Journal of Medical Internet Research - International Scientific Journal for Medical Research, Information and Communication on the Internet #Preprint #PeerReviewMe: Warning: This is a unreviewed preprint. Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn. Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period. Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author). Background: Previous studies suggest that patient's online health information seeking affects medical consultations and patient-doctor relationships. An up-to-date picture of patients’ online health information seeking behaviors can inform and prepare frontline healthcare professionals to collaborate, facilitate or empower patients to access and manage health information online. Objective: This study explores the prevalence, patterns, and predictors of online health information seeking behaviors among primary care patients in Hong Kong, and the relations between online health information seeking and eHealth literacy. Methods: Patients attending a university primary care clinic in Hong Kong were asked to complete a questionnaire survey on their demographic backgrounds; health status; frequency and pattern of online health information seeking; contents, sources, and reasons of online health information seeking, and their eHealth literacy. eHealth literacy was measured by the validated eHealth Literacy Scale (eHEALS). Regression analyses explored various demographic and behavioral predictors to online health information seeking, and predictors to eHealth literacy. Results: 97.3% (1162/1194) respondents used the Internet, of which 87.4% (1016/1162) had used the Internet to find health information. Most respondents (66.0%, 665/1008) searched once monthly or more. Few (26.7%, 271/1016) asked their doctor about health information found online, but most doctors (56.1%, 152/271) showed little or no interest at all. The most sought topic was symptom (81.6%, 829/1016), the top reason was noticing new symptoms or change in health (70.1%, 712/1016), the most popular source is online encyclopedia (70.0%, 711/1016), and the top reason for choosing a source was convenience (55.4%, 563/1016). Poisson regression analysis identified high eHEALS score, fair or poor self-rated health, having chronic medical condition, and using the Internet several times a day as significant predictors to online health information seeking. Multiple regression analysis identified lower age, better self-rated health, more frequent Internet use, more frequent online health information seeking, and more types of health information sought as significant predictors to higher eHealth literacy. Conclusions: OHIS is prevalent among primary care patients in Hong Kong, but only a minority shared the information with doctors. Websites were chosen more for convenience than for accuracy or authoritativeness. Doctors should recognize, facilitate and empower patients for OHIS.

BACKGROUND Previous studies have suggested that patients’ online health information seeking affects their medical consultations and patient-doctor relationships. An up-to-date picture of patients’ online health information-seeking behaviors can inform and prepare frontline health care professionals to collaborate, facilitate, or empower their patients to access and manage health information found online. OBJECTIVE This study explores the prevalence, patterns, and predictors of online health information-seeking behaviors among primary care patients in Hong Kong, and the relations between online health information seeking and electronic health (eHealth) literacy. METHODS Patients attending a university primary care clinic in Hong Kong were asked to complete a questionnaire survey on their demographic backgrounds; health status; frequency and pattern of online health information seeking; contents, sources, and reasons for online health information seeking; and their eHealth literacy. eHealth literacy was measured by the validated eHealth Literacy Scale (eHEALS). Regression analyses explored various demographic and behavioral predictors to online health information seeking, and predictors to eHealth literacy. RESULTS In all, 97.32% (1162/1194) respondents used the internet, of which 87.44% (1016/1162) had used the internet to find health information. Most respondents (65.97%, 665/1008) searched once monthly or more. Few (26.88%, 271/1008) asked their doctor about health information found online, but most doctors (56.1%, 152/271) showed little or no interest at all. The most sought topic was symptom (81.59%, 829/1016), the top reason was noticing new symptoms or change in health (70.08%, 712/1016), the most popular source was online encyclopedia (69.98%, 711/1016), and the top reason for choosing a source was convenience (55.41%, 563/1016). Poisson regression analysis identified high eHEALS score, fair or poor self-rated health, having a chronic medical condition, and using the internet several times a day as significant predictors of online health information seeking. Multiple regression analysis identified lower age, better self-rated health, more frequent internet use, more frequent online health information seeking, and more types of health information sought as significant predictors to higher eHealth literacy. CONCLUSIONS Online health information seeking is prevalent among primary care patients in Hong Kong, but only a minority shared the information with doctors. Websites were chosen more for convenience than for accuracy or authoritativeness. Doctors should recognize patients’ online health information-seeking behavior, and facilitate and empower them to search for high-quality online health information.

BACKGROUND As the population of survivors of cancer grows and new information technologies become widespread in Hong Kong, survivors of cancer increasingly seek ongoing care and support information from large language models (LLMs). While these tools provide immediate conversational responses, they carry substantial risks of generating inaccurate, unsafe, or generic medical advice. Evaluating LLM response performance is particularly critical in Hong Kong’s bilingual health care context. OBJECTIVE This study aimed to evaluate the clinical utility, safety, understandability, and readability of 4 popular commercial LLMs answering cancer survivorship questions in English and Traditional Chinese and to explore whether a structured prompting strategy improves response performance. METHODS We conducted a 2-phase evaluation study combining quantitative scoring with qualitative classification of unsafe responses. In phase 1, 25 expert-curated questions spanning 5 survivorship domains were input into 4 LLMs: Google Gemini 3.1 Pro, HK Chat-0.6.2, DeepSeek-V3.2, and Kimi-K2.5. A Delphi expert panel evaluated the baseline responses for clinical utility using a 5-point scale and assessed safety via binary categorization. Understandability was measured using the Patient Education Materials Assessment Tool (PEMAT). Readability was assessed via the Flesch-Kincaid Grade Level for English and lexical richness for Traditional Chinese. In phase 2, experts developed an optimized context, role, audience, format, task, and tone (CRAFT) prompt, and the resulting LLM responses were compared with baseline outputs in an exploratory within-sample analysis using paired statistical tests. RESULTS Baseline evaluations identified Google Gemini 3.1 Pro as the most clinically useful model in English (mean 3.99/5, SD 0.16) and Traditional Chinese (mean 4.21/5, SD 0.20). Unconstrained models demonstrated language-dependent safety vulnerabilities: English errors (up to 24%) centered on service mismatches and definitive interpretations, whereas Traditional Chinese errors (up to 20%) involved prescriptive disease management and unverified adjunctive therapies. Within each language, the models producing the most readable text were not those achieving the highest clinical utility; this relationship was not formally tested. Applying the CRAFT prompt to Gemini 3.1 Pro significantly improved clinical utility across both languages (English: mean 4.68, SD 0.18, P<.001; Traditional Chinese: 4.66, SD 0.23, P<.001). After prompting, the 2 unsafe English baseline responses were reclassified as safe (English 23/25 to 25/25 safe; Traditional Chinese 25/25 at both time points). This transition was not statistically significant on a post hoc exact McNemar test (P=.50) and is reported as descriptive. CONCLUSIONS LLMs exhibited language-divergent clinical safety vulnerabilities when answering cancer survivorship queries, and within each language, the most readable models were not the most clinically useful. In an exploratory within-sample comparison, expert-designed CRAFT prompting improved clinical utility and corrected the unsafe outputs seen at baseline without reducing understandability. Safe deployment of LLMs in multilingual oncology contexts requires strict validation combined with structured, role-restrictive constraints. CLINICALTRIAL

Introduction Intersectionality, as applied to health, provides a framework for understanding how overlapping social identities, such as ethnicity, gender and socioeconomic status, shape differential health outcomes and healthcare experiences. Individuals who occupy multiple marginalised identities often experience compounded disadvantages. Ethnic minority (EM) populations, defined here as social groups who are numerically smaller and/or socially marginalised within a given national or regional context based on ethnicity, race, culture, language, ancestry, or related heritage (often overlapping with racialised identities), frequently experience systemic exclusion, racism and structural barriers that also contribute to persistent disparities in morbidity, mortality and healthcare access. However, much health research relies on single-axis analyses, which can obscure within-group variation and may inadvertently reinforce inequities. Despite growing recognition of intersectionality, its empirical application to EM health remains limited. This systematic review aims to synthesise evidence on how intersectionality theory has been conceptualised and applied to understand health outcomes and healthcare utilisation among EM populations globally. Methods and analysis This systematic review will follow Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and the PRISMA-Equity Extension. Comprehensive searches of peer-reviewed databases and grey literature, including MEDLINE, EMBASE, CINAHL Plus, Scopus, Web of Science and ProQuest will identify quantitative, qualitative and mixed-method empirical studies published from 2000 to 2026. Eligible studies must implement intersectionality theory to EM health or healthcare utilisation outcomes. Two independent reviewers will screen titles, abstracts and full-text articles. Data extraction will also be conducted independently and discrepancies will be discussed with input from a third reviewer. Guided by the PROGRESS-Plus equity framework, findings will be synthesised thematically, implementing a mixed-method convergent integrated synthesis design. Ethics and dissemination This review only analyses previously published data, thus it does not require ethical approval. Findings will be presented at relevant academic or professional conferences and submitted for publication in a peer-reviewed journal. PROSPERO registration number CRD420251248887.

Background: As the population of survivors of cancer grows and new information technologies become widespread in Hong Kong, survivors of cancer increasingly seek ongoing care and support information from large language models (LLMs). While these tools provide immediate conversational responses, they carry substantial risks of generating inaccurate, unsafe, or generic medical advice. Evaluating LLM response performance is particularly critical in Hong Kong’s bilingual health care context. Objective: This study aimed to evaluate the clinical utility, safety, understandability, and readability of 4 popular commercial LLMs answering cancer survivorship questions in English and Traditional Chinese and to explore whether a structured prompting strategy improves response performance. Methods: We conducted a 2-phase evaluation study combining quantitative scoring with qualitative classification of unsafe responses. In phase 1, 25 expert-curated questions spanning 5 survivorship domains were input into 4 LLMs: Google Gemini 3.1 Pro, HK Chat-0.6.2, DeepSeek-V3.2, and Kimi-K2.5. A Delphi expert panel evaluated the baseline responses for clinical utility using a 5-point scale and assessed safety via binary categorization. Understandability was measured using the Patient Education Materials Assessment Tool (PEMAT). Readability was assessed via the Flesch-Kincaid Grade Level for English and lexical richness for Traditional Chinese. In phase 2, experts developed an optimized context, role, audience, format, task, and tone (CRAFT) prompt, and the resulting LLM responses were compared with baseline outputs in an exploratory within-sample analysis using paired statistical tests. Results: Baseline evaluations identified Google Gemini 3.1 Pro as the most clinically useful model in English (mean 3.99/5, SD 0.16) and Traditional Chinese (mean 4.21/5, SD 0.20). Unconstrained models demonstrated language-dependent safety vulnerabilities: English errors (up to 24%) centered on service mismatches and definitive interpretations, whereas Traditional Chinese errors (up to 20%) involved prescriptive disease management and unverified adjunctive therapies. Within each language, the models producing the most readable text were not those achieving the highest clinical utility; this relationship was not formally tested. Applying the CRAFT prompt to Gemini 3.1 Pro significantly improved clinical utility across both languages (English: mean 4.68, SD 0.18, P<.001; Traditional Chinese: 4.66, SD 0.23, P<.001). After prompting, the 2 unsafe English baseline responses were reclassified as safe (English 23/25 to 25/25 safe; Traditional Chinese 25/25 at both time points). This transition was not statistically significant on a post hoc exact McNemar test (P=.50) and is reported as descriptive. Conclusions: LLMs exhibited language-divergent clinical safety vulnerabilities when answering cancer survivorship queries, and within each language, the most readable models were not the most clinically useful. In an exploratory within-sample comparison, expert-designed CRAFT prompting improved clinical utility and corrected the unsafe outputs seen at baseline without reducing understandability. Safe deployment of LLMs in multilingual oncology contexts requires strict validation combined with structured, role-restrictive constraints.