing approach in preliminary assessments of behavioral digital interventions and protocols. Conclusions This study outlined an original approach to the early development and iterative testing of digital healthy coping interventions for type 2 DM. Indeed, the intervention was well-accepted and proved its effectiveness in the definition and refinement of the initial protocol and of the user experience with a VC before directly involving real patients in its subsequent use and testing.
Collapse
Affiliation(s)
- Giulia Bassi
- Department of Developmental and Socialization Psychology, University of Padova, Padova, Italy.,Centre Digital Health & Wellbeing, Fondazione Bruno Kessler, Trento, Italy
| | - Ivan Donadello
- KRDB Research Centre, Free University of Bozen-Bolzano, Bolzano, Italy
| | - Silvia Gabrielli
- Centre Digital Health & Wellbeing, Fondazione Bruno Kessler, Trento, Italy
| | - Silvia Salcuni
- Department of Developmental and Socialization Psychology, University of Padova, Padova, Italy
| | - Claudio Giuliano
- Centre Digital Health & Wellbeing, Fondazione Bruno Kessler, Trento, Italy
| | - Stefano Forti
- Centre Digital Health & Wellbeing, Fondazione Bruno Kessler, Trento, Italy
| |
Collapse
|