Designing ECAs to Improve Robustness of Human-Machine Dialogue

Mencía, Beatriz López; Pardo, David; Trapote, Álvaro Hernández; Gómez, Luis A. Hernández

doi:10.4018/978-1-60960-617-6.ch003

Cited by 4 publications

(1 citation statement)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We opted to include a digital counselor because this interface can increase interest, connection, and retention [59,60]. There is objective evidence that digital counselors must possess key characteristics to be effective [61][62][63]. Certain animations in facial features and gestures convey empathy, support, and acceptance while also directing attention to behavior change [64].…”

Section: Relevant Engaging Digital Counselor Development and Integrationmentioning

confidence: 99%

A Digital Counselor-Delivered Intervention for Substance Use Among People With HIV: Development and Usability Study

Hutton,

Aggarwal,

Gillani

et al. 2023

JMIR Form Res

View full text Add to dashboard Cite

Background Substance use disorders are prevalent and undertreated among people with HIV. Computer-delivered interventions (CDIs) show promise in expanding reach, delivering evidence-based care, and offering anonymity. Use in HIV clinic settings may overcome access barriers. Incorporating digital counselors may increase CDI engagement, and thereby improve health outcomes. Objective We aim to develop and pilot a digital counselor–delivered brief intervention for people with HIV who use drugs, called “C-Raven,” which is theory grounded and uses evidence-based practices for behavior change. Methods Intervention mapping was used to develop the CDI including a review of the behavior change research in substance use, HIV, and digital counselors. We conducted in-depth interviews applying the situated-information, motivation, and behavior skills model and culturally adapting the content for local use with people with HIV. With a user interaction designer, we created various digital counselors and CDI interfaces. Finally, a mixed methods approach using in-depth interviews and quantitative assessments was used to assess the usability, acceptability, and cultural relevance of the intervention content and the digital counselor. Results Participants found CDI easy to use, useful, relevant, and motivating. A consistent suggestion was to provide more information about the negative impacts of drug use and the interaction of drug use with HIV. Participants also reported that they learned new information about drug use and its health effects. The CDI was delivered by a “Raven,” digital counselor, programmed to interact in a motivational interviewing style. The Raven was perceived to be nonjudgmental, understanding, and emotionally responsive. The appearance and images in the intervention were perceived as relevant and acceptable. Participants noted that they could be more truthful with a digital counselor, however, it was not unanimously endorsed as a replacement for a human counselor. The C-Raven Satisfaction Scale showed that all participants rated their satisfaction at either a 4 (n=2) or a 5 (n=8) on a 5-point Likert scale and all endorsed using the C-Raven program again. Conclusions CDIs show promise in extending access to care and improving health outcomes but their development necessarily requires integration from multiple disciplines including behavioral medicine and computer science. We developed a cross-platform compatible CDI led by a digital counselor that interacts in a motivational interviewing style and (1) uses evidence-based behavioral change methods, (2) is culturally adapted to people with HIV who use drugs, (3) has an engaging and interactive user interface, and (4) presents personalized content based on participants’ ongoing responses to a series of menu-driven conversations. To advance the continued development of this and other CDIs, we recommend expanded testing, standardized measures to evaluate user experience, integration with clinician-delivered substance use treatment, and if effective, implementation into HIV clinical care.

show abstract

Section: Relevant Engaging Digital Counselor Development and Integrationmentioning

confidence: 99%

A Digital Counselor-Delivered Intervention for Substance Use Among People With HIV: Development and Usability Study

Hutton,

Aggarwal,

Gillani

et al. 2023

JMIR Form Res

View full text Add to dashboard Cite

show abstract

Analyzing the determinants to accept a virtual assistant and use cases among cancer patients: a mixed methods study

Bussel

Odekerken‐Schröder

et al. 2022

BMC Health Serv Res

View full text Add to dashboard Cite

Background Technological progress in artificial intelligence has led to the increasing popularity of virtual assistants, i.e., embodied or disembodied conversational agents that allow chatting with a technical system in a natural language. However, only little comprehensive research is conducted about patients' perceptions and possible applications of virtual assistant in healthcare with cancer patients. This research aims to investigate the key acceptance factors and value-adding use cases of a virtual assistant for patients diagnosed with cancer. Methods Qualitative interviews with eight former patients and four doctors of a Dutch radiotherapy institute were conducted to determine what acceptance factors they find most important for a virtual assistant and gain insights into value-adding applications. The unified theory of acceptance and use of technology (UTAUT) was used to structure perceptions and was inductively modified as a result of the interviews. The subsequent research model was triangulated via an online survey with 127 respondents diagnosed with cancer. A structural equation model was used to determine the relevance of acceptance factors. Through a multigroup analysis, differences between sample subgroups were compared. Results The interviews found support for all factors of the UTAUT: performance expectancy, effort expectancy, social influence and facilitating conditions. Additionally, self-efficacy, trust, and resistance to change, were added as an extension of the UTAUT. Former patients found a virtual assistant helpful in receiving information about logistic questions, treatment procedures, side effects, or scheduling appointments. The quantitative study found that the constructs performance expectancy (ß = 0.399), effort expectancy (ß = 0.258), social influence (ß = 0.114), and trust (ß = 0.210) significantly influenced behavioral intention to use a virtual assistant, explaining 80% of its variance. Self-efficacy (ß = 0.792) acts as antecedent of effort expectancy. Facilitating conditions and resistance to change were not found to have a significant relationship with user intention. Conclusions Performance and effort expectancy are the leading determinants of virtual assistant acceptance. The latter is dependent on a patient’s self-efficacy. Therefore, including patients during the development and introduction of a VA in cancer treatment is important. The high relevance of trust indicates the need for a reliable, secure service that should be promoted as such. Social influence suggests using doctors in endorsing the VA.

show abstract

Analysing the determinants to accept a virtual assistant and use cases among cancer patients: a mixed methods study (Preprint)

Bussel¹,

Odekerken‐Schröder²,

Ou³

et al. 2022

Preprint

View full text Add to dashboard Cite

BACKGROUND Technological progress in artificial intelligence has led to the increasing popularity of virtual assistants, i.e., embodied conversational agents that allow chatting with a technical system in a natural language. However, only little comprehensive research is conducted about patients' perceptions and possible applications of virtual assistant in healthcare with serious illnesses. Therefore, acceptance remains laggard. OBJECTIVE This research aims to investigate the key acceptance factors and value-adding use cases of a virtual assistant for patients diagnosed with cancer. METHODS Qualitative interviews with eight former patients and four doctors of a Dutch radiotherapy institute were conducted to establish an appropriate theoretical acceptance model and gain insights into possible use cases. The unified theory of acceptance and use of technology (UTAUT) was used to systematise perceptions and was inductively modified as a result of the interviews. The subsequent research model was triangulated via an online survey with 127 respondents diagnosed with cancer. A structural equation model was used to determine the relevance of acceptance factors. Through a multigroup analysis, differences between sample subgroups were compared. RESULTS During the interviews, support was found for the four factors of the UTAUT. Additionally, three factors, self-efficacy, trust, and resistance to change, were added as an extension of the UTAUT. Participants found a virtual assistant helpful in receiving information about logistic questions, treatment procedures, side effects, or scheduling appointments. The quantitative study found that the constructs performance expectancy (ß=.399), effort expectancy (ß=.258), social influence (ß=.114), and trust (ß=.210) significantly influenced behavioural intention to use a virtual assistant, explaining 80% of its variance. Self-efficacy (ß=.792) acts as antecedent of effort expectancy. Facilitating conditions and resistance to change were not found to have a significant relationship with user intention. No different relevance of factors was found between subgroups of age, gender, and experience. However, descriptive statistics indicate a higher intention of younger patients to use a virtual assistant. CONCLUSIONS Performance and effort expectancy are the leading determinants of virtual assistant acceptance. The latter is dependent on a patient’s self-efficacy, highlighting the need for close involvement during the development and introduction. The high relevance of trust indicates the need for a reliable, secure service that should be promoted as such. Social influence suggests using doctors in endorsing the abovementioned characteristics. CLINICALTRIAL N/A

show abstract

Designing ECAs to Improve Robustness of Human-Machine Dialogue

Cited by 4 publications

References 43 publications

A Digital Counselor-Delivered Intervention for Substance Use Among People With HIV: Development and Usability Study

A Digital Counselor-Delivered Intervention for Substance Use Among People With HIV: Development and Usability Study

Analyzing the determinants to accept a virtual assistant and use cases among cancer patients: a mixed methods study

Analysing the determinants to accept a virtual assistant and use cases among cancer patients: a mixed methods study (Preprint)

Contact Info

Product

Resources

About