Junyu Liu

Junyu Liu

Auto-generated profile, based on publicly available information

Junyu Liu

Kyoto University

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
9 May 2022 in JMIR Infodemiology
Verified
Journal Article
Open access
QNQian Niu
Qian Niu
JLJunyu Liu
Junyu Liu
MKMasaya Kato
Masaya Kato
YSYuki Shinohara
Yuki Shinohara
NMNatsuki Matsumura
Natsuki Matsumura
MNMomoko Nagai-Tanima
Momoko Nagai-Tanima
Background: COVID-19 vaccines are considered one of the most effective ways for containing the COVID-19 pandemic, but Japan lagged behind other countries in vaccination in the early stages. A deeper understanding of the slow progress of vaccination in Japan can be instructive for COVID-19 booster vaccination and vaccinations during future pandemics. Objective: This retrospective study aims to analyze the slow progress of early-stage vaccination in Japan by exploring opinions and sentiment toward the COVID-19 vaccine in Japanese tweets before and at the beginning of vaccination. Methods: We collected 144,101 Japanese tweets containing COVID-19 vaccine-related keywords between August 1, 2020, and June 30, 2021. We visualized the trend of the tweets and sentiments and identified the critical events that may have triggered the surges. Correlations between sentiments and the daily infection, death, and vaccination cases were calculated. The latent dirichlet allocation model was applied to identify topics of negative tweets from the beginning of vaccination. We also conducted an analysis of vaccine brands (Pfizer, Moderna, AstraZeneca) approved in Japan. Results: The daily number of tweets continued with accelerating growth after the start of large-scale vaccinations in Japan. The sentiments of around 85% of the tweets were neutral, and negative sentiment overwhelmed the positive sentiment in the other tweets. We identified 6 public-concerned topics related to the negative sentiment at the beginning of the vaccination process. Among the vaccines from the 3 manufacturers, the attitude toward Moderna was the most positive, and the attitude toward AstraZeneca was the most negative. Conclusions: Negative sentiment toward vaccines dominated positive sentiment in Japan, and the concerns about side effects might have outweighed fears of infection at the beginning of the vaccination process. Topic modeling on negative tweets indicated that the government and policy makers should take prompt actions in building a safe and convenient vaccine reservation and rollout system, which requires both flexibility of the medical care system and the acceleration of digitalization in Japan. The public showed different attitudes toward vaccine brands. Policy makers should provide more evidence about the effectiveness and safety of vaccines and rebut fake news to build vaccine confidence.

11 February 2025 in Journal of Medical Internet Research
Verified
Journal Article
Open access
JLJunyu Liu
Junyu Liu
QNQian Niu
Qian Niu
MNMomoko Nagai-Tanima
Momoko Nagai-Tanima
TATomoki Aoyama
Tomoki Aoyama
Background: Despite the reinstatement of proactive human papillomavirus (HPV) vaccine recommendations in 2022, Japan continues to face persistently low HPV vaccination rates, which pose significant public health challenges. Misinformation, complacency, and accessibility issues have been identified as key factors undermining vaccine uptake. Objective: This study aims to examine the evolution of public attitudes toward HPV vaccination in Japan by analyzing social media content. Specifically, we investigate the role of misinformation, public health events, and cross-vaccine attitudes (eg, COVID-19 vaccines) in shaping vaccine hesitancy over time. Methods: We collected tweets related to the HPV vaccine from 2011 to 2021. Natural language processing techniques and large language models (LLMs) were used for stance analysis of the collected data. Time series analysis and latent Dirichlet allocation topic modeling were used to identify shifts in public sentiment and topic trends over the decade. Misinformation within opposed-stance tweets was detected using LLMs. Furthermore, we analyzed the relationship between attitudes toward HPV and COVID-19 vaccines through logic analysis. Results: Among the tested models, Gemini 1.0 pro (Google) achieved the highest accuracy (0.902) for stance analysis, improving to 0.968 with hyperparameter tuning. Time series analysis identified significant shifts in public stance in 2013, 2016, and 2020, corresponding to key public health events and policy changes. Topic modeling revealed that discussions around vaccine safety peaked in 2015 before declining, while topics concerning vaccine effectiveness exhibited an opposite trend. Misinformation in topic "Scientific Warnings and Public Health Risk" in the sopposed-stance tweets reached a peak of 2.84% (47/1656) in 2012 and stabilized at approximately 0.5% from 2014 onward. The volume of tweets using HPV vaccine experiences to argue stances on COVID-19 vaccines was significantly higher than the reverse. Conclusions: Based on observation on the public attitudes toward HPV vaccination from social media contents over 10 years, our findings highlight the need for targeted public health interventions to address vaccine hesitancy in Japan. Although vaccine confidence has increased slowly, sustained efforts are necessary to ensure long-term improvements. Addressing misinformation, reducing complacency, and enhancing vaccine accessibility are key strategies for improving vaccine uptake. Some evidence suggests that confidence in one vaccine may positively influence perceptions of other vaccines. This study also demonstrated the use of LLMs in providing a comprehensive understanding of public health attitudes. Future public health strategies can benefit from these insights by designing effective interventions to boost vaccine confidence and uptake.

9 June 2022 in Journal of Medical Internet Research
Verified
Journal Article
Open access
QNQian Niu
Qian Niu
JLJunyu Liu
Junyu Liu
MKMasaya Kato
Masaya Kato
MNMomoko Nagai-Tanima
Momoko Nagai-Tanima
TATomoki Aoyama
Tomoki Aoyama
Background: The global public health and socioeconomic impacts of the COVID-19 pandemic have been substantial, rendering herd immunity by COVID-19 vaccination an important factor for protecting people and retrieving the economy. Among all the countries, Japan became one of the countries with the highest COVID-19 vaccination rates in several months, although vaccine confidence in Japan is the lowest worldwide. Objective: We attempted to find the reasons for rapid COVID-19 vaccination in Japan given its lowest vaccine confidence levels worldwide, through Twitter analysis. Methods: We downloaded COVID-19–related Japanese tweets from a large-scale public COVID-19 Twitter chatter data set within the timeline of February 1 and September 30, 2021. The daily number of vaccination cases was collected from the official website of the Prime Minister’s Office of Japan. After preprocessing, we applied unigram and bigram token analysis and then calculated the cross-correlation and Pearson correlation coefficient (r) between the term frequency and daily vaccination cases. We then identified vaccine sentiments and emotions of tweets and used the topic modeling to look deeper into the dominant emotions. Results: We selected 190,697 vaccine-related tweets after filtering. Through n-gram token analysis, we discovered the top unigrams and bigrams over the whole period. In all the combinations of the top 6 unigrams, tweets with both keywords “reserve” and “venue” showed the largest correlation with daily vaccination cases (r=0.912; P<.001). On sentiment analysis, negative sentiment overwhelmed positive sentiment, and fear was the dominant emotion across the period. For the latent Dirichlet allocation model on tweets with fear emotion, the two topics were identified as “infect” and “vaccine confidence.” The expectation of the number of tweets generated from topic “infect” was larger than that generated from topic “vaccine confidence.” Conclusions: Our work indicates that awareness of the danger of COVID-19 might increase the willingness to get vaccinated. With a sufficient vaccine supply, effective delivery of vaccine reservation information may be an important factor for people to get vaccinated. We did not find evidence for increased vaccine confidence in Japan during the period of our study. We recommend policy makers to share accurate and prompt information about the infectious diseases and vaccination and to make efforts on smoother delivery of vaccine reservation information.

22 July 2021 in medRxiv
Verified
Preprint
Open access
PDF file
QNQian Niu
Qian Niu
JLJunyu Liu
Junyu Liu
MKMasaya Kato
Masaya Kato
YSYuki Shinohara
Yuki Shinohara
NMNatsuki Matsumura
Natsuki Matsumura
MNMomoko Nagai-Tanima
Momoko Nagai-Tanima
Background The global public health and socioeconomic impacts of COVID-19 have been substantial. To achieve herd immunity, widespread use of the vaccine is required, and it is therefore critical for government and public health agencies to understand public perceptions of the vaccine to help sustain subsequent vaccinations. Objective This study aims to explore the opinions and sentiments of tweets about COVID-19 vaccination among Twitter users in Japan, both before and at the beginning of the COVID-19 vaccination program. Methods We collected 144,101 Japanese tweets containing COVID-19 vaccine-related keywords from Japanese Twitter users between August 1, 2020, and June 30, 2021. Specifically, we identified temporal changes in the number of tweets and key events that triggered a surge in the number of tweets. In addition, we performed sentiment analysis, and calculated the correlation between different sentiments and the number of deaths, infections, and vaccinations. We also built latent Dirichlet allocation (LDA) topic models to identify commonly discussed topics in a large sample of tweets. We also provided a word cloud of high-frequency unigram and bigram tokens as additional evidence, and conducted further analysis on three different vaccine brands. Results The overall number of tweets has continued to increase since the start of mass vaccination in Japan. Sentiments were generally neutral, but negative sentiment was more significant than positive sentiment. Before and after the first vaccination in Japan, the correlations of negative/positive sentiment with death, infection, and vaccination cases changed significantly. Public concerns revolved around three themes: information on vaccine reservations and vaccinations in Japan; infection and mutation of COVID-19 in Japan; and prevention measures, vaccine development and supply, and vaccination status in other countries. Furthermore, public attention to the three brands of vaccines has a temporal shift as clinical trials move forward. Conclusions The number of tweets and changes in sentiment might be driven by major news events in relation to the COVID-19 vaccine, with negative sentiments dominating positive sentiments overall. Death and infection cases correlated significantly with negative sentiments, but the correlation fell after vaccinations began as morbidity and mortality decreased. The public’s attention to different vaccine brands had a temporal change during their clinical trial process, and although the discussion points are slightly different, the core remains effective and secure.

21 May 2026 in JMIR Infodemiology
Verified
Journal Article
Open access
JLJunyu Liu
Junyu Liu
SYSiwen Yang
Siwen Yang
DMDexiu Ma
Dexiu Ma
QNQian Niu
Qian Niu
ZZZequn Zhang
Zequn Zhang
TATomoki Aoyama
Tomoki Aoyama
Background: Human papillomavirus (HPV) vaccine hesitancy remains a significant public health challenge in Japan, where proactive vaccination recommendations were suspended between 2013 and 2021. The resulting information gap between medical institutions and vaccine-hesitant populations is exacerbated by misinformation on social media platforms. Traditional public health communication strategies cannot address individual queries while simultaneously monitoring population-level discourse. Objective: This study aimed to develop and conduct a feasibility evaluation of a dual-purpose artificial intelligence agent system that delivers verified HPV vaccine information to the public through a conversational interface while generating infoveillance reports for medical institutions based on user interactions and social media discourse. Methods: We implemented a system with 3 components: a vector database integrating 139,803 documents, including academic papers, Japanese government sources, news media, and social media posts; a retrieval-augmented generation chatbot using a ReAct agent architecture with iterative multitool orchestration across 5 specialized knowledge sources; and an automated report generation system with modules for news analysis, research synthesis, social media sentiment analysis, including stance classification and topic modeling, and user interaction pattern identification. System performance was assessed using both automated and manual evaluation protocols on a scale from 0 to 5. Results: The entire system functioned as expected. For single-turn evaluation, the chatbot achieved mean scores of 4.83 (SD 0.67; 95% CI 4.71-4.93) for relevance, 4.89 (SD 0.53; 95% CI 4.79-4.97) for routing, 4.50 (SD 1.29; 95% CI 4.27-4.70) for reference quality, 4.90 (SD 0.62; 95% CI 4.78-4.99) for correctness, and 4.88 (SD 0.54; 95% CI 4.78-4.96) for professional identity, with an overall mean of 4.80. Multiturn evaluation yielded higher mean scores: 4.94 for context memory (SD 0.25; 95% CI 4.84-5.00) and an overall mean of 4.98, with topic centering and identity achieving 5.00. The report generation system achieved high scores across all sections: 4.83 for completeness (SD 0.37; 95% CI 4.73-4.94), 4.88 for correctness (SD 0.33; 95% CI 4.77-4.96), and 4.12 for helpfulness (SD 0.48; 95% CI 3.98-4.27). Reference validity achieved perfect scores (5.00) across all periods, with citation correctness averaging 4.21 (SD 0.58; 95% CI 3.96-4.46). Conclusions: This feasibility study demonstrated that an integrated artificial intelligence agent system can support both public HPV vaccine communication and social media infoveillance in a Japanese-language context. Prospective deployment with real users is needed to assess actual public health impact.

22 September 2025
Verified
Book Chapter
Closed access
PDF file
JLJunyu Liu
Junyu Liu
LYLawrence K. Q. Yan
Lawrence K. Q. Yan
TWTianyang Wang
Tianyang Wang
QNQian Niu
Qian Niu
MNMomoko Nagai-Tanima
Momoko Nagai-Tanima
TATomoki Aoyama
Tomoki Aoyama