Mapping Indonesian Public Discourse on Oral Health: A Content Analysis of Social Media Using Latent Dirichlet Allocation Topic Modelling Authors: Rosa Amalia, Iwan Dewanto, Anggit Wirasto, Leny Pratiwi Arie Sandy Published: 2026 ABSTRACT BackgroundThe potential of social media that offers public health insights in middle-income countries remains underexplored. This study examines patterns of oral health discourse on Indonesian social media and highlighting such insights into promotive and preventive oral health strategies.MethodsA discourse and content analysis was conducted on user-generated posts from Quora and Threads over an 18-month period (July 1, 2023-January 31, 2025). Posts were retrieved through automated data extraction and computational text retrieval based on keyword searches and coded into 19 predefined categories. Latent Dirichlet allocation (LDA) was applied to identify underlying thematic structures. Associations between platforms and discussion topics were tested using chi-square and Monte Carlo simulations. Comparative analyses examined platform-specific trends, thematic frequencies and inquiry types, with findings cross-referenced against national survey data.ResultsA total of 858 posts (541 from Quora, 317 from Threads) were analysed. The most frequently discussed issues were mouth ulcers and toothache. LDA revealed 10 dominant themes, including oral hygiene practices, children’s dental visits, gum swelling and postsurgical recovery. Quora discussions were dominated by symptom-focused inquiries, whereas Threads contained more personal and emotionally expressive narratives. Statistical testing confirmed a significant association between platform type and topics discussed (P < .001).ConclusionThis study demonstrated the utility of automated data extraction and LDA-based thematic modelling in uncovering community-level oral health patterns. Social media provides an immediate context-relevant perspective for monitoring oral health needs, supporting targeted communication strategies and responsive public health interventions.Clinical relevanceThis study introduces a complementary approach to conventional epidemiological monitoring. The findings can guide policymakers and practitioners in developing targeted preventive initiatives and public engagement strategies for improved oral health outcomes. Introduction Oral health is essential to general well-being, yet it often receives less attention in public health priorities. The World Health Organization (WHO), in its 2022 Global Oral Health Status Report, underscored the urgent need to situate oral health within the broader global health agenda to achieve universal health coverage. Indeed, the report indicates that approximately 3.5 billion individuals worldwide are affected by oral diseases. Among these, untreated dental caries in permanent teeth is the most prevalent condition.1 In Indonesia, recent data from the Basic Health Research (RISKESDAS) 2018 and the Indonesian Health Survey (SKI) 2023 reveal that a significant proportion of the population is affected by dental or oral health problems. The data indicates that more than half of the population experiences these issues, with percentages of 57.6% and 56.9% recorded in RISKESDAS and SKI, respectively. However, a comparatively low percentage of individuals affected by these problems seek care from dental professionals, with only 10.2% and 8.1% of those affected reporting that they had consulted a dentist in RISKESDAS and SKI, respectively.2,3 These statistics underscore the urgent need for more effective oral health strategies. Nevertheless, conventional survey methodologies often exhibit limitations in terms of scope and real-time insights, potentially resulting in the oversight of ‘under-voiced’ public concerns.4 In this context, the adoption of digital methodologies functions not merely as an alternative, but as a practical solution to these limitations. The accelerated proliferation of social media platforms offers a unique opportunity to confront these challenges. These platforms serve as repositories of user-generated content, thereby reflecting individuals' attitudes, behaviours and lived experiences related to oral health.5 Beyond functioning as communication channels, social media can also serve as research instruments. The sheer magnitude and uninterrupted flux of digital data empower passive monitoring, thereby enabling public health researchers to discern emergent trends, discern misinformation and appraise community sentiment, all without necessitating direct reporting from individuals.6 Real-time analysis of social media data offers a scalable and cost-efficient way to strengthen population health surveillance. Although oral health researchers have initiated an examination of online discourse concerning treatment expenses, hygiene practices and educational campaigns,7 this field continues to lag behind other areas of healthcare. Conversely, social media has been extensively utilised for the purpose of tracking the emergence of infectious diseases, monitoring vaccine hesitancy and evaluating mental health trends.8,9 This discrepancy is especially pronounced in middle-income countries such as Indonesia, where oral diseases are prevalent and conventional surveillance systems frequently encounter delays and inadequate coverage. Given the pervasive use of social media among various age demographics in Indonesia, a significant number of individuals are turning to digital platforms to disseminate information regarding symptoms, seek expert counsel and articulate their concerns.10 These online discussions provide timely insight into the dissemination of misinformation, care-seeking behaviours and unmet needs offering a strategic opportunity to translate digital data into more responsive and equitable oral health policies.11 To better understand how these dynamics, unfold in discussion-based environments, this study focuses on 2 platforms Quora and Threads that emphasise question-and-answer and narrative exchange. Although platforms such as Facebook, TikTok and Instagram have larger user bases in Indonesia.12 Quora and Threads were selected because they support more detailed, text-based discussions rather than brief or purely visual content. Although this choice reflects a specific demographic scope, it enables deeper exploration of user experiences and reasoning that may be less visible on visual-centred platforms. This study therefore aims to (1) to identify dominant oral health topics discussed on Quora and Threads (2) to compare discourse characteristics between platforms and (3) to evaluate the alignment of these discussions with national oral health indicators and global priorities. The integration of computational text analysis with public health perspectives is a novel approach that emphasises the potential of data-driven methodologies to enhance oral health promotion and disease prevention. Methods This study employed a content analysis integrating automated data extraction, qualitative coding and computational text modelling to examine patterns and themes in user-generated oral health discussions on social media. Publicly available content from Quora and Threads was systematically collected over an 18-month period (July 1, 2023-January 31, 2025) using ethical automated extraction techniques. While conceptually analogous to structured manual browsing, the process was programmatically implemented to ensure consistency, scalability and reproducibility across a large volume of posts. The study aimed to identify recurrent themes as well as platform-specific differences in oral health communication. A particular emphasis was placed on platform-specific architectures, encompassing multi-page navigation (pagination) and dynamically rendered content, which necessitates user-triggered interaction to attain full visibility. The extraction process was conducted using Python as the primary programming environment. The parsing and organisation of HTML content was facilitated by the utilisation of BeautifulSoup, thereby enabling the structured retrieval of data that is publicly accessible. In the context of Threads, where substantial content is dynamically loaded, the utilisation of Selenium was instrumental in simulating standard user interactions and ensuring the comprehensive capture of visible discussions.13 These tools were applied within a standardised and documented workflow to enhance procedural transparency and enable replication under comparable digital conditions. Data cleaning procedures were designed to strengthen analytic validity while maintaining clarity of individual-level discourse. Duplicate entries, incomplete posts and irrelevant materials were removed prior to analysis. Only original user-generated content was retained, while advertisements, automated bot-generated posts and reposted material were excluded to minimise noise and potential bias. Ethical safeguards guided all stages of the study. Data collection was limited strictly to publicly accessible content. Platform-specific guidelines and robots.txt directives were observed throughout the process to ensure responsible digital research conduct.14 No attempts were made to access restricted or private data, and all collected materials were anonymised prior to analysis, with no personally identifiable information retained. Posts were included in the analysis if they met the following criteria: (1) publicly accessible content from Threads and Quora; (2) contained relevant keywords related to oral health; (3) published within a predefined 18-month study period; and (4) consisted of original user-generated content, excluding advertisements, automated bot posts or reposted material. The study size was determined by the total number of eligible posts retrieved from Quora and Threads, ensuring that the full range of relevant discussions within the defined scope was captured. The data collection process was primarily supported by keyword searches, which were instrumental in ensuring the relevance and accuracy of the conversations concerning oral health. The development of these keywords was informed by a focus group discussion (FGD) with Dental Public Health experts, ensuring congruence with the study's objectives. The selected keywords included phrases such as ‘oral hygiene’, ‘dental cavities’, ‘gingivitis’, ‘dental appointments’ and ‘tooth decay’, among others. The implementation of keyword searches is imperative for ensuring the relevance and precision of the identification of discussions related to oral health. This collaborative approach was developed to minimise researcher bias in defining relevant oral health discourse. It also improved the precision of the search process, allowing for a more accurate and inclusive collection of public conversations on the topic. The selected keywords were subsequently integrated into automated data retrieval scripts using query parameters and regular expressions. This approach was implemented to ensure that only relevant posts were collected. The extracted data was systematically stored in a structured database (SQLite), a lightweight system for managing structured data. The database contained information such as postdate, platform and content including title, body text and comments. The structural design of the system facilitated the efficient querying and processing of data for subsequent analysis.15 Data preprocessing began with normalisation by converting all text to lowercase and removing special characters, symbols and URLs. Language detection was performed using the Python ‘langdetect’ library to filter posts to include only those in Indonesian language (Bahasa). Duplicate entries were identified and removed based on hashed content, and posts were manually reviewed to ensure relevance to the study’s focus on oral health. Cohen's Kappa statistic was used to assess inter-rater reliability between the 2 independent reviewers, achieving a score of 0.85, indicating high consistency.16 The data construction process (Table 1) began with an initial retrieval of 1596 posts, consisting of 758 from Quora and 838 from Threads. To ensure the quality of the analytic corpus, a 2-stage screening process was conducted. First, 295 posts were removed through automated cleaning, which included the exclusion of duplicate entries based on content hashing, automated bot-generated material and non-Indonesian text identified via the langdetect library. Second, a manual screening of the remaining 1301 posts was performed to ensure clinical relevance. During this stage, 443 posts were excluded because they lacked personal experiential narratives or used dental terms metaphorically (eg, ‘heartbreak feels worse than a toothache’). This rigorous process resulted in a final dataset of 858 posts (541 from Quora and 317 from Threads). The higher exclusion rate observed in Threads (62%) compared to Quora (28%) is attributed to the platform's more informal and conversational nature, which generated a greater proportion of posts using oral health terms in nonclinical or metaphorical contexts.Table 1Data construction and filtering stages of social media posts from Quora and Threads.Table 1 dummy alt textStage of filteringQuoraThreadsTotalInitial retrieval (raw data)7588381596Automated cleaning (duplicates, bots, non-Indonesian)−84−211−295Manual relevance screening (clinical relevance & metaphors)−133−310−443Analytic corpus541317858 Table 1Data construction and filtering stages of social media posts from Quora and Threads.Table 1 dummy alt textStage of filteringQuoraThreadsTotalInitial retrieval (raw data)7588381596Automated cleaning (duplicates, bots, non-Indonesian)−84−211−295Manual relevance screening (clinical relevance & metaphors)−133−310−443Analytic corpus541317858 Data construction and filtering stages of social media posts from Quora and Threads. To identify recurring themes within the data, a subset of the data was manually coded. Using NVivo 15 for the analysis of qualitative data, 19 primary categories were identified. The coding process was executed in stages, with regular feedback from Dental Public Health experts to ensure accuracy and consistency. The frequency of keywords within each theme was calculated and adjusted based on the total number of posts, with results visualised using Python libraries. The results were compiled into tables displaying the number of oral health complaint posts for each identified category. While the predefined coding categories captured expert-informed dimensions of oral health discourse, this approach remained limited by prior assumptions. To overcome potential researcher bias and explore emergent, data-driven structures beyond expert-defined categories, the study also applied latent Dirichlet allocation (LDA), a machine learning method used to uncover hidden themes in large text datasets. Unlike previous qualitative content analyses based on 19 predesigned codes, LDA allowed for an open-ended exploration within large textual datasets, without requiring predefined categories. This approach not only validated the initial coding but also helped surface new discussion themes that may have been overlooked. Similar approaches have been employed in recent health informatics research.17 The resulting topics were subsequently mapped to identify gaps between public discussions and current practices in oral health. The selection of 10 topics (k = 10) was determined through an iterative optimisation process. We evaluated multiple LDA models with the number of topics (k) ranging from 5 to 20. The final model (k = 10) was selected because it achieved the optimal Coherence Score (Cv), indicating the highest degree of semantic similarity between keywords within each topic while maintaining distinct boundaries between themes. Furthermore, we conducted a qualitative interpretability check in collaboration with Dental Public Health experts to ensure that each of the 10 topics represented a unique and meaningful clinical or behavioural dimension of oral health. The obtained results were compared with data from 2 official key sources Basic Health Survey (Riset Kesehatan Dasar/RISKESDAS) year 2018 and the Indonesian Health Survey (Survei Kesehatan Indonesia/SKI) year 2023 to provide a broader context and validate the findings from the social media analysis.2,3 Both surveys offer comprehensive insights into the health status of the Indonesian population, including detailed information on oral health. Finally, aggregated data were presented in figure summarising the number of posts per category across Quora and Threads. To analyse whether the distribution of oral health topics varied significantly between the 2 platforms, a chi-square test of independence was conducted. The test’s assumptions were met as all expected cell counts were above the minimum required threshold of 5. To strengthen the reliability of the results and account for any limitations of conventional chi-square assumptions, a Monte Carlo simulation with 10,000 replicates was also performed. It is imperative to acknowledge that, given the nature of data collection through automated data extraction, there were no missing data points to address in this analysis. These methodological decisions align with current best practices in statistical analysis and have been supported by recent literature in the field.18 This study was granted an exemption from Ethics Committee of Faculty of Dentistry and Dental Hospital Universitas Gadjah Mada, under exemption number 21/UN1/KEP/FKG-RSGM/EC/2025, as the study relied solely on publicly accessible social media data and did not include any interaction with human subjects or the collection of personally identifiable information. Results All posts that met the predefined inclusion criteria and passed the initial preprocessing stages were included in the analysis. The initial search for social media posts related to dental problems revealed that not all retrieved content held direct relevance to clinical dental health. Many entries turned out to be general discussions, promotional content, entertainment-related posts or personal reflections that lacked any specific mention of dental complaints. In this process, some posts that discussed clinical topics such as educational explanations about specific dental conditions or promoting oral health awareness were also identified. While these posts were informative, they did not contain personal complaints or direct experiential narratives, and thus were excluded. Accordingly, each post was subject to a manual review to ascertain its relevance. The meticulous filtration process of such content was imperative to guarantee that the final dataset represented authentic user expressions of dental health problems rather than general information or opinion sharing. Only posts that clearly expressed personal complaints related to dental health and had potential clinical relevance were included in the final analysis. This approach was necessary to better capture public expressions of dental health concerns and contribute to a deeper understanding of real-world issues faced by individuals. The process of extracting relevant data from social media posts requires a high level of careful and critical judgment, particularly given the large volume of posts that may appear relevant based on keywords but are actually unrelated to clinical issues. One example of this challenge can be seen in the keyword ‘toothache’. While the term was useful in identifying posts that might involve dental pain, many of the results were not clinically relevant. For example, some entries used ‘toothache’ metaphorically or within a figurative context, such as ‘I feel heartbroken; it feels worse than a toothache’, or described unrelated incidents like ‘My lips and teeth hurt after my phone hit my face while I was gaming’. These examples show how keyword-based searches can surface content that mentions dental symptoms but doesn’t reflect actual dental health concerns. Following the application of the inclusion and exclusion criteria, a total of 858 posts were subjected to analysis: 541 from Quora and 317 from Threads. A comparative analysis of the subjects addressed on these 2 platforms revealed distinct patterns in the mention of oral health issues, with significant variations in frequency across categories. The aforementioned findings are illustrated in Figure 1. The most prevalent subject of discourse across both platforms pertained to mouth ulcers, with 121 mentions on Quora compared to 138 mentions on Threads. This suggests that Threads has generated a greater level of interest in discussing mouth ulcers, possibly due to the platform's more casual tone, which encourages users to share personal experiences. The second most discussed issue was toothache, which received 159 mentions on Quora compared to 58 on Threads. This indicates a higher volume of discussion about toothaches on Quora, which may reflect the platform's focus on more in-depth, medical-oriented queries. By comparison, dental medication was mentioned 77 times on Quora, but only 10 times on Threads, highlighting a significantly lower emphasis on medical treatments on the latter platform.Fig. 1Mentions of oral health issues on Quora and Thread.Fig 1 dummy alt text Mentions of oral health issues on Quora and Thread. A comparative analysis of the discourse on Quora and Threads reveals that dental decay has been a recurring topic of discussion, with 52 mentions on Quora and 46 on Threads. This observation suggests that the prevalence of concern for this issue is comparable across both platforms. However, a closer examination reveals that Quora has seen slightly more engagement with this topic, suggesting a slightly higher level of discourse activity on the former platform. With regard to the subject of impacted teeth, a comparative analysis of mentions on Quora and Threads reveals a nearly equivalent level of interest across both platforms, with 20 and 21 mentions, respectively. When considering fewer common topics, periodontal issues received 15 mentions on Quora and only 8 on Threads. This discrepancy indicates that periodontal concerns are infrequently addressed on Threads, which may mirror the more informal nature of the platform. Conversely, Quora, with its more detailed and technical discussions, appears to attract inquiries on more specialised dental subjects. Themes concerning dental interventions, including postextraction care, received a total of 15 inquiries on Quora and 12 on Threads. Conversely, discourse on dental calculus was minimal, with a mere 8 mentions on Quora and 6 on Threads. Furthermore, fixed orthodontic treatments were referenced 10 times on Quora and 2 times on Threads, thereby underscoring the specialised nature of orthodontic concerns. Regarding children's oral health, tooth abscess in children was mentioned 37 times on Quora, with no mentions on Threads, while mouth ulcers in children appeared 3 times on Quora and also saw no mention on Threads. Topics like dental issues during pregnancy (4 mentions on Quora) and dental care for the elderly (1 mention on Quora) were less frequently discussed but did appear on Quora. Bruxism (teeth grinding) and dry mouth saw minimal mention, with bruxism mentioned once on Threads and dry mouth discussed 6 times on both Quora and Threads, indicating that these issues may be niche or less frequently raised in public forums. Latent Dirichlet Allocation (LDA) topic modelling was used to identify underlying themes in Indonesian-language social media posts about oral health. Because the content was written in informal (everyday Bahasa) the resulting topics reflect how people naturally talk about dental issues online. These conversations offer insight into common complaints, care-seeking behavior and public perceptions of oral health problems. The analysis produced 10 key topics, each represented by high-probability keywords (Figure 2). The figure shows the top 10 words for each topic, ranked by how strongly they contribute to that topic. All keywords are shown in Bahasa to preserve the original context. The interpretation of each of the 10 topic models is presented in detail below.1.Topic 0: Daily oral hygiene practicesDiscussions focused on routine brushing, especially before bed or on weekends. Common issues include a stinging sensation on the tongue, suggesting discomfort associated with brushing practices.2.Topic 1: Children’s dental appointmentThis topic captures reflects parents’ concerns about scheduling and the safety of dental visits for their children.3.Topic 2: Oral diseases and infectionsUsers discuss dental diseases, infections and their potential systemic effects. Tooth extraction appears as a common treatment response to advanced conditions.4.Topic 3: Home remedies for oral discomfortPosts mention simple self-care strategies such as saltwater rinses and vitamin supplements primarily used to address dry mouth or mild oral discomfort, often in children.5.Topic 4: Swollen gums and clinical interventionTalks about gum pain and swelling that often leads to clinic visits and extractions.6.Topic 5: Postsurgical bleeding and recoveryThis topic centres on experiences following oral surgery, especially wisdom tooth removal. Users commonly report bleeding, discomfort and recovery concerns.7.Topic 6: Microorganisms and oral-systemic healthUsers link viral or bacterial infections to oral conditions, often speculating about triggers such as food or beverages (eg, coffee), revealing lay interpretations of disease causality.8.Topic 7: Vitamins and nutritional deficiencyPosts emphasises the importance of nutrients for oral health, especially for children and older adults.9.Topic 8: Mouth ulcers and symptom managementMouth ulcers are a frequent concern, with users discussing triggers, such as food and drink, and self-medication practices aimed at symptom relief and healing.10.Topic 9: Medication use during pregnancyExpectant mothers express uncertainty around the use of medications for dental pain or nausea. Concerns focus on safety, signalling a need for clearer guidance.Fig. 2Topic distribution and salient terms from LDA-based topic modelling on social media discussions.Fig 2 dummy alt text 1.Topic 0: Daily oral hygiene practicesDiscussions focused on routine brushing, especially before bed or on weekends. Common issues include a stinging sensation on the tongue, suggesting discomfort associated with brushing practices. Topic 0: Daily oral hygiene practices Discussions focused on routine brushing, especially before bed or on weekends. Common issues include a stinging sensation on the tongue, suggesting discomfort associated with brushing practices. 2.Topic 1: Children’s dental appointmentThis topic captures reflects parents’ concerns about scheduling and the safety of dental visits for their children. Topic 1: Children’s dental appointment This topic captures reflects parents’ concerns about scheduling and the safety of dental visits for their children. 3.Topic 2: Oral diseases and infectionsUsers discuss dental diseases, infections and their potential systemic effects. Tooth extraction appears as a common treatment response to advanced conditions. Topic 2: Oral diseases and infections Users discuss dental diseases, infections and their potential systemic effects. Tooth extraction appears as a common treatment response to advanced conditions. 4.Topic 3: Home remedies for oral discomfortPosts mention simple self-care strategies such as saltwater rinses and vitamin supplements primarily used to address dry mouth or mild oral discomfort, often in children. Topic 3: Home remedies for oral discomfort Posts mention simple self-care strategies such as saltwater rinses and vitamin supplements primarily used to address dry mouth or mild oral discomfort, often in children. 5.Topic 4: Swollen gums and clinical interventionTalks about gum pain and swelling that often leads to clinic visits and extractions. Topic 4: Swollen gums and clinical intervention Talks about gum pain and swelling that often leads to clinic visits and extractions. 6.Topic 5: Postsurgical bleeding and recoveryThis topic centres on experiences following oral surgery, especially wisdom tooth removal. Users commonly report bleeding, discomfort and recovery concerns. Topic 5: Postsurgical bleeding and recovery This topic centres on experiences following oral surgery, especially wisdom tooth removal. Users commonly report bleeding, discomfort and recovery concerns. 7.Topic 6: Microorganisms and oral-systemic healthUsers link viral or bacterial infections to oral conditions, often speculating about triggers such as food or beverages (eg, coffee), revealing lay interpretations of disease causality. Topic 6: Microorganisms and oral-systemic health Users link viral or bacterial infections to oral conditions, often speculating about triggers such as food or beverages (eg, coffee), revealing lay interpretations of disease causality. 8.Topic 7: Vitamins and nutritional deficiencyPosts emphasises the importance of nutrients for oral health, especially for children and older adults. Topic 7: Vitamins and nutritional deficiency Posts emphasises the importance of nutrients for oral health, especially for children and older adults. 9.Topic 8: Mouth ulcers and symptom managementMouth ulcers are a frequent concern, with users discussing triggers, such as food and drink, and self-medication practices aimed at symptom relief and healing. Topic 8: Mouth ulcers and symptom management Mouth ulcers are a frequent concern, with users discussing triggers, such as food and drink, and self-medication practices aimed at symptom relief and healing. 10.Topic 9: Medication use during pregnancyExpectant mothers express uncertainty around the use of medications for dental pain or nausea. Concerns focus on safety, signalling a need for clearer guidance. Topic 9: Medication use during pregnancy Expectant mothers express uncertainty around the use of medications for dental pain or nausea. Concerns focus on safety, signalling a need for clearer guidance. Topic distribution and salient terms from LDA-based topic modelling on social media discussions. The 10 topics identified through LDA were grouped into 3 broader clusters to make the findings easier to understand and more meaningful for public health. Although LDA detects patterns of words statistically, the results can appear scattered without a clear structure. Therefore, the topics were reorganised into 3 main domains (1) symptom management and acute care, (2) preventive behaviours and health knowledge and (3) vulnerable populations. These clusters reflect 3 common patterns in online discussions. This approach helps translate machine-generated results into a clearer public health framework. Within this structure, bruxism and fixed orthodontic treatment were placed in the Preventive Behaviours and Health Knowledge cluster. Bruxism is usually discussed in relation to stress, sleep problems and the use of night guards, focusing more on causes and management than on acute disease. Fixed orthodontic treatment, especially when related to aesthetic correction of malocclusion, represents planned and elective care. It reflects long-term decisions about function and appearance rather than responses to pain or infection. Although orthodontic care is not preventive in a strict clinical sense, it shows intentional health investment and behavior-based decision making. As demonstrated in Table 2, this structured clustering was systematically compared with RISKESDAS 2018 and SKI 2023 indicators, revealing both convergence with nationally reported disease burdens and important experiential and perceptual gaps not captured in conventional survey instruments.Table 2Clustering of LDA topics and alignment with national survey indicators.Table 2 dummy alt textClusterLDA topicCore discourse focusRelated conditions in social mediaReported in RISKESDAS 2018/SKI 2023InterpretationSymptom management and acute careTopic 3: Home remedies for oral discomfortSelf-treatment for oral discomfortDry MouthMild oral discomfortNot reportedNot directly reportedInitial self-management prior to seeking professional careTopic 4: Swollen gums and clinical interventionPain, swellingPeriodontal issuesToothacheOral biopsyReportedReportedNot reportedClinical attendance primarily prompted by acute symptomsTopic 5: Postsurgical bleeding and recoveryAnxiety after extractionPostextraction bleedingImpacted teethNot reportedNot reportedLived treatment experiencesTopic 8: Mouth ulcers and symptom managementUlcer triggers and self-medicationMouth ulcers (adult and children)ReportedFrequently reported acute oral conditionPreventive behaviours and health knowledgeTopic 0: Daily oral hygiene practicesBrushing routines and discomfortDental calculus,Brushing frequencyUse of fixed orthodonticNot reportedReportedNot reportedPreventive behavior discourseTopic 2: Oral diseases & infectionsDisease awareness, extraction as outcomeDental decayTooth abscessReportedNot reportedReflects disease knowledge framingTopic 6: Microorganisms, oral-systemic and functional health linkMicrobial causality beliefsInfection narrativesBruxismNot directly reportedNot reportedNonprofessional interpretations of biological causesTopic 7: Vitamins & Nutritional deficiencyNutritional prevention focusNutritional deficiency for children and elderly oral careNot reportedPreventive literacy and belief systemsVulnerable populationsTopic 1: Children’s dental appointmentParental anxiety, scheduling concernsTooth abscessTooth eruptionAcute PulpitisNot reportedNot reportedNot reportedParental uncertainty regarding paediatric dental care decisionsTopic 9: Medication use during pregnancyDrug safety concernsDental issues during pregnancyNot reportedInformational gap in maternal oral care Table 2Clustering of LDA topics and alignment with national survey indicators.Table 2 dummy alt textClusterLDA topicCore discourse focusRelated conditions in social mediaReported in RISKESDAS 2018/SKI 2023InterpretationSymptom management and acute careTopic 3: Home remedies for oral discomfortSelf-treatment for oral discomfortDry MouthMild oral discomfortNot reportedNot directly reportedInitial self-management prior to seeking professional careTopic 4: Swollen gums and clinical interventionPain, swellingPeriodontal issuesToothacheOral biopsyReportedReportedNot reportedClinical attendance primarily prompted by acute symptomsTopic 5: Postsurgical bleeding and recoveryAnxiety after extractionPostextraction bleedingImpacted teethNot reportedNot reportedLived treatment experiencesTopic 8: Mouth ulcers and symptom managementUlcer triggers and self-medicationMouth ulcers (adult and children)ReportedFrequently reported acute oral conditionPreventive behaviours and health knowledgeTopic 0: Daily oral hygiene practicesBrushing routines and discomfortDental calculus,Brushing frequencyUse of fixed orthodonticNot reportedReportedNot reportedPreventive behavior discourseTopic 2: Oral diseases & infectionsDisease awareness, extraction as outcomeDental decayTooth abscessReportedNot reportedReflects disease knowledge framingTopic 6: Microorganisms, oral-systemic and functional health linkMicrobial causality beliefsInfection narrativesBruxismNot directly reportedNot reportedNonprofessional interpretations of biological causesTopic 7: Vitamins & Nutritional deficiencyNutritional prevention focusNutritional deficiency for children and elderly oral careNot reportedPreventive literacy and belief systemsVulnerable populationsTopic 1: Children’s dental appointmentParental anxiety, scheduling concernsTooth abscessTooth eruptionAcute PulpitisNot reportedNot reportedNot reportedParental uncertainty regarding paediatric dental care decisionsTopic 9: Medication use during pregnancyDrug safety concernsDental issues during pregnancyNot reportedInformational gap in maternal oral care Clustering of LDA topics and alignment with national survey indicators. To examine whether the distribution of oral health topics differed between Quora and Threads, a chi-square test of independence was conducted. The test showed a significant association between platform and topic distribution (χ² = 52.4, df = 18, P < .001). To validate the result, a Monte Carlo simulation with 10,000 iterations was used, confirming the association's significance (2-sided P < .001). Discussion This study examined oral health discourse on Indonesian social media, revealing how digital platforms reflect community health concerns, behavioural patterns and surveillance gaps. Analysis of 858 user-generated posts from Quora and Threads showed that discussions were predominantly symptom-driven, with toothache, mouth ulcers and dental decay emerging as the most salient topics. These findings underscore a persistent focus on pain and treatment rather than prevention which a pattern consistent with the curative orientation of oral health behaviour in many low and middle-income contexts. The limited presence of oral hygiene education and preventive messages in online discussions signals an important opportunity for targeted digital health promotion; it also suggests that preventive knowledge is not spreading naturally online and requires more intentional intervention. The underrepresentation of vulnerable groups such as pregnant women, older adults and individuals with disabilities suggests structural invisibility rather than lack of need. This gap represents a missed opportunity for structured digital intervention. Latent Dirichlet allocation (LDA) analysis revealed 10 distinct thematic diversity, demonstrates that social media serves as both a personal health diary and a public forum for experiential learning. In order to render the findings more accessible to a broader public health audience, the 10 identified LDA topics were organised into 3 clusters. Table 2 highlights how the clustered LDA findings relate to nationally reported oral health conditions in RISKESDAS 2018 and SKI 2023. The comparison shows clear overlap in major problems such as toothache, mouth ulcers, periodontal disease and dental decay, indicating that social media discussions reflect real and measurable disease burdens in the population. This alignment supports the external validity of the digital findings and suggests that online discourse is not random, but closely connected to existing epidemiological patterns.19 At the same time, the table reveals important gaps. Several issues frequently discussed online such as postextraction recovery, medication concerns during pregnancy, bruxism, fixed orthodontic treatment and parental anxiety are not explicitly captured in national survey instruments. This suggests that while national surveys effectively measure disease prevalence, they may overlook experiential, behavioural and perception-based dimensions of oral health. Therefore, Table 2 demonstrates the complementary value of social media analysis as an additional layer of surveillance that captures how people experience, interpret and respond to oral health problems beyond clinical indicators. Taken together, these findings indicate that while preventive themes exist within online discourse, they remain overshadowed by curative and episodic concerns; reflecting the same structural tendencies identified in the broader communication landscape. Similar to findings from earlier studies, preventive topics tend to receive far less attention in online discussion than symptom-related concerns. As a result, themes related to preventive and promotive aspects including fluoride use, oral hygiene education and early detection were markedly underrepresented.20,21 The observed similarity in clustering patterns is further corroborated by the findings from other digital health analyses, such as Twitter topic modelling centred on fluoride-free dentistry.22,23 These observations underscore the potential to strategically integrate preventive measures, including dietary guidance and timely dental visits, within high-engagement symptom discussions. This approach is designed to systematically transition the discourse towards a focus on long-term prevention. Marked contrasts were observed between the 2 platforms. Quora tended to host structured, formal discussions, often supported by clinical reasoning or evidence, whereas Threads encouraged more casual, emotionally expressive conversations cantered on empathy and shared personal experiences. These differences highlight the importance of tailoring oral health communication to each platform’s distinct culture and user expectations. As shown in previous research, Q&A forums generally facilitate information-rich exchanges, while social networks favour concise and emotionally engaging interactions.24,25 From a strategic standpoint, these differences imply that oral health communication should be platform-specific. Monte Carlo simulations confirmed significant content differences across platforms, suggesting that audience segmentation and tailored messaging may be essential for effective communication. Quora-like platforms can be leveraged for detailed educational content and professional Q&A formats, while emotionally driven platforms like Threads are better suited for narrative-based health promotion and peer-driven motivation. Recognising and optimising these affordances is critical to ensure message and promote positive behavioural engagement across diverse digital audiences. However, the effectiveness of any digital strategy ultimately depends on the quality and authenticity of the data on which it is built. Therefore, even though data collection was performed through keyword-based automated extraction, manual screening remained critical to filter out irrelevant contexts of oral health terms. This reflects broader concerns about misinformation in unstructured online content and underscores the need for rigorous verification processes.26,27 The comparison between topics emerging from social media and official national datasets such as RISKESDAS 2018 and SKI 2023 reveals important informational gaps that may influence future oral health policy. The efficacy of national surveys in tracking prevalent health conditions, such as dental caries and general toothache, has been well-documented. Nevertheless, discourse on social media has revealed a more extensive array of public concerns that are not currently reflected in government indicator. For example, conversations about dental medications, impacted teeth and postextraction complications appear frequently online but remain absent from national reporting. This finding suggests the potential for surveillance systems to evolve beyond the scope of disease prevalence, encompassing individuals' experiences with treatment and self-managed pain. This proposed broadening of surveillance systems, from tracking disease prevalence to encompassing individuals' treatment experiences and self-managed pain, reflects how communities actually navigate oral health challenges in daily life. In addition, discussions related to dental care for older adults, oral health during pregnancy and primary tooth eruption in children highlight unmet informational needs among specific population groups that are not explicitly detailed in national reports. Emerging topics such as bruxism, oral biopsies and dental calculus also indicate substantial public interest or anxiety around conditions often considered specialist issues. These areas represent strategic opportunities for preventive education before cases progress to more complex and costly treatments. Consequently, social media can serve as a valuable source of data, capturing experiences with treatment and unmet needs that may not be fully addressed by traditional epidemiological methods. Real-time digital monitoring can function as an early-warning system, revealing ‘silent epidemics’ of untreated pain, cost-related care avoidance and anxiety driven by misinformation. In light of the growing number of individuals who seek counsel via online sources due to financial constraints or accessibility issues, the utilisation of digital instruments, such as AI-powered chatbots, authenticated expert responses and targeted initiatives, is poised to assume greater significance in the enhancement of preventive oral healthcare in Indonesia. Despite global efforts to promote a life-course approach, online discussions remain largely short-term and problem-focused. Nutrition and oral-systemic health are mentioned, but often without clear links to long-term prevention or healthy ageing, and sometimes based on personal assumptions rather than evidence. Integrating social media analysis into early stages of national oral health monitoring could provide real-time insights and support more inclusive strategies aligned with the WHO Global Oral Health Action Plan.28 This gap between discourse and policy priorities represents a missed opportunity for digital health promotion. Embedding preventive messages into high-engagement topics such as daily oral hygiene or common symptoms could help realign public perception with long-term goals. Practical steps may include the utilisation of WhatsApp for the purpose of providing triage support, collaboration with trusted influencers to address issues of misinformation, and the creation of short videos explaining oral hygiene milestones, with a focus on specific ages and demographics. The objective of these efforts is to engage younger audiences. The platform differences observed in this study further support the need for tailored communication. Threads featured more content on children’s dental development, suggesting an audience of parents seeking quick advice, while Quora contained more detailed accounts of elderly care and complex dental procedures. Overall, the discourse leaned toward curative concerns, with prevention, cancer screening, orthodontics and disability-inclusive care underrepresented. Similar findings have been documented in studies of children’s oral health, where targeted caregiver engagement is key to improving preventive behaviours.29,30 These consistent patterns across population highlight the need for robust analytical frameworks capable of capturing both explicit and latent dimensions of public discourse. The strengths of this study lie in its analytical triangulation, integrating qualitative content analysis, computational text analytics (LDA) and statistical testing to yield a comprehensive and contextually grounded understanding of oral health discourse. The research design combines several complementary analytical approaches. It leverages the precision of predefined, expert-validated keywords to ensure accurate data capture and categorisation. Unsupervised LDA was then applied to uncover latent themes beyond pre-existing categories, while chi-square testing validated by Monte Carlo simulation was used to robustly quantify platform-specific differences. This methodological approach enables a comprehensive, and multi-layered analysis of online oral health discourse. The predefined coding provides a validated framework for classifying known concerns based on expert judgement. the LDA method reveals the organic and naturally emerging themes within public discussions, offering insights beyond predetermined categories. Statistical testing then adds rigor by confirming that the observed patterns are not due to chance. This systemic mapping of public oral health discourse offers a novel and robust perspective for identifying gaps in alignment with global health priorities and for designing more targeted platform-specific communication strategies. However, several limitations should be acknowledged. The lack of demographic information (gender, age and location) hindered the execution of subgroup analysis and limited the ability to design targeted and equitable interventions. While digital trace data offer valuable real-time insights, they do not provide the structured demographic detail typically available in national surveys. Future research could address this gap through ethical methods such as language pattern analysis or anonymised location estimation, thereby strengthening the policy relevance of digital epidemiology without compromising privacy. In addition, the exclusion of private or closed-group posts may have limited access to potentially rich but inaccessible perspectives. Although inter-rater reliability was high, interpreting user-generated content remains inherently complex and context-dependent. To preserve analytic integrity, a strict data cleaning process was implemented. Nonindividual content, including advertisements and memes, was excluded to ensure the analysis focused on authentic personal experiences. While such materials may offer insight into misinformation or social trends in future studies, their removal was necessary here to maintain clarity and validity in examining public discourse. Finally, as a recommendation, future studies could adopt a cross-platform design to compare different social media environments. This approach would reach a broader range of people and provide more nuanced insights into user behaviour by analysing differences in language and the length of their posts. The findings from this study can serve as valuable input for the government, particularly in planning and conducting national surveys. The incorporation of social media analysis as a component of the preliminary research phase has the potential to facilitate the identification of prevalent health concerns that are being frequently discussed by the public. This could lead to a deeper understanding of how people describe common issues related to oral health. In conclusion, social media is no longer a peripheral data source but an essential tool for contemporary oral health surveillance. This study reveals not only gaps in oral health equity but also the digital means to address them in real time. Despite their inherent limitations, these platforms should be considered an essential complementary layer within modern oral health surveillance systems in Indonesia's oral health landscape. Combining insights from social media with structured epidemiological data offers valuable evidence for policy formulation and supports the design of health promotion program that respond to the real needs of the community. Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.