Introduction

The coronavirus disease 2019 (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is raging throughout the world. As of February 21, 2021, the cumulative confirmed cases reported to the World Health Organization (WHO) exceeded 110,763,000, the cumulative number of deaths was 2,455,000, and the COVID-19 pandemic was still the most severe global health emergency [1]. Upon the outbreak of COVID-19, China was the first country to share SARS-CoV-2 genome sequence data with the WHO and the international community [2]. With the help of scientific technologies and resources, the Chinese government adopted the most comprehensive, stringent and thorough prevention and control measures in an attempt to bring the virus under control and finally achieved an initial triumph. The new generation of information technology, represented by big data and artificial intelligence (AI), has been widely applied in epidemic prevention and control, diagnosis and treatment as well as management decisions and has played valuable roles in the fight against the COVID-19 pandemic. The COVID-19 pandemic presents a unique background to the emergency response, diagnosis and treatment of sudden public health events in that the COVID-19 pandemic has had different forms along with the evolution of the disease, and it has also required taking the ever-changing epidemic situation, the treatment conditions of patients and their response to various treatment decisions into consideration. Although science and technology have developed, there are still challenges with respect to epidemic surveillance, accurate prevention and control, effective diagnosis and treatment, and timely judgement. Epidemic assessment usually relies on the real-time monitoring of disease control systems, and big data and related technologies promote multi-dimensional data integration and epidemiological analysis. In particular, the qualitative interpretation of AI-powered lung imaging includes the description of lung imaging changes over time, the prediction of clinical manifestations and outcomes, and the assessment of the effects of disease and treatment on systemic organs [59].

The first novel coronavirus strain was successfully isolated [60] on January 24, 2020, which means that the vaccine strain could be cultured to prepare a vaccin [61, 62]. With the improvement of science and technology, based on the accumulating data on severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) and the support and promotion of relevant national policies, five COVID-19 vaccines have initiated human trials in China, and advanced AI technology has played an important role in accelerating the speed and improving the quality of vaccine research and development [63, 64].

First, in the discovery stage of vaccines, it often takes several years to perform manual analysis. However, machine learning can complete data analysis and matching of potential pharmacological relationships, screen drug action targets from massive data, find biomarkers, simulate synthesis of lead compounds, and conduct multiple rounds of analysis on the structure–activity relationship of vaccines, which greatly improve the research speed and effectiveness of vaccines [65, 66]. Second, the application of AI in pharmacokinetics, safety pharmacology and toxicology, preparation development and phase I, II and III clinical trials accelerates the research speed and improves the quality and effectiveness [67,68,69]. For example, in pharmacokinetic studies, a neural network system has been used to simulate the absorption, distribution, metabolism and excretion in animals and produce the optimized pharmacological structure to improve the accuracy and speed of experiments [70,71,72]. In toxicology studies, a deep learning method was used to analyse the toxicity mechanism, mode, pathway and effect of various vaccines, which could reduce the pressure of clinical trials and save study cost [65, 73]. Third, AI can promote intelligent manufacturing for large-scale vaccine production. Based on the deep integration of new-generation information and communication technology and advanced manufacturing technology, a new production mode with functions featuring self-perception, self-learning, self-decision-making, self-implementation and self-adaption functions throughout production, management and service can greatly improve the production, quality and efficiency of vaccines and promote the rapid entry of vaccines into the market [74, 75]. Finally, AI can intelligently track and monitor the use of vaccines after marketing. A vaccine monitoring system for epidemic prevention can be established to realize remote, digital and dynamic tracking and monitoring and collect data for use in a timely manner, thus providing guidance for further research and development, production and application of vaccines [76,77,78].

AI in epidemic prevention and control and management decisions

Immediately after the COVID-19 outbreak, the central authorities led by Chinese President ** put people’s life and health first and fought to block the spread of the virus, putting forward the overall principles of supporting confidence, strengthening unity, ensuring science-based control and treatment, and implementing targeted measures [79]. After the COVID-19 virus database developed by the China National Center for Bioinformation (CNCB) was officially launched, the genome and variant information of the COVID-19 virus were released in time, [80] which strongly promoted international cooperation in epidemic prevention and control [81,82,83]. Based on the big data of the epidemic situation in Wuhan and the whole country, the water, land and air transportation in Wuhan were closed in time, and the government rallied 346 national medical teams, consisting of 42,600 medical workers and 925 public health professionals, to the immediate aid of Hubei Province and Wuhan city. [84].

In the process of epidemic prevention and control, the application of big data and AI provides support in timely research and judgement of the epidemic trend, epidemiological investigation, identification of every infected person, and close contact tracing and isolation [85, 86]. By establishing a large-scale cross-regional, cross-departmental, cross-industrial and cross-structural shared database to provide risk data for epidemic prevention and control in accordance with the law, accurately identify different risk populations, and predict epidemic risks in different regions, AI provides effective services for orderly movement of people and resumption of work and production [87, 88]. Through real-time 5G video conversation, epidemiological investigation teams in remote mountainous areas can interact with high-level experts on a platform thousands of kilometres away [89,90,91,92]. With the authorization of individual citizens, [93] the personal “Health QR Code” and “Communication Big Data Travel Card” have been promoted nationwide as certificates for personnel travel, resumption of work and education, daily life and access to public places. Traffic control and classified disposal are carried out based on the query results, and accurate identification, precise implementation and precise prevention and control are realized at different levels in different regions [94]. An “epidemic map” is drawn using big data technology, and the specific location, distance and number of people for epidemic transmission are indicated by using the name, address and location of the community members, which provides convenience for the public to prevent COVID-19 [95,96,97].

Six months after the WHO announced that the COVID-19 pandemic constituted a “public health emergency of international concern” on January 31, 2020, the WHO issued a statement for the third time on August 1, 2020, declaring that the COVID-19 pandemic was still a “public health emergency of international concern”. It is expected that the epidemic will last for a long time, and it is necessary for countries around the world to strengthen cooperation and take long-term response measures [98]. China’s successful experience in epidemic prevention and control has been recognized, learned from and applied by many other countries, and their fight against the virus has also achieved good results [99,100,101].

Challenge and future

Although big data and AI have been successfully applied in the prevention and control of the COVID-19 pandemic, they still face obstacles and challenges before widespread clinical application [102]. With the global prevalence of COVID-19, the management of epidemic data and medical data has been a major obstacle to the development of intelligent prevention and control and clinical solutions [103, 104]. Big data is a necessary prerequisite for AI training [105]. Standardized data are particularly important in the field of health care, especially multi-region, multi-system and multi-source heterogeneous data [106]. Although epidemic data disclosure is an encouraging step forward, the research community has not yet reached a consensus on specific data sets, and there is a lack of shared and collaborative data sets that are proven to be universal, repeatable and standardized for sharing and collaboration. Governments, professional organizations and institutions at all levels should be encouraged to share validated data to support the development of AI algorithms [107]. Another obstacle is the interpretability of AI, the ability to question the reasons behind particular outcomes and the expectation of failure [108, 109]. With the emergence of deep learning and prediction models, applications must be constantly updated according to the use of real-world data (RWD) in training. The application of AI requires trained professionals, who are expensive in time and cost; [110, 111] however, it also raises concerns about personal privacy and data security [112, 113].

The ultimate goal of health care is to prevent and control diseases. Establishment of an accurate risk model is vital to guide the strategy of risk adjustment. Therefore, AI systems must meet strict clinical test requirements, which will become an important direction of future development [114, 115]. With the expanded sources of big data, the data sources are not limited to electronic medical records (EMRs) and electronic health records (EHRs). Currently, data from wearable devices, mobile phones, social media and others are also available [116, 117]. AI is very suitable for integrating parallel information flows from biology, the population and society to improve the prediction models and medical intervention for patient outcomes, including quality control and risk assessment [118, 119].

As the application potential of big data and AI in the medical field is increasingly confirmed, there are still many directions for the transition from current application to conventional clinical practice. For example, for medical image analysis, the accuracy and prediction ability of the AI method need to be significantly improved. To replace the workflow of clinicians, it needs to be proven that the effectiveness of the AI method is better than that of human experts in a control study [120]. In addition, the application of AI in monitoring health resources and outcomes may improve efficiency and reduce costs. Like any innovative technology, the development of big data and AI will certainly be beyond the imagination of humankind [121].

Conclusion

On February 4, 2020, the Ministry of Industry and Information Technology of China released an initiative that called on giving full play to the utility of AI in the fight against the novel coronavirus outbreak [38] Facing the challenge of COVID-19, breakthroughs in the application of AI have been accelerated, and it has been again verified that AI technology has a profound effect on current medical health and human behaviour and will play a greater role in the fight against the COVID-19 pandemic and other events that may be faced in the future [122, 123]. At present, the main applications and trends are as follows. first, big data analysis will be more intelligent. It will be easier to find epidemics at the early stage, track close contacts, improve diagnosis and treatment efficiency, predict the possible evolution of viruses in the future and develop more effective and long-lasting vaccines by analysing massive and real-time data with machine learning and deep learning [124,125,126]. Second, epidemic prediction will be more accurate. Most AI algorithms are prediction oriented, and the unique skill of AI-assisted epidemiological research will be used to establish a system that can accurately predict when and where future outbreaks will occur and how human behaviours will change to improve the ability to detect and respond to epidemic risks [127, 128]. Third, detection and prevention of the pandemic will be automated. Computer vision technology will be used to screen individuals with COVID-19 symptoms such as fever, and facial recognition technology will be used to track the activities of individuals with symptoms in a crowd and inform relevant departments or managers of the statistical data and probability of virus transmission [129, 130].