ROLE OF ARTIFICIAL INTELLIGENCE AND BIG DATA IN COVID-19 PANDEMIC OUTBREAK : CHALLENGES, LESSONS AND RECOMMENDATIONS : PART II

As the COVID-19 outbreak is booming, various approaches have been taken to control this outbreak. For example, lockdown, social distancing, screen and testing at a large scale. In this way, regulatory authorities occupy a crucial role in defining policies that can encourage the involvement of residents, scientists and researchers, industry, giant techs and large firms, as well as harmonizing the approaches executed by different entities to avoid any barriers and obstacles in the way of preventing the COVID-19 disease. Regarding this challenge, many attempts have been made from the first confirmed COVID-19 to current situation. An example is from the quarantine policy in Korea which effective from 1st April 2020. More specifically, all passengers entering Korea are required to be quarantined for 14 days at registered addresses or designated facilities. Moreover, all passengers are required to take daily self-diagnosis twice a day and send the reports using self-diagnosis apps installed in their mobile phones. An “AI monitoring calling system” has been implemented by the Seoul Metropolitan Government to automatically check the health conditions of the people who do not have any mobile phones and/or have not installed the self-diagnosis apps. Another effort is from the collaboration between the Zhejiang Provincial Government and Alibaba DAMO Academy who developed an AI platform for automatic the COVID-19 testing and analysis with an accuracy of 96% within 20 seconds only. Alibaba DAMO Academy has worked with many Hospitals in China to build AI systems for detecting the COVID-19 infected cases, where Alibaba DAMO Academy is responsible for designing AI algorithms and hospitals are responsible for providing CT scans of COVID-19 cases. This system has been used by more than 20 hospitals in China.

In order to make AI and big data platforms and applications a trustful solution to fight the COVID-19 virus, a critical challenge arises from the lack of standard datasets. As surveyed results shows that many AI algorithms and big data platforms have been proposed, but they are not tested using the same dataset. However, we cannot decide which algorithm is better for the virus detection since two datasets with different numbers of samples are used. Furthermore, most datasets found in the literature have been made thanks to individual efforts, e.g., the authors collect some datasets available in the Internet, then unify them to create their own dataset and evaluate their proposed algorithms. To overcome this challenge, government, giant firms and health organizations (e.g., WHO and CDC) play a key role as they can collaboratively work for high quality and big datasets. A variety of data sources can be provided by these entities, e.g., X-ray and CT scans from the hospitals, satellite data, personal information and reports from self-diagnosis apps.

Right now, important things are keeping people healthy and soon controlling the situation; however, how personal data secure and private is still needed and should be investigated. A showcase of this challenge is the scandal of the Zoom video conferencing app over its security and privacy issues. In the pandemic, authorities may request their people to share their personal information, for example, GPS location, CT scans, diagnosis reports, travel trajectory, and daily activities, which is needed to control the situation, make upto date policies, and decide immediate actions. Data is a must to guarantee the success of any AI and big data platforms; however, normally people do not want to share their personal information, if not officially requested. There is a trade off: privacy/security and performance. Many technologies are available to solve the privacy and security issues during the COVID-19 pandemic. We consider some potential solutions which can serve as research directives in the future. ● Blockchain: Basically, blockchain is defined as a decentralized, immutable and public database, where each transaction is verified by all the nodes in the network, which is enabled by consensus algorithms. Blockchain has found its success for various healthcare applications, so it is possible to deploy blockchain based solutions to improve the user security and data privacy during the COVID-19 outbreak period. MiPasa is one of the projects that combines two emerging technologies from IBM: blockchain and cloud computing. The purpose of this project is to provide reliable, accessible, and high-quality data to the communities. ● Federated learning (FL): Traditionally, the data should be collected and stored centrally to training Deep Learning(DL) models. Federated learning offers a new solution in which a majority of personal data is not required to be transmitted to the central server. Applying to the AI based framework using mobile phones for diagnosing COVID-19 case, each phone can train its own Deep Learning(DL) model using local data. Mobile phones transmit their trained models to a central server, which is responsible for aggregating to make a global model that is then disseminated to all the mobile phones. We note that a phone is not limited to mobile phone only, it can be a server at the local health department, whereas the aggregation server can be a global cloud like Microsoft Azure and Amazon Web Services. ● Incentive mechanisms: A large and reliable dataset lays a foundation for AI and big data platforms contend with the COVID-19 outbreak. Therefore, there is a need for incentive design to call the participation of more people and entities in contributing their own data. Incentives are needed due to the following reasons: 1) a massive quantity of data are available from people/entities, who may be not requested by the governments to provide their data, and 2) the quality of data should be guaranteed in order to improve the accuracy and performance of learning models. Such incentive mechanisms for healthcare, wireless communications, transportation, etc.

Reviewing the state-of-the-art literature, we find that AI and big data technologies play a key role in combating the COVID-19 pandemic through a variety of appealing applications, ranging from outbreak tracking, virus detection to treatment and diagnosis support. On the one hand, AI is able to provide viable solutions for fighting the COVID-19 pandemic in several ways. For example, AI has proved very useful for supporting outbreak prediction, coronavirus detection as well as infodemiology and infoveillance by leveraging learning-based techniques such as Machine Learning (ML) and Deep Learning (DL) from COVID-19 centric modelling, classification, and estimation. Moreover, AI has emerged as an attractive tool for facilitating vaccine and drug manufacturing. By using the datasets provided by healthcare organizations, governments, clinical labs and patients, AI leverages intelligent analytic tools for predicting effective and safe vaccine/drug against COVID-19, which would be beneficial from both the economic and scientific perspectives. On the other hand, big data has been proved its capability to tackle the COVID-19 pandemic. Big data potentially provides various promising solutions to help fight the COVID-19 pandemic. By combining with AI analytics, big data helps us to understand the COVID-19 in terms of virus structure and disease development. Big data can help healthcare providers in various medical operations from early diagnosis, disease analysis to prediction of treatment outcomes. With its great potentials, the integration of AI and big data can be the key enabler for governments in fighting the potential COVID-19 outbreak in the future. Some recommendations can be considered to promote the COVID-19 fighting. First recommendation, AI and big data-based algorithms should be optimized further to enhance the accuracy and reliability of the data analytics for better COVID-19 diagnosis and treatment. Second recommendation, AI and big data can be incorporated with other emerging technologies to offer newly effective solutions for fighting COVID-19. For example, data analytic tools from Oracle cloud computing have been leveraged to design vaccine(a new vaccine candidate) against the COVID-19 virus. More interestingly, recent studies have showed that 5G wireless technologies (e.g., drones, IoT, localization) can be utilized to combat the COVID-19 pandemic via a number of applications, such as delivery of testing samples, goods transport, social distancing, and people’s movement monitoring for outbreak tracking. Final recommendation, non-technology measures such as social distancing restrictions still play a highly important role in slowing down the virus spread and thus need to be implemented effectively under the management of government agencies, aiming to control the COVID-19 pandemic in the future.