Young Adult Perspectives on Artificial Intelligence–Based Medication Counseling in China: Discrete Choice Experiment (Preprint)

Abstract

<sec> <title>BACKGROUND</title> As artificial intelligence (AI) permeates the current society, the young generation is becoming increasingly accustomed to using digital solutions. AI-based medication counseling services may help people take medications more accurately and reduce adverse events. However, it is not known which AI-based medication counseling service will be preferred by young people. </sec> <sec> <title>OBJECTIVE</title> This study aims to assess young people’s preferences for AI-based medication counseling services. </sec> <sec> <title>METHODS</title> A discrete choice experiment (DCE) approach was the main analysis method applied in this study, involving 6 attributes: granularity, linguistic comprehensibility, symptom-specific results, access platforms, content model, and costs. The participants in this study were screened and recruited through web-based registration and investigator visits, and the questionnaire was filled out online, with the questionnaire platform provided by Questionnaire Star. The sample population in this study consisted of young adults aged 18-44 years. A mixed logit model was used to estimate attribute preference coefficients and to estimate the willingness to pay (WTP) and relative importance (RI) scores. Subgroups were also analyzed to check for heterogeneity in preferences. </sec> <sec> <title>RESULTS</title> In this analysis, 340 participants were included, generating 8160 DCE observations. Participants exhibited a strong preference for receiving 100% symptom-specific results (β=3.18, 95% CI 2.54-3.81; &lt;i&gt;P&lt;/i&gt;&amp;lt;.001), and the RI of the attributes (RI=36.99%) was consistent with this. Next, they showed preference for the content model of the video (β=0.86, 95% CI 0.51-1.22; &lt;i&gt;P&lt;/i&gt;&amp;lt;.001), easy-to-understand language (β=0.81, 95% CI 0.46-1.16; &lt;i&gt;P&lt;/i&gt;&amp;lt;.001), and when considering the granularity, refined content was preferred over general information (β=0.51, 95% CI 0.21-0.8; &lt;i&gt;P&lt;/i&gt;&amp;lt;.001). Finally, participants exhibited a notable preference for accessing information through WeChat applets rather than websites (β=0.66, 95% CI 0.27-1.05; &lt;i&gt;P&lt;/i&gt;&amp;lt;.001). The WTP for AI-based medication counseling services ranked from the highest to the lowest for symptom-specific results, easy-to-understand language, video content, WeChat applet platform, and refined medication counseling. Among these, the WTP for 100% symptom-specific results was the highest (¥24.01, 95% CI 20.16-28.77; US $1=¥7.09). High-income participants exhibited significantly higher WTP for highly accurate results (¥45.32) compared to low-income participants (¥20.65). Similarly, participants with higher education levels showed greater preferences for easy-to-understand language (¥5.93) and video content (¥12.53). </sec> <sec> <title>CONCLUSIONS</title> We conducted an in-depth investigation of the preference of young people for AI-based medication counseling services. Service providers should pay attention to symptom-specific results, support more convenient access platforms, and optimize the language description, content models that add multiple digital media interactions, and more refined medication counseling to develop AI-based medication counseling services. </sec>

Ähnliche Arbeiten