The gut–liver axis has evolved from a mechanistic framework to a translational paradigm at the intersection of microbiology, immunology, and hepatology (Kim et al., 2024; Li et al., 2025). This bidirectional interface—linking microbial metabolism, mucosal immunity, and hepatic signaling—has redefined the pathophysiology of chronic liver diseases and catalyzed new therapeutic frontiers, from dietary modulation to fecal microbiota transplantation (FMT) (Tilg et al., 2022). However, as public interest rises, the translation of these scientific concepts into online health information often occurs without systematic oversight, increasing the risk of oversimplification and misinterpretation (Porcari et al., 2023).

Short-video platforms such as TikTok, Bilibili, and Xiaohongshu now constitute the world’s fastest-growing channels of biomedical diffusion, democratizing access to information while simultaneously eroding the boundary between expert knowledge and popular belief (Qin et al., 2024; (Monroe) Meng et al., 2024). Algorithmic optimization often prioritizes virality over veracity, elevating engagement metrics at the expense of informational rigor (Metzler and Garcia, 2024). In the context of microbiome-based therapies—particularly FMT, which traverses experimental, ethical, and regulatory domains—this dynamic is not merely a communication concern but a potential vector for clinical misinformation and behavioral risk (Hou et al., 2025).

Importantly, the gut–liver axis is especially prone to consumer-level distortion. Under the banner of “detox,” “microbiome resetting,” or “liver cleansing,” this scientific construct is frequently co-opted by marketers of functional beverages, probiotic supplements, and herbal remedies—often stripped of mechanistic accuracy or clinical grounding (Anand and Mande, 2022; Strand, 2022). These narratives borrow scientific legitimacy while bypassing regulatory scrutiny, contributing to overclaimed health benefits and microbiome-related misinformation. These trends highlight the need to evaluate how gut–liver–related information is presented within high-traffic digital ecosystems.

Despite the exponential growth of digital health research, limited attention has been paid to how algorithmic platforms translate scientific complexity into lay perceptions—or how engagement metrics such as likes, shares, and comments may inflate the perceived credibility of content regardless of its accuracy (De Santis et al., 2023). The gut–liver axis, situated at the intersection of translational science and online discourse, offers a compelling lens to interrogate this disconnect between digital visibility and biomedical validity (Tripathi et al., 2018; Lin et al., 2025). Yet no study has systematically assessed the quality, reliability, or platform-specific differences of short-form video content on this topic, representing a notable research gap.

To address this epistemic and ethical gap, we conducted a cross-platform, quality-of-information audit of short videos related to gut microbiota and liver disease. Using validated instruments (mDISCERN, JAMA benchmarks, GQS) alongside engagement-quality correlation analyses, we aimed to evaluate: (i) the overall quality and reliability of gut–liver axis content across major short-video platforms; (ii) differences among TikTok, Bilibili, and Xiaohongshu—three platforms with distinct content norms, audience demographics, and creator ecosystems; and (iii) whether engagement metrics align with or diverge from educational value (Mohammadzadeh Azarabadi et al., 2025). By clarifying these dimensions, the study provides empirical evidence to guide digital-health governance, expert participation, and public-facing microbiome communication.

Building on the statistical differences identified across the three platforms, with Bilibili showing consistently higher informational quality and TikTok generating substantially greater user engagement, our findings highlight a persistent disconnect between visibility and content quality. Our findings highlight a persistent disconnect between visibility and content quality in gut–liver axis–related short-form videos. Despite TikTok’s high engagement, it consistently ranked lowest across all educational indices. In contrast, Bilibili videos—characterized by longer duration and expert-driven production—achieved the highest quality scores (all p < 0.001) (Sun et al., 2025). Notably, engagement metrics showed no correlation with content quality, reinforcing that virality does not equal scientific rigor. The only content feature moderately associated with higher quality was video duration (r = 0.335, p < 0.01), suggesting that longer videos offer more room for context and citation—an often-overlooked factor on short-form platforms (van Kessel et al., 2022).

Subgroup analysis underscored the pivotal role of creator expertise in shaping both content quality and audience response. Videos produced by healthcare professionals or affiliated institutions consistently outperformed those from non-professionals across all evaluated domains, with average SUM scores of 9.03 versus 3.87, respectively (p < 0.001). Professionally authored videos also achieved higher levels of user approval, including likes and favorites, contradicting the notion that evidence-based content is less appealing (Kong et al., 2021). These findings refute the assumption that evidence-based content lacks appeal and underscore the dual value of professional contributors in ensuring both scientific accuracy and audience receptivity—particularly vital for complex topics like the gut–liver axis, where simplification may distort clinical understanding.

For instance, fecal microbiota transplantation (FMT), still under clinical evaluation for most liver-related conditions, is often misrepresented in short videos as a general “gut detox” tool—ignoring regulatory constraints, donor-screening protocols, and infection risks (Ma et al., 2025). Similarly, over-the-counter probiotics and so-called “gut-liver detox teas” are frequently marketed as cure-alls, supported by vague microbiome rhetoric unbacked by clinical evidence (Pan et al., 2025). This commodification of microbiome therapies undermines informed decision-making and may delay proper treatment.

Our findings challenge the presumed link between popularity and educational value. While engagement indicators—likes, comments, shares, and favorites—were highly intercorrelated, none aligned with video quality scores (Bortolotti, 2016). This suggests that algorithmic amplification often rewards emotional appeal over scientific substance. In many short-video ecosystems, algorithmic amplification prioritizes emotionally appealing or entertainment-oriented content, which suppresses informational depth by rewarding brief and attention-driven viewer responses. This dynamic tends to elevate highly engaging but educationally limited material. Longer videos function as an exception because their extended duration allows creators to provide context, incorporate evidence, and construct more coherent explanations. This structural flexibility helps clarify the moderate positive association between video duration and educational quality observed in our analysis, despite the broader algorithmic tendency to favor immediacy over informational rigor. Notably, longer duration emerged as the only quantitative predictor of quality, reinforcing the importance of temporal space for conveying complex health information (Mercer Moss et al., 2016).

Platform-level differences in user demographics and engagement behaviors may also influence visibility patterns. Social media platforms vary widely in terms of user base size, activity level, age distribution, educational background, and professional orientation. These factors may shape algorithmic amplification by determining which types of content generate rapid engagement signals. For example, platforms with younger or entertainment-oriented users may preferentially circulate visually stimulating short videos, whereas platforms with more academically oriented users may better support higher-quality educational content. Although user characteristics were not the primary focus of this study, acknowledging these differences provides important context for interpreting cross-platform disparities.

Correcting this imbalance requires coordinated interventions. At the creator level, incentives such as educational grants, CME credit, or institutional branding could encourage expert participation. Platforms should embed verifiable creator credentials—through medical badges, institutional tags, or tiered quality labels—and evolve algorithms to incorporate quality-weighted amplification. Recommendation systems designed to “serve societal good” have argued for integrating long-term value and content fidelity into feed logic (Lasser and Poechhacker, 2025; Milli et al., 2025). In addition, scalable safeguards—such as real-time peer-review tagging systems analogous to academic article review—may provide a viable path to maintain diversity without sacrificing reliability.

These efforts should be coupled with user-focused strategies to bolster digital health literacy. Campaigns addressing cognitive biases, trust heuristics, and information hygiene can improve public resilience against misinformation (Yuen et al., 2024; Ziapour et al., 2024). This is particularly urgent in microbiome and metabolic health, where patient decisions hinge on accurate understanding of multisystemic interactions. To our knowledge, this is the first multi-dimensional, cross-platform audit of gut–liver axis–related science videos using validated quality instruments. Unlike prior studies focusing on a single platform or disease, our work offers a broader lens into how platform dynamics, creator identity, and algorithmic visibility co-shape digital health narratives. These insights hold direct implications for hepatology, microbiome-based interventions, and translational education—domains where public understanding is critical to clinical uptake (Schweitzer et al., 2025).

Nonetheless, some limitations should be acknowledged. First, the cross-sectional design captures only a single timepoint, making it difficult to assess how content quality may evolve over time or in response to policy changes. Second, although we used validated scoring tools and ensured dual independent review, some degree of subjectivity is inevitable in quality assessment (Gijsen et al., 2020). Third, the study included only Chinese-language content from three major platforms (TikTok, Bilibili, and Xiaohongshu) and did not incorporate other ecosystems such as WeChat Video Accounts or YouTube, which may operate under different governance structures and user demographics. Fourth, all data reflect platform behavior as of October 2025; subsequent algorithmic updates may influence content visibility, engagement dynamics, and creator practices. Finally, cross-linguistic and cross-cultural variations were not evaluated, which may limit the generalizability of our findings to regions beyond the Mandarin-speaking context. Future research incorporating multilingual and cross-cultural analyses would provide a more comprehensive understanding of these dynamics.

Looking ahead, future research could explore several promising directions. AI-based tools may help detect misinformation and track semantic drift in real time. Longitudinal studies could shed light on how platform-level content changes in response to shifting algorithms or public health events. In addition, experimental studies might evaluate whether expert verification, visible credibility labels, or algorithm-level nudges can improve the visibility and trustworthiness of high-quality content. Ultimately, interdisciplinary collaboration—among clinicians, data scientists, platform designers, and regulators—will be key to aligning online health information ecosystems with public health priorities.

Beyond the practical considerations of platform design and governance, our findings point to a deeper epistemic challenge that shapes scientific communication in algorithmic environments. When relevance and visibility are driven by engagement metrics rather than informational rigor, public understanding of complex biomedical concepts becomes filtered through attention-oriented curation. This process not only influences what audiences encounter but also affects how scientific knowledge is prioritized, simplified, or marginalized within digital spaces. Addressing these epistemic distortions will require interdisciplinary collaboration to ensure that insights related to the gut–liver axis and biomedical science more broadly can circulate in a manner that preserves complexity, contextual integrity, and translational relevance.

In conclusion, this cross-platform analysis demonstrates a clear disconnect between the popularity and educational quality of short-form videos related to the gut–liver axis. While content from healthcare professionals consistently exhibited higher informational value, engagement metrics failed to reflect this, highlighting the limitations of algorithm-driven visibility. To improve digital health communication, platforms should prioritize content quality through expert verification and algorithmic adjustments, while parallel efforts to enhance public digital literacy are crucial for empowering evidence-based health decisions in the digital age.