Electronic Theses and Dissertation

Pendahuluan: infodemik selama pandemi covid-19 memengaruhi persepsi publik terhadap vaksinasi melalui penyebaran informasi di media daring. narasi terkait keamanan, efek samping, efektivitas, dan status kehalalan vaksin berkembang secara luas dan berpotensi membentuk keraguan masyarakat terhadap vaksinasi. namun, bukti empiris mengenai hubungan antara narasi infodemik media daring, perilaku pencarian informasi publik, dan capaian vaksinasi masih terbatas, khususnya pada fase awal program vaksinasi covid-19 di indonesia. penelitian ini bertujuan mengidentifikasi karakteristik narasi infodemik dalam media daring, menganalisis isu-isu utama yang merepresentasikan keraguan publik terhadap vaksinasi, mengevaluasi dinamika minat pencarian publik terkait aspek kehalalan dan keamanan vaksin, serta menganalisis hubungan antara dinamika pencarian informasi publik dengan capaian vaksinasi. metode: penelitian observasional ini menggunakan pendekatan infodemiologi komputasional untuk menganalisis media daring dan perilaku pencarian publik terkait vaksinasi covid-19. dataset berupa judul berita daring terkait vaksinasi covid-19 di indonesia dikumpulkan selama januari–maret 2021 melalui google news. tahapan analisis meliputi text preprocessing, pemodelan topik berbasis transformer menggunakan bertopic untuk mengidentifikasi struktur narasi infodemik, validasi menggunakan latent dirichlet allocation (lda), serta analisis bi-gram dan tri-gram untuk memperoleh frasa dominan yang merepresentasikan dimensi penerimaan vaksinasi. hasil integrasi pemodelan topik dan analisis n-gram digunakan untuk menghasilkan kata kunci dan kueri terpilih yang relevan dengan dimensi keraguan vaksinasi covid-19. kueri tersebut kemudian digunakan pada google trends untuk memantau dinamika minat pencarian publik melalui data relative search volume (rsv). selanjutnya, data rsv dianalisis bersama data cakupan vaksinasi nasional menggunakan uji stasioneritas, cross-correlation, dan uji kausalitas granger. hasil: analisis dengan bertopic mengidentifikasi dua kelompok narasi infodemik dominan dalam pemberitaan media daring terkait vaksinasi covid-19di indonesia, yaitu isu kehalalan vaksin serta keamanan dan efek samping vaksin. pola narasi tersebut menunjukkan konsistensi dengan menggunakan lda, bi-gram dan tri-gram. beberapa kata kunci dominan seperti “vaksin halal”, “fatwa mui”, dan “efek samping vaksin” kemudian digunakan sebagai kueri google trends untuk memantau dinamika minat pencarian publik selama periode awal program vaksinasi nasional. hasil analisis menunjukkan bahwa peningkatan pencarian terkait status halal vaksin cenderung mendahului peningkatan cakupan vaksinasi dosis pertama, sedangkan pencarian terkait efek samping dan keamanan vaksin menunjukkan pola yang lebih fluktuatif. analisis deret waktu menunjukkan adanya keterkaitan berdasarkan pola waktu antara minat pencarian publik dan capaian vaksinasi, meskipun hubungan tersebut bersifat terbatas dan lebih terlihat pada fase awal vaksinasi dosis pertama dibandingkan dosis kedua. uji kausalitas granger tidak menunjukkan hubungan prediktif yang signifikan (p>0,05). diskusi: dinamika infodemik menunjukkan keterkaitan berdasarkan pola waktu yang lebih terlihat pada fase awal penerimaan vaksinasi, khususnya pada vaksinasi dosis pertama. integrasi bertopic, lda, analisis n-gram, dan google trends menunjukkan potensi sebagai pendekatan infodemiologi komputasional untuk mengidentifikasi indikasi dini keraguan masyarakat terhadap vaksinasi berbasis data digital. temuan ini mengindikasikan bahwa minat pencarian publik lebih merefleksikan faktor penerimaan masyarakat terhadap vaksinasi, sedangkan keberlanjutan dan penyelesaian vaksinasi lebih dipengaruhi oleh pelaksanaan program vaksinasi dan akses terhadap layanan kesehatan. kata kunci: infodemiologi komputasional; bertopic; latent dirichlet allocation; google trends; vaksinasi covid-19; keraguan vaksin; analisis deret waktu.

Introduction: The COVID-19 infodemic significantly influenced public perceptions of vaccination through the widespread dissemination of information across online media platforms. Narratives related to vaccine safety, side effects, effectiveness, and halal status circulated extensively and had the potential to shape public vaccine hesitancy. However, empirical evidence examining the relationship between online infodemic narratives, public information-seeking behavior, and vaccination coverage remains limited, particularly during the early phase of the COVID-19 vaccination program in Indonesia. This study aimed to identify the characteristics of infodemic narratives in online media, analyze the major issues representing public vaccine hesitancy, evaluate the dynamics of public search interest related to vaccine halal status and safety, and examine the relationship between public information-seeking behavior and vaccination coverage. Methods: This observational study employed a computational infodemiology approach to analyze online media narratives and public search behavior related to COVID-19 vaccination. A dataset consisting of online news headlines related to COVID-19 vaccination in Indonesia was collected from Google News between January and March 2021. The analytical stages included text preprocessing, transformer-based topic modeling using BERTopic to identify infodemic narrative structures, validation using Latent Dirichlet Allocation (LDA), and bi-gram and tri-gram analyses to extract dominant phrases representing dimensions of vaccine acceptance. The integration of topic modeling and N-gram analysis was subsequently used to generate selected keywords and search queries relevant to COVID-19 vaccine hesitancy dimensions. These queries were then applied to Google Trends to monitor public search interest dynamics using Relative Search Volume (RSV) data. Furthermore, RSV data were analyzed alongside national vaccination coverage data using stationarity testing, cross-correlation analysis, and Granger causality testing. Results: BERTopic analysis identified two dominant infodemic narrative groups within online media coverage related to COVID-19 vaccination in Indonesia, namely vaccine halal-related issues and narratives concerning vaccine safety and side effects. These narrative patterns demonstrated consistency with the findings obtained from LDA, bi-gram, and tri-gram analyses. Several dominant keywords, including “halal vaccine,” “MUI fatwa,” and “vaccine side effects,” were subsequently used as Google Trends queries to monitor public search interest during the early phase of the national vaccination program. The findings indicated that increased searches related to vaccine halal status tended to precede increases in first-dose vaccination coverage, whereas searches concerning vaccine side effects and safety demonstrated more fluctuating patterns throughout the observation period. Time-series analysis revealed a temporal association between public search interest and vaccination coverage; however, this relationship was limited and more apparent during the early phase of first-dose vaccination compared with the second dose. Granger causality analysis did not demonstrate a statistically significant predictive relationship (p>0.05). Discussion: Infodemic dynamics demonstrated temporal associations that were more apparent during the early phase of vaccine acceptance, particularly for first-dose vaccination. The integration of BERTopic, LDA, N-gram analysis, and Google Trends showed potential as a computational infodemiology approach for identifying early indications of vaccine hesitancy using digital data sources. These findings suggest that public search interest primarily reflects demand-side factors related to vaccine acceptance, whereas the continuation and completion of vaccination are more strongly influenced by vaccine program implementation and access to healthcare services. Keywords: computational infodemiology; BERTopic; Latent Dirichlet Allocation; Google Trends; COVID-19 vaccination; vaccine hesitancy; time-series analysis.