Key changes to improve social presence of a virtual health assistant promoting colorectal cancer screening informed by a technology acceptance model

Embodied Conversational Agents for Chronic Diseases: Scoping Review

BACKGROUND

Embodied conversational agents (ECAs) are computer-generated animated humanlike characters that interact with users through verbal and nonverbal behavioral cues. They are increasingly used in a range of fields, including health care.

OBJECTIVE

This scoping review aims to identify the current practice in the development and evaluation of ECAs for chronic diseases.

METHODS

We applied a methodological framework in this review. A total of 6 databases (ie, PubMed, Embase, CINAHL, ACM Digital Library, IEEE Xplore Digital Library, and Web of Science) were searched using a combination of terms related to ECAs and health in October 2023. Two independent reviewers selected the studies and extracted the data. This review followed the PRISMA-ScR (Preferred Reporting Items of Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) statement.

RESULTS

The literature search found 6332 papers, of which 36 (0.57%) met the inclusion criteria. Among the 36 studies, 27 (75%) originated from the United States, and 28 (78%) were published from 2020 onward. The reported ECAs covered a wide range of chronic diseases, with a focus on cancers, atrial fibrillation, and type 2 diabetes, primarily to promote screening and self-management. Most ECAs were depicted as middle-aged women based on screenshots and communicated with users through voice and nonverbal behavior. The most frequently reported evaluation outcomes were acceptability and effectiveness.

CONCLUSIONS

This scoping review provides valuable insights for technology developers and health care professionals regarding the development and implementation of ECAs. It emphasizes the importance of technological advances in the embodiment, personalized strategy, and communication modality and requires in-depth knowledge of user preferences regarding appearance, animation, and intervention content. Future studies should incorporate measures of cost, efficiency, and productivity to provide a comprehensive evaluation of the benefits of using ECAs in health care.

Factors Influencing the Acceptability, Acceptance, and Adoption of Conversational Agents in Health Care: Integrative Review

BACKGROUND

Conversational agents (CAs), also known as chatbots, are digital dialog systems that enable people to have a text-based, speech-based, or nonverbal conversation with a computer or another machine based on natural language via an interface. The use of CAs offers new opportunities and various benefits for health care. However, they are not yet ubiquitous in daily practice. Nevertheless, research regarding the implementation of CAs in health care has grown tremendously in recent years.

OBJECTIVE

This review aims to present a synthesis of the factors that facilitate or hinder the implementation of CAs from the perspectives of patients and health care professionals. Specifically, it focuses on the early implementation outcomes of acceptability, acceptance, and adoption as cornerstones of later implementation success.

METHODS

We performed an integrative review. To identify relevant literature, a broad literature search was conducted in June 2021 with no date limits and using all fields in PubMed, Cochrane Library, Web of Science, LIVIVO, and PsycINFO. To keep the review current, another search was conducted in March 2022. To identify as many eligible primary sources as possible, we used a snowballing approach by searching reference lists and conducted a hand search. Factors influencing the acceptability, acceptance, and adoption of CAs in health care were coded through parallel deductive and inductive approaches, which were informed by current technology acceptance and adoption models. Finally, the factors were synthesized in a thematic map.

RESULTS

Overall, 76 studies were included in this review. We identified influencing factors related to 4 core Unified Theory of Acceptance and Use of Technology (UTAUT) and Unified Theory of Acceptance and Use of Technology 2 (UTAUT2) factors (performance expectancy, effort expectancy, facilitating conditions, and hedonic motivation), with most studies underlining the relevance of performance and effort expectancy. To meet the particularities of the health care context, we redefined the UTAUT2 factors social influence, habit, and price value. We identified 6 other influencing factors: perceived risk, trust, anthropomorphism, health issue, working alliance, and user characteristics. Overall, we identified 10 factors influencing acceptability, acceptance, and adoption among health care professionals (performance expectancy, effort expectancy, facilitating conditions, social influence, price value, perceived risk, trust, anthropomorphism, working alliance, and user characteristics) and 13 factors influencing acceptability, acceptance, and adoption among patients (additionally hedonic motivation, habit, and health issue).

CONCLUSIONS

This review shows manifold factors influencing the acceptability, acceptance, and adoption of CAs in health care. Knowledge of these factors is fundamental for implementation planning. Therefore, the findings of this review can serve as a basis for future studies to develop appropriate implementation strategies. Furthermore, this review provides an empirical test of current technology acceptance and adoption models and identifies areas where additional research is necessary.

TRIAL REGISTRATION

PROSPERO CRD42022343690; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=343690.

Application of machine learning for risky sexual behavior interventions among factory workers in China

Introduction

Assessing the likelihood of engaging in high-risk sexual behavior can assist in delivering tailored educational interventions. The objective of this study was to identify the most effective algorithm and assess high-risk sexual behaviors within the last six months through the utilization of machine-learning models.

Methods

The survey conducted in the Longhua District CDC, Shenzhen, involved 2023 participants who were employees of 16 different factories. The data was collected through questionnaires administered between October 2019 and November 2019. We evaluated the model's overall predictive classification performance using the area under the curve (AUC) of the receiver operating characteristic (ROC) curve. All analyses were performed using the open-source Python version 3.9.12.

Results

About a quarter of the factory workers had engaged in risky sexual behavior in the past 6 months. Most of them were Han Chinese (84.53%), hukou in foreign provinces (85.12%), or rural areas (83.19%), with junior high school education (55.37%), personal monthly income between RMB3,000 (US$417.54) and RMB4,999 (US$695.76; 64.71%), and were workers (80.67%). The random forest model (RF) outperformed all other models in assessing risky sexual behavior in the past 6 months and provided acceptable performance (accuracy 78%; sensitivity 11%; specificity 98%; PPV 63%; ROC 84%).

Discussion

Machine learning has aided in evaluating risky sexual behavior within the last six months. Our assessment models can be integrated into government or public health departments to guide sexual health promotion and follow-up services.

Can virtual human clinicians help close the gap in colorectal cancer screening for rural adults in the United States? The influence of rural identity on perceptions of virtual human clinicians

Rural adults experience disparities in colorectal cancer screening, a trend even more distinct among rural Black adults. Healthcare disruptions caused by COVID-19 exacerbated inequities, heightening attention on virtual communication strategies to increase screening. Yet little is known about how rural adults perceive virtual human clinicians (VHCs). Given that identifying as rural influences perceived source credibility often through appearance judgments, the goal of this pilot was to explore how to develop VHCs that individuals highly identified with rurality find attractive. Between November 2018 and April 2019, we tested a culturally tailored, VHC-led telehealth intervention delivering evidence-based colorectal cancer prevention education with White and Black adults (N = 2079) in the United States recruited through an online panel who were non-adherent to screening guidelines and between 50 and 73 years of age. Participants were randomized on three factors (VHC race-matching, VHC gender-matching, Intervention type). Ordinal logistic regression models examined VHC appearance ratings. Participants with a high rural identity (AOR = 1.12, CI = [1.02, 1.23], p =.02) rated the VHCs more attractive. High rural belonging influenced VHC attractiveness for Black participants (AOR = 1.22, CI = [1.03, 1.44], p =.02). Also, Black participants interacting with a Black VHC and reporting high rural self-concept rated the VHC as more attractive (AOR = 2.22, CI = [1.27, 3.91], p =.01). Findings suggest adults for whom rural identity is important have more positive impressions of VHC attractiveness. For patients with strong rural identities, enhancing VHC appearance is critical to tailoring colorectal cancer prevention interventions.

Communicating risk to promote colorectal cancer screening: a multi-method study to test tailored versus targeted message strategies

Colorectal cancer (CRC) screening rates are suboptimal, partly due to poor communication about CRC risk. More effective methods are needed to educate patients, but little research has examined best practices for communicating CRC risk. This multi-method study tests whether tailoring CRC risk information increases screening intentions. Participants (N = 738) were randomized with a 2:2:1 allocation to tailored, targeted, and control message conditions. The primary outcome was intention to screen for CRC (yes/no). Additional variables include perceived message relevance, perceived susceptibility to CRC, and free-text comments evaluating the intervention. A chi-square test determined differences in the proportion of participants who intended to complete CRC screening by condition. A logistic-based path analysis explored mediation. Free-text comments were analyzed using advanced topic modeling analysis. CRC screening intentions were highest in the tailored intervention and significantly greater than control (P = 0.006). The tailored message condition significantly increased message relevance compared with control (P = 0.027) and targeted conditions (P = 0.002). The tailored condition also increased susceptibility (P < 0.001) compared with control, which mediated the relationship between the tailored condition and intention to screen (b = 0.04, SE = 0.02, 95% confidence interval = 0.02, 0.09). The qualitative data reflect similar trends. The theoretical mechanisms and practical implications of tailoring health education materials about CRC risk are discussed.