Towards AI-Driven Healthcare: Systematic Optimization, Linguistic Analysis, and Clinicians' Evaluation of Large Language Models for Smoking Cessation Interventions.

Creating intervention messages for smoking cessation is a labor-intensive process. Advances in Large Language Models (LLMs) offer a promising alternative for automated message generation. Two critical questions remain: 1) How to optimize LLMs to mimic human expert writing, and 2) Do LLM-generated messages meet clinical standards? We systematically examined the message generation and evaluation processes through three studies investigating prompt engineering (Study 1), decoding optimization (Study 2), and expert review (Study 3). We employed computational linguistic analysis in LLM assessment and established a comprehensive evaluation framework, incorporating automated metrics, linguistic attributes, and expert evaluations. Certified tobacco treatment specialists assessed the quality, accuracy, credibility, and persuasiveness of LLM-generated messages, using expert-written messages as the benchmark. Results indicate that larger LLMs, including ChatGPT, OPT-13B, and OPT-30B, can effectively emulate expert writing to generate well-written, accurate, and persuasive messages, thereby demonstrating the capability of LLMs in augmenting clinical practices of smoking cessation interventions.

The Effects of Embedded Skin Cancer Interventions on Sun-Safety Attitudes and Attention Paid to Tan Women on Instagram.

Background and Objectives: Because of high skin cancer risks for young women, it is vital that effective interventions reach and influence this demographic. Visual social media platforms, like Instagram, are popular with young women and are an appropriate intervention site; yet, they also host competing images idealizing tan skin. The present study tested the ability of digital sun-safety interventions to affect self-control-related emotions and visual attention to subsequent tan-ideal images as well as sun-safety attitudes. Methods: Women were recruited from a large public Mid-Atlantic university in the United States. Participants (N = 120) were randomly assigned to view an appearance benefits intervention, a self-control emotions intervention, or a control message, each designed to look like an Instagram sponsored story. After self-reporting self-compassion and anticipated pride, participants then viewed seven pairs of Instagram posts featuring either tan or pale women while an eye tracker assessed visual attention. Finally, participants self-reported their responses to questions assessing sun-safety-related norms, efficacy, and attitudes. Results: A mixed design analysis of covariance revealed that women who first viewed the appearance benefits intervention story spent less time visually fixated on Instagram images of tan women than did those who viewed the self-control emotions intervention or control message (p = 0.005, η p 2 = 0.087). Regressions also revealed interactions between the intervention conditions and feelings of anticipated pride on both visual attention and sun-safety attitudes. Conclusion: Sponsored stories on Instagram can promote sun-safety attitudes, depending on the emotional responses they generate. Additionally, sponsored interventions can affect subsequent visual attention.

Social Media and Emotional Burnout Regulation During the COVID-19 Pandemic: Multilevel Approach.

BACKGROUND: In February 2020, the Chinese government imposed a complete lockdown of Wuhan and other cities in Hubei Province to contain a spike of COVID-19 cases. Although such measures are effective in preventing the spread of the virus, medical professionals strongly voiced a caveat concerning the pandemic emotional burnout at the individual level. Although the lockdown limited individuals' interpersonal communication with people in their social networks, it is common that individuals turn to social media to seek and share health information, exchange social support, and express pandemic-generated feelings. OBJECTIVE: Based on a holistic and multilevel perspective, this study examines how pandemic-related emotional exhaustion enacts intrapersonal, interpersonal, and hyperpersonal emotional regulation strategies, and then evaluates the effectiveness of these strategies, with a particular interest in understanding the role of hyperpersonal-level regulation or social media-based regulation. METHODS: Using an online panel, this study sampled 538 Chinese internet users from Hubei Province, the epicenter of the COVID-19 outbreak in China. Survey data collection lasted for 12 days from February 7-18, 2020, two weeks after Hubei Province was placed under quarantine. The sample had an average age of 35 (SD 10.65, range 18-78) years, and a majority were married (n=369, 68.6%). RESULTS: Using structural equation modeling, this study found that intrapersonal-level (B=0.22; ß=.24; P<.001) and interpersonal-level (B=0.35; ß=.49; P<.001) emotional regulation strategies were positively associated with individuals' outcome reappraisal. In contrast with intrapersonal and interpersonal regulations, hyperpersonal (social media-based) regulation strategies, such as disclosing and retweeting negative emotions, were negatively related to the outcome reappraisal (B=-1.00; ß=-.80; P<.001). CONCLUSIONS: Consistent with previous literature, intrapersonal-level regulation (eg, cognitive reappraisal, mindfulness, and self-kindness) and interpersonal-level supportive interaction may generate a buffering effect on emotional exhaustion and promote individuals' reappraisal toward the stressful situation. However, hyperpersonal-level regulation may exacerbate the experienced negative emotions and impede reappraisal of the pandemic situation. It is speculated that retweeting content that contains pandemic-related stress and anxiety may cause a digital emotion contagion. Individuals who share other people's negative emotional expressions on social media are likely to be affected by the negative affect contagion. More importantly, the possible benefits of intrapersonal and interpersonal emotion regulations may be counteracted by social media or hyperpersonal regulation. This suggests the necessity to conduct social media-based health communication interventions to mitigate the social media-wide negative affect contagion if lockdown policies related to highly infectious diseases are initiated.