Exploring the determinants of global vaccination campaigns to combat COVID-19

In the context of the COVID-19 pandemic, global vaccine campaigns are a mass vaccination intervention conducted through routine service provision for individuals who have reached a specified age. However, obtaining a high uptake rate to reach herd immunity may be undermined by various social motivations. To scrutinize the practical and dynamic strategies for a successful vaccination campaign, we map out the determinants that exacerbate vaccine hesitancy by leveraging the capacity of rich metadata from Twitter. Here, we uncover the collective propensities underlying dynamic social motivations and the uneven distribution of vaccines across the globe. Our findings suggest that profiling the status quo of public perceptions and engaging in introspection about vaccine-promoting policies in due course are integral components of preparedness against the ongoing pandemic. Simultaneously, we propose several recommendations to remind governments of the importance of building confidence in vaccination in a targeted way, and we assert that national barriers should be abandoned and that international responsibility should be assumed.


Introduction
T he coevolution of science and vaccine-promoting policies during an epidemic is a long-standing concern . However, our systematic understanding of the factors influencing the success of global vaccination remains limited, which may undermine global vaccination efforts to curb vaccinepreventable diseases. In the ongoing fight against the COVID-19 pandemic, vaccine hesitancy has emerged as a significant public health problem, topping the list of threats to global health (WHO, 2019). We must confront the fact that we need a more nuanced approach to mapping out the factors that exacerbate the lack of willingness to be vaccinated or anti-vaccination movements and reconcile our preparedness policies to address such societal disparities in due course.
We are now facing the ongoing priority of vaccination promotion, which requires dynamic and pertinent strategies to increase confidence in the new COVID-19 vaccines (Volpp et al., 2021). Conversely, inappropriate policies without scientific insights will affect how long it takes to achieve herd immunity across the globe (Forman et al., 2021). Crucially, unveiling the underlying social motivations of social mobilization is a keystone in dynamically regulating scientific decisions to bridge the gaps in societal disparities in COVID-19 vaccination. A growing number of recent studies have investigated the willingness to be vaccinated and the propagation of anti-vaccination movements in several countries (Murphy et al., 2021;Bradshaw et al., 2020;Edelstein et al., 2020) and have identified demographic, socioeconomic, and behavioural factors that are linked with vaccine acceptance (Kalam et al., 2021). The influential factors include gender, age and marital status (Al-Mohaithef and Padhi, 2020); level of education and ethnic origin (Edelstein et al., 2020;Bell et al., 2020); psychological factors and beliefs (Burki, 2020); and previous vaccination (Szilagyi et al., 2021). Moreover, previous survey results indicate that growing concerns over the efficacy and safety of vaccines, misinformation and declining levels of public trust in politics may reshape vaccine risk perceptions of the public Knipe et al., 2020;Larson, 2018;Zhang et al., 2019), which have been considered influential determinants of lower levels of acceptance (Lin et al., 2020). As a part of local and global immunization efforts, we need to think globally and act locally: Can the new vaccines be rapidly produced in various countries meet the standards? (Dodd et al., 2021). Do certain policies draw on unvetted and unsubstantiated evidence? What is the most urgent priority in addressing the public's mistrust of vaccines? How should vaccines be allocated to achieve herd immunity? (Roope et al., 2020). To answer these open questions, actors from each government should have open minds to modestly engage in introspection and rededicate themselves to leverage the full strength of vaccine policies (DeRoo et al., 2020).
Substantial global vaccine policies are grounded in scientific insights into the profiling of public perceptions of vaccines. Hence, the attitudes of vaccinating groups towards COVID-19 vaccines are a global concern that has arisen in many places in Europe, Asia and America, such as the USA (Loomba et al., 2021), the UK (Loomba et al., 2021), Spain (Eguia et al., 2021), Chile  and China . From a national perspective, in different countries, there is a high degree of heterogeneity in the public's opinions of vaccines (Dodd et al., 2021). Gaps in vaccine coverage across countries could potentially undermine global efforts against the pandemic and delay the ensuing societal and economic recovery. Currently, as large-scale societal labs, social media continue to stockpile seminal patterns of both individual and collective behaviours that go far beyond engaging in personal gossip (Mourad et al., 2020). Rich collections of digital social imprints are formidable proxies for determining the dynamic patterns of practical appeal. Hence, it is necessary to understand the social motivations of respondents in the global vaccination debate on Twitter.
To that end, this globally based study aims to articulate the paradigm of public engagement in COVID-19 vaccination to scrutinize the practical and dynamic strategies of preparedness to mitigate the pandemic. This study uncovers the collective propensities in social behaviours and reveals the variations in social motivations for global vaccination to assist authorities in making scientific and dynamic decisions.

Methods
Twitter is thought of as a natural behavioural experiment that profiles social responses to global vaccination. As one of the ten most visited websites on the internet worldwide, it is a crucial public opinion battlefield. In this study, after deweighting, we used the Twitter application programming interface (API) to retrieve and collect more than 19 million Twitter messages posted worldwide from November 2020 to April 2021. Through word frequency statistical analysis and manual induction, we obtained some topics that Twitter users paid more attention to, such as side effects, vaccine diplomacy, variants and so on; established a topic-based codebook suitable for our research; and further depicted public opinion transmission with regard to vaccinerelated topics. The sampling of original metadata, the sample sizes of the data and the keyword codebook are included in Supplementary Tables S1, S2 and S3.
Model for measuring the dimensions of vaccination. To locate the concrete dimensions that impact global vaccination, we utilized the Shapley Additive exPlanations (SHAP) model (Rodríguez-Pérez and Bajorath, 2020) and gradient boosting decision trees (GBDT) (Friedman, 2001) to calculate the contribution of keywords on Google Trends to vaccination topics and to indirectly reflect global opinions towards vaccination after people receive specific rich information.
In the interpretability of the model and the context of activity predictions, SHAP values can be rationalized as a fair or reasonable allocation of the importance of a feature given a particular model output (Rodríguez-Pérez and Bajorath, 2020). Accordingly, SHAP values represent estimates of feature importance (magnitude of the contribution) as well as the direction (sign). The SHAP model interprets the predicted value of the model as the sum of the attributed values of each input feature, which is expressed as where g is the explanatory model and x′ represents the structured data. M is the number of input features, ; i 2 R is the attribution value (SHAP value) of each feature, and ∅ 0 is the constant of the explanatory model. In light of the experiment (Lundberg et al., 2018), SHAP values consistently attribute feature importance, better align with human intuition, and better recover influential features. In addition, GBDT, such as extreme gradient boosting (XGBoost) (Chen and Guestrin, 2016) and categorical feature gradient boosting (CatBoost) (Dorogush et al., 2017), have been widely applied in various fields in recent years (Xiao et al., 2017;Xia et al., 2017) and possess the benefits of capturing interactions among features without explicitly defining them. Therefore, by combining the SHAP value method with GBDT models, it is possible to provide a detailed explanation of the contribution to vaccination topics (Meng et al., 2021). The theoretical descriptions of the SHAP scale, GBDT models and keywords are included in the Supplementary Information.
Model of global emotional response. In addition to our focus on global opinions on vaccination, we assessed the overall and finegained levels of collective emotions. We used multidimensional sentiment analysis to calculate the emotional classification of different dimensions of Twitter data based on bidirectional encoder representations from transformers (BERT) (Devlin et al., 2019). As a pretrained language model, BERT has shown its effectiveness in language representation and feature engineering, which is considered a great breakthrough in the field of natural language processing (NLP), refreshing the best results for 11 NLP tasks (Sun et al., 2019). Therefore, by making use of BERT, we could better identify fine-grained emotions of specific aspects related to vaccination. We utilized a five-layer structure for emotion analysis, including a word embedding layer based on BERT, a sequential layer for capturing semantic context, a bidirectional long short-term memory (LSTM) layer, a fully connected layer based on rectified linear units (ReLU) (Xavier et al., 2011) and a softmax layer to obtain the final emotional polarity.
Furthermore, we operationalized normalized mutual information (NMI) (Danon et al., 2005) as concrete indicators to assess the probable relationship between public emotions and vaccination. For two different types of A and B, the element of N, N ij , is the number of nodes in real community i that appear in found community j. A measure of similarity between the partitions, based on information theory, is then obtained: where the number of real communities is denoted by C A and the number of found communities is denoted by C B . The sum over row i of matrix N ij is denoted by N i. , and the sum over column j is denoted by N .j . The theoretical descriptions of the scale of emotion analysis and NMI are included in the SI. The sample sizes of the data are included in Supplementary Table S4.

Results
Vaccine-related topics on Twitter. People's attitudes towards vaccination always vary with different events, and the pattern of diversity, which is still unknown, is crucial for policy making. To fill these gaps, we reveal the implications of the public voice embedded in online behaviours. With an average of 199 million daily active users (Business of apps, 2021). Twitter can be regarded as a social experiment (Hofman et al., 2021). Compared with traditional media, netizen responses are highly contingent on sociocultural structures (Gigerenzer et al., 2011), which could reflect human behaviours. That is, the transmission of topics on Twitter can be inferred to collective human behaviours. According to the butterfly effect, even seemingly minor aspects of response can shape individual behaviours and scale up to a notable paradigm in collective human behaviours (Lorenz-Spreen et al., 2020). In the context of the COVID-19 pandemic, by leveraging rich metadata from Twitter (19 million tweets covering at least 20 countries), we reveal the paradigm shifts in global public opinion on the hottest vaccine-related topics.
People have reached a consensus on health policies, which need to dynamically adapt to public engagement. Bearing this idea in mind, we aimed to map out some substantive paradigms as hallmark references for policy making in a targeted way. In the first stage, the population of netizens last November was divided based on diverse vaccine-related topics with limited influences ( Supplementary Fig. S1). Subsequently, the number of tweets involved in vaccine-related topics dramatically increased over time. The initial paradigm indicates that the public was engaged in three main topics: potential side effects (0.948 million, 5%), politicians (0.981 million, 5.2%) and different groups of people (e.g., pregnant women, children, obese people) (0.385 million, 2%). In the second stage, the dominant paradigm shifted to the topics of politicians and vaccine diplomacy. The topic of vaccine diplomacy (0.055 million, 0.3%) gradually emerged in January 2021, wherein a relative minority of the population was involved ( Fig. 1). At the end of February, SARS-CoV-2 variants-B.1.1.7 (alpha variant), B.1.351 (beta variant), P.1 (gamma variant) (Mahase, 2021), and B.1.617 (delta variant) (Thompson and Pickrell, 2021)-began to spread globally, causing a devastating second wave of COVID-19. Subsequently, in the third stage, Fig. 1 The landscape of public opinion transmission with regard to vaccine-related topics. a The dynamic landscape of global opinion transmission could be divided into three stages that include the following topics: different groups of people (green), side effects (blue), politicians (red), vaccine diplomacy (magenta), variants (cyan), and other topics (grey). The size of the dots denotes the degree of topic discussion. The lines between the dots represent the mutual influence across different topics. b The number of tweets involved in different topics. Data were retrieved between 27 November 2020 and 19 April 2021. Topics were categorized based on the keywords in tweets (see the SI).
Vaccine-related topics on Google Trends indexes. The impact of many concrete dimensions on global vaccination remains unclear . Different countries have diversified dimensions of concern; however, there are still some common aspects, and reasonable global policies are the only way to achieve herd immunity. To examine the pertinent impacts, we further employed the SHAP model to calculate the contribution of different dimensions on Google Trends indexes to vaccination topics and to explore the underlying social motivations.
In the wake of COVID-19, vaccination is a prerequisite for preparedness against the pandemic, without which the result will be complete failure. Throughout the pandemic, there has been much attention and opposition to different aspects of vaccination (Wadman and You, 2017). Hence, it is very challenging to achieve herd immunity, and in this respect, there is still a long way to go. We curated 47 keywords and 10 clusters of the hottest topics on Twitter. Here, two boosting models (CatBoost and XGBoost) were used to analyse the different contributions to reduce the impact of model errors. Integrating the results of the two boosting models, we observe that some consistent paradigms show similar pro-vaccination attitudes after the people were exposed to information on certain types of vaccines, related vaccine trials, vaccine accessibility, the immunity of humans, and certain organizations (Fig. 2). Admittedly, the volunteers who received either a trial or a placebo were low-risk (Ewen, 2020), healthy adults rather than those belonging to high-risk populations (e.g., pregnant women (Beate, 2021), obese people (Kwok et al., 2020;Popkin et al., 2020), syndrome patients (Koirala et al., 2020); hence, there are also concerns about the safety of different groups being vaccinated. The public cares very much about the uncertainties of vaccines, for example, how long the vaccines' effectiveness will last (SHAP value: −0.31 to 0.23), how the government will distribute vaccines to the public (SHAP value: −0.49 to 0.44), whether asymptomatic people can be prevented from spreading the disease (SHAP value: −0.14 to 0.10) and how well the vaccines work in different groups, especially pregnant women (SHAP value: −0.17 to 0.10) and children (SHAP value: −0.06 to 0.05) (Subbaraman, 2021). These dimensions will become the motivations that affect people's willingness to be vaccinated. In particular, side effects, such as blood clotting (SHAP value: −0.11 to 0.11), autism (SHAP value: −0.16 to 0.13), and delirium (SHAP value: −0.41 to 0.26), could have considerable effects on the acceptance of COVID-19 vaccines. These negative dimensions are vulnerable to becoming driving forces that result in low global vaccination rates.
Global emotions on Twitter. Spatiotemporal sentiment analysis is a cutting-edge tool for characterizing the public's responses to vaccination. To glimpse the global view to further propose targeted policies, we performed sentiment analysis on rich metadata from Twitter based on geographical information to reveal the specific emotions expressed by people in regard to vaccination. Based on the expected start time of vaccination in most countries (Tim, 2020;CNN, 2020), we divided the metadata into two periods bounded by December 2020. Based on an existing Twitter data report (Business of apps, 2021; Briggs, 2020), we extracted the following six typical countries with high user activity in the metadata: the USA, the UK, Canada, Australia, France and India.
The different trends of emotional fluctuations could temporally reflect the collective responses to COVID-19 vaccines in specific countries (Fig. 3). Specifically, the emotion of worry over vaccines in the USA (avg., 57.7%), the UK (avg., 67.3%), Canada (avg., 60.6%) and Australia (avg., 63.5%) tend to be the same (Fig. 3a). In part, these countries have similar cultures, geopolitics, and concerns about vaccines, for instance, side effects (e.g., autism (Geschwind, 2009), blood clotting (Kupferschmidt and Vogel, 2021), myocarditis ( Kim et al., 2021) and Bell's palsy (Colella et al., 2021)). In France, which is a non-English-speaking country, the emotions of worry and happiness of the population (0.079 million, 0.4%) are much lower than those of the population in the USA, the UK, Canada and Australia. Overall, the population in French shows a downward trend in terms of worry and an upward trend in terms of neutral emotion. In India, the emotional trends are markedly different from those in the other countries. People generally show more happiness than sadness, and positive emotions increased to a peak before mid-March . Then, positive emotions began to decline dramatically when the Delta variant (B.1.617) triggered a new wave of the pandemic with India as the epicenter (Vaidyanathan, 2021).
Based on the idea that the world is a global community with a shared future, defeating COVID-19 calls for joint efforts that we all need to make in due course (Hotez and Narayan, 2021). In the wake of COVID-19, people always seek assistance from their domestic government and the international community. In the context of the current political backlash, uncovering the current concerns of netizens is crucial for local authorities to make the right decisions. To fill this gap, we first illustrate fine-grained emotions across countries (Fig. 4). Overall, British citizens, Canadians and Indians are more likely to express their views on the international pandemic situation. Comparably, many netizens in some countries, such as those in the UK, Canada and the USA, worry about what is happening in their own countries. Interestingly, the majority of netizens in Australia and India tend to consume international outlets rather than domestic news, especially outlets based in the UK, China, Canada and the USA. It can be understood that they tend to pay close attention to countries with soft power for vaccination, such as the permanent members of the UN Security Council or geographically close countries (Pannu and Barry, 2021). For example, French and Indian populations are mostly concerned about news updates in the UK and China, respectively. In addition, Australia is not mentioned much by other countries, partly because the COVID-19 situation in Australia was very different from that in other countries (Roberts, 2021).
As vaccination progresses, mapping out the degree of vaccine hesitancy is crucial for further rolling out vaccines to achieve herd immunity. Last December, the first COVID-19 vaccine from the pharmaceutical companies Pfizer and BioNTech was granted emergency-use authorization (Ledford et al., 2020). As of August 2021, more than 5 billion doses had been administered worldwide (Edouard et al., 2021). As a representative indicator, collective emotional polarity reflects the status of vaccine hesitancy, which may undermine global efforts in regard to vaccine promotion. In general, the emotions of netizens in the six countries tended to be stable, with worry, neutral emotion, happiness, love and sadness having average values of 0.63, 0.59, 0.47, 0.21 and 0.26, respectively (Fig. 5). The results remind us that public engagement is still a difficult task. Although the positive emotions of some populations tend to increase (Institute of Global Health Innovation, 2021), it should not be simply emphasized that the increase in positive emotions with vaccination means that the pandemic situation tends to be optimistic.
Global vaccine production capacity and vaccination status.  We divided Phase I and Phase II based on the sampling frequency, which was changed from once every 5 days to 3 times a day for self-verification, using dashed lines and transparency to distinguish. All curves are smoothed by Savitzky-Golay filtering to reconcile the differences in regional time zones. The results indicate that a higher sampling frequency in Phase II could articulate the nature of the public's emotional fluctuations. Worry and neutral emotions generally dominate people's expressions on Twitter, which is consistent with social sentiments (Hou et al., 2021) (see the SI). vaccination process has heightened people's anxieties and augmented vaccine hesitancy (Fadda et al., 2020). Global responses to the roll-out and distribution of vaccines remain unknown. Simultaneously, the uneven distribution of vaccines has put vaccine diplomacy on the agenda (Hotez and Narayan, 2021). Thus, we aimed to reveal people's opinions of vaccine diplomacy, which first came to public attention in December 2020 on Twitter. Vaccine diplomacy is vulnerable to becoming a diplomatic strategy on the bilateral table for politicians. We arranged and plotted the situation of the estimated vaccine production capacity and vaccination status (Fig. 6).
At present, in the international community, several countries, including China, the United States, Russia, the United Kingdom, Australia, Canada, and India, have developed COVID-19 vaccines ( Supplementary Fig. S4). Most countries have adopted multiple research methods simultaneously to ensure the speed and effect of vaccine research. China and the USA are conducting nearly 20 vaccine trials simultaneously, and the USA has the largest vaccine production capacity (Fig. 6). The USA and the UK have high productivity, high vaccination coverage and large export coverage ( Supplementary Fig. S5), which indicates that productivity is positively correlated with national power (Pannu and Barry, 2021). Comparably, the total vaccination numbers of China and the USA are very high, which reflects the strong execution capability of the governments of these countries. In contrast, lowand middle-income countries (LMICs) are mostly left behind by developed communities (Holder, 2021) and are devoid of vaccines. Hence, the complex bottlenecks in the allocation of approved vaccines require urgent attention.

Discussion
We conducted a study on the real-time public awareness of and behavioural responses to COVID-19 vaccines and global vaccination from November 2020 to April 2021, utilizing rich random-sampling metadata from Twitter (more than 19 million messages after deweighting). This study reveals the nature of the public opinion landscape and the underlying motivations behind the paradigms of public appeals in the context of the COVID-19 vaccination process.

Principal findings
Diverse topics, different voices. By revealing the paradigm shifts in global public opinion on the hottest vaccine-related topics, our results strongly suggest that concerns about vaccine diplomacy and the different policies of various countries cannot be ignored, as they may trigger people's discussion based on nationalism (Katz et al., 2021), which is consistent with the previous experiment (Lazarus et al., 2020). The findings also indicate that the public needs to be aware that the pandemic is global . Admittedly, in the COVID-19 pandemic, every nation is relevant. Compared with the first stage, currently, the spotlight is no longer on politicians, which indicates that they could not play a critical role in vaccination promotion, as expected.
Subsequently, through the results of the third stage, we found that variant-related topics, regardless of whether there was a strong positive sentiment, a negative sentiment, or just a general announcement, could trigger extensive discussion and involve an increasingly larger population (Tregoning, 2021). Consistent with the study (Forman et al., 2021), people who are sceptical with regard to the efficacy of vaccines in resisting new variants undermine confidence in vaccination and further exacerbate vaccine hesitancy. The findings remind us that authorities need to dynamically formulate policies to prepare for the influence of negative information on people and to avoid increasing vaccine hesitancy. For instance, the Ugandan government drafted researchers into a COVID-19 scientific committee to provide information for and prepare the country's prevention and treatment strategy for the pandemic in a timely manner. New Zealand brought epidemiologists, immunologists and public health experts into a Technical Advisory Group, which met twice per week to provide dynamic advice to prevent vaccine hesitancy (Haldane et al., 2021). The transmission of Twitter topics represents the public's expectations for solutions, from which rules should be found to provide targeted dynamic policies.
Different dimensions of global vaccination. Using the Google Trends indexes, we conducted a global opinions study on different dimensions of vaccination. People's intention to receive the COVID-19 vaccines hinges on complex dimensions, in which the role of government trust (Lazarus et al., 2020), the medical industry and vaccine companies all play a part (De Figueiredo et al., 2020;Davis et al., 2022). The results also indicate that people always have positive attitudes towards vaccination when they receive positive information (Hamel et al., 2020). Conversely, some negative dimensions, such as side effects, are vulnerable to becoming driving forces that result in low global vaccination rates. For instance, some people still believe rumours that vaccines cause autism (Geschwind, 2009) and increase the burden on the immune system. Similar studies in different countries using different methods have reached the same conclusions that side effects can diminish people's willingness to be vaccinated Paul et al., 2021). Blood clotting disorder has recently been triggered by vaccinations (Kupferschmidt and Vogel, 2021), which has aggravated people's worry and even caused panic. Additionally, people show different reactions to news coverage on national topics and their local leaders, especially in Russia and India. Such patterns indicate that as an integral component of preparedness, it is important to analyse the concerns and attitudes of people from different countries with regard to vaccination to achieve global herd immunity as soon as possible. For example, as the largest buyer of vaccines, India has already purchased 196 million doses of domestically manufactured vaccines, alleviating people's potential anxiety about vaccine distribution. Sri Lanka introduced a hotline for the public to submit complaints about difficulties in receiving essential services to keep abreast of public needs (Haldane et al., 2021).
Global responses to COVID-19 vaccines. Through rich randomsampling metadata from Twitter, we revealed the specific emotions expressed by people in regard to vaccination. The results indicate that populations of some countries, such as the USA, the UK, Canada and Australia, have certain similarities in their emotional attitudes towards vaccination (Fig. 3), and in combination with previous survey results (Ward et al., 2020), the findings show that emotions related to COVID-19 vaccines are partly linked to national policies.
Regarding the importance of dynamic governmental decisionmaking in a timely manner (Fig. 1), we further determined the possible negative impacts on social mobilization. After March, the rapid rise of highly transmissible variants across the globe put the world on watch again (Callaway, 2021;Mallapaty et al., 2021). Extending previous survey results (Loomba et al., 2021), our findings show that the public's intention to be vaccinated against COVID-19 is vulnerable to various external events, especially when members of the public receive misinformation. For example, some political figures (Hornsey et al., 2020;Andrew, 2021), as well as many high-profile partisan media outlets and social media influencers (Jeff, 2021), have propagated a great amount of misinformation harming vaccination promotion. The reason lies in the tendency of populations to react more strongly to negative information than to positive information to a certain extent (Soroka et al., 2019). In this case, misinformation is more likely to be amplified and spread. These findings underscore the tremendous undertaking of social mobilization, which must persuade hesitant people to be vaccinated if we are to achieve herd immunity to return to normalcy. Even then, pro-vaccination populations who said they would "definitely or probably" be vaccinated must follow through, and populations characterized by vaccine hesitancy or anti-vaccination attitudes who said that they did not plan to get vaccinated but could change their mind must be persuaded and motivated to act, which is consistent with the studies that vaccination willingness shows intertemporal change . Therefore, the looming challenge implies that the roll-out of pertinent policies is crucial for increasing vaccine coverage. The WHO has launched an initiative called "Myth Busters", in which the organization directly addresses misconceptions and misinformation (WHO, 2020). Australia has established strong responses with one voice as the source for accurate and clear information, partly avoiding the propagation of misinformation (Böl, 2021).
Simultaneously, due to the knock-on effects of global efforts to contain the long pandemic, the impact on people's mental health cannot be ignored (Kohrt, 2021). It is thought that the spread of new variants will reshape our routine daily life and potentially pose a threat to mental health (Kola et al., 2021). For example, in just the first few months of the pandemic in the USA, there was a significant increase in depression (18% rise) (Heffner et al., 2021). Women in Chile have been psychologically affected by the COVID-19 pandemic . However, it is gratifying that people show more positive emotions with increasing vaccination rates. This trend is meaningful for addressing pre-existing mental healthcare needs. Governments should pay more attention to the implications of citizens' mental health and implement psychosocial programmes following the best practices of the WHO and other international organizations. For instance, the World Psychiatric Association has published a position paper on the responsibility of psychiatrists to advocate for the needs of patients to governments and public health policy agencies in this global health crisis (Stewart and Appelbaum, 2020).
As a bonus of the experiment of fine-grained emotions across countries (Fig. 4), the results also reflect the many nuances of current bilateral relations. For instance, Americans generally worry about the way the Chinese government is fighting against COVID-19, while they take an open mind to engage in modest introspection with regard to local preparedness (Nature, 2021). From a macroscopic perspective, as concrete indicators, these emotional polarities indicate that the perceptions among countries are not positive. At present, the current multipolar interconnected world is far more complex than ever before (Hotez and Narayan, 2021). All countries should set aside their differences and fight shoulder to shoulder against this common threat to health. For instance, the decisive reason for Serbia's current success in vaccination is China, which has provided Serbia with millions of vaccine doses (Von Michael, 2021).
The results of further experiments on the relationship between emotions and vaccination status reveal that anti-vaccination attitudes among hesitant groups who have not been vaccinated for various reasons are harder to shift (Stacy and Kevin, 2021) (Fig. 5). In particular, the emergence of SARS-CoV-2 variants (Thompson and Pickrell, 2021) and some potential side effects of the vaccines, such as blood clotting (Karen, 2021) and myocarditis , have further affected hesitant people's acceptance of vaccination to a certain extent (Supplementary Fig. S3). Partially vaccinated populations remain hesitant about the timing of protection, which influences their decision to be fully vaccinated (Jillian et al., 2021). In addition, a large proportion of the vaccinated population reported that they were uninformed about the need to continue to take prophylactic measures, including mask wearing and avoiding crowds, which indicates that inconsistent public health messages are relevant for COVID-19 vaccine refusal (Paul et al., 2021). This means that authorities need to make augmented educational efforts to ensure that each of us is preparing to be fully vaccinated and is following prophylactic measures. For instance, the National Foundation for Infectious Diseases in the USA convened a multidisciplinary expert roundtable and developed a new report, COVID-19 Communications: Promoting Prevention Measures and Vaccine Confidence, to help leaders across sectors and in the government communicate effectively and encourage the public to follow evidence-based public health prevention measures to control the spread of COVID-19 (NFID, 2021).
Vaccine diplomacy: A challenge to restoring global order. Thus far, we are still in a dilemma-the vaccine production capacity is insufficient, and this has become the primary bottleneck in the global distribution of vaccines (Forman et al., 2021). To address the allocation of vaccines, on the one hand, wealthy countries need to abandon vaccine nationalism, which perpetuates the long history of powerful countries securing vaccines and therapeutics at the cost of LMICs (Katz et al., 2021;Adhanom, 2021). On the other hand, regarding political backlash, the public naturally expects more from the five permanent members of the UN Security Council. Under these circumstances, it is incumbent on these permanent members to assume greater international responsibility. Powerful countries, such as the permanent members, must lead by example and carry out effective programmes of international cooperation and assistance (Mallapaty et al., 2021). Such cooperation is not only a matter of social justice but also a sound pragmatic response for ending such a pandemic. At the same time, it is in part a reflection of the importance of financing needs for COVID-19, and consistent with some studies (Neumann et al., 2021), the pandemic might have ended considerably sooner if effective therapies and vaccine platforms for COVID-19 had been invested in earlier and more widely. Accordingly, powerful countries need to innovate low-cost manufacturing and supply chains to expand global production. Decisively controlling the pandemic requires the provision of billions of vaccines to the populations of LMICs (Hotez and Narayan, 2021). As Ursula von der Leyen, president of the European Commission, explained, "A global pandemic requires a world effort to end it-none of us will be safe until everyone is safe." (Ghebreyesus and Leyen, 2020).
Strategies to promote vaccination. Based on the guiding principle of the coevolution of science and vaccine-promoting policy , we further emphasize the significance of formulating dynamic policies in a timely manner to assist authorities in making reasonable decisions. The global community has often proven itself to be unprepared to respond to worldwide health emergencies that have profound societal and economic implications (Yamada et al., 2016). Hence, we propose the following recommendations for government action based on the revealed status quo and other national and international public health guidance for preparedness for emergencies during vaccine promotion (Table 1).
Oscillating volition of the public. With concrete indicators, our study is characterized by a striking feature: the collective response to COVID-19 vaccines is vulnerable to various external events during the vaccine-promoting process. The public's intention to be vaccinated is therefore not static; it is highly responsive to current information and sentiment around COVID-19 vaccines, as well as the state of the pandemic and the perceived risk of contracting the disease. Governments or medical authorities must be prepared to explain the possible individual consequences of vaccine-related negative emergencies to the public in a timely manner to reassure the public.
Crisis behind the high vaccination rates. While large-scale vaccine rejection threatens global vaccination, large-scale acceptance with local vaccine rejection can also have negative consequences for herd immunity. For instance, the UK and the USA already have high vaccination rates; however, parts of the remaining unvaccinated populations possess a stronger vaccine hesitancy tendency . Simultaneously, populations who cannot be vaccinated due to medical conditions are constantly exposed to danger and discrimination, which can occur at any time without vaccine protection. In contrast, Israel, which has a high vaccination rate and has sustained a continuous decline in incidence rates in all age groups (Balicer and Ohana, 2021), may be crossing over to the other side of the pandemic. However, it is difficult to extrapolate from the Israeli experience a universally "safe" vaccine uptake threshold because such a threshold is likely to be setting specific. To promote global vaccination to reach herd immunity thresholds, it is necessary to adopt targeted persuasion for people with different attitudes towards COVID-19 vaccines. For those who are still hesitant about vaccines, this means finding trusted vaccines or medical advocates to provide education and communication, partnering with community or opinion leaders, and overcoming identity hurdles. For those who are still on the fence, this means further popularizing vaccination knowledge, creating reminders and salience via social media, creating triggers to prompt an "act now" mentality, and adding incentives and convenience. For those who have a positive attitude, this means ensuring the accessibility and ease of vaccination, for example, through proactive outreach through call centres and easy-to-use mobile device apps and electronic health records to expedite intake at vaccination sites. For those who cannot be vaccinated due to medical conditions, the government should introduce strong policies for protection, the community should focus on the publicity of people with medical conditions and improve the selfprotection awareness of these people, and medical institutions should allocate more resources to these people.
Public mistrust in governments. Governments are heavily promoting vaccines to achieve vaccination rates, and while this initiative is reasonably effective, the poor misinformation response has multiplied the public's doubts about vaccines. Misinformation about the pandemic has circulated widely on social media platforms (Loomba et al. 2021). For example, there is misinformation that 5 G mobile networks are linked to the virus, that vaccine trial participants have died after taking a candidate COVID-19 vaccine, and that the pandemic is a conspiracy or a bioweapon (Megget, 2020;Pennycook et al. 2020). Governments should make more efforts to clarify this misinformation. Notably, misinformation will not disappear because of a strong crackdown by the government. In contrast, it will become more hidden, making those who are already involved in the misinformation more convincing. If misinformation is not properly addressed, the government's credibility with the public will be greatly diminished and detrimental to any subsequent policy implemented by the government. The government cannot just crack down and obscure misinformation; rather, it should use propaganda tools to clarify it. To build confidence in vaccination, it is important that regulators, companies and their research partners keep the promises they have made to ensure transparency, publish data and engage in an open discussion on those data as they arrive.
Imbalanced vaccine distribution. As permanent members of the UN Security Council, China, the USA, the UK and Russia have made efforts to fight the pandemic. They have carried out vaccine experiments in many countries or have been recognized by The Lancet (Logunov et al. 2021;Jones and Roy, 2021). Notably, although the success of vaccine development has brought victory in the fight against COVID-19, the distribution of global vaccines is severely polarized (Mallapaty et al. 2021). Many LMICs are still experiencing difficulties in supplying the vaccine; however, highincome countries (HICs) control almost a billion doses of COVID-19 vaccines more than they actually need, which has attracted criticism from LMICs and the international community, which has said that the actions of HICs have undermined the efforts of the international community to fight the pandemic and that the hoarding of vaccines violates declared human rights commitments (Wouters et al. 2021). Affected by the outbreak, it is conceivable that many countries are thinking only about their own vaccine use and are hoarding. However, the international community is still facing the next wave of the pandemic. With its rapid international circulation in modern society, the outbreak will not disappear because of the success of a single country in the fight against the disease. Abandoning national barriers and shouldering responsibility, especially in the case of the permanent members of the UN Security Council, are the foundation of successfully resisting the COVID-19 pandemic.

Conclusion
In the critical period when a new pandemic wave could undermine global efforts at any time, articulating social paradigms of collective responses to COVID-19 vaccination is paramount for achieving herd immunity. With an emphasis on the lens of the leading social media platform, Twitter, we scrutinized collective responses to COVID-19 vaccines and pertinent discourses in sociocultural paradigms to uncover the collective propensities and potential consequences.
In response to the public's wavering volition in regard to vaccination, governments should be prepared to resolutely build confidence in vaccination and avoid vaccine hesitancy. Simultaneously, despite the high vaccination rates in some HICs, such as the USA and the UK, the population with anti-vaccination attitudes cannot be ignored, and it is crucial to adopt targeted measures for people with different opinions towards COVID-19 vaccines. To achieve a global victory against the pandemic, LMICs where government enforcement is lacking, vaccines are in short supply and injection rates are low deserve more attention in the current international confrontation. HICs should shoulder the Table 1 Overview of challenges of and strategies for promoting global vaccination.

Challenges Strategies
Oscillating volition of the public Speed of communication is necessary in an emergency. Public health literacy should be enhanced. Public health bodies should maintain robust national, subnational, and local epidemiological databases, updated and reported daily. Crisis behind the high vaccination rates There should be targeted persuasion for people with different attitudes.
Mental health outreach services should be expanded. Appropriate measures should be taken to protect members of vulnerable groups such as the elderly, poor, migrant, and homeless populations.

Public mistrust in governments
The government should do more to clarify misinformation. Authorities should provide the public health grounds for their decisions. Pandemic preparedness teams consisting of public health and medical experts should be established to coordinate national responses. Imbalanced vaccine distribution National barriers should be abandoned, and international responsibilities should be assumed. More attention should be given to densely populated, low-resource areas.
There should be collaboration with other countries, the WHO, and other international bodies to respond to the pandemic.
responsibility for global resistance to the pandemic, seeking cooperation and addressing the intractable problem of vaccine shortages and injection.

Data availability
Data and code associated with this study are available in the GitHub (https://github.com/yinfulian/vaccination-campaign.git).