ROLE OF ARTIFICIAL INTELLIGENCE AND BIG DATA IN COVID-19 PANDEMIC OUTBREAK: PART I

The very first infected novel coronavirus case (COVID-19) was found in Hubei, China in December 2019. The COVID-19 pandemic has spread over many countries and areas in the world, and has significantly affected every aspect of our daily lives.

Artificial Intelligence(AI) thriving technology for many intelligent applications in various fields. Some high-profile examples of AI are autonomous vehicles (e.g., self-driving car and drones) in automotive, medical diagnosis and telehealth in healthcare, cybersecurity systems (e.g., malware and botnet detection), AI banking in finance, image processing and natural language processing in computer vision. Among many branches of AI, machine learning (ML) and deep learning (DL) are two important approaches. Generally, ML refers to the ability to learn and extract meaningful patterns from the data, and the performance of ML-based algorithms and systems are heavily dependent on the representative features. In the meanwhile, DL is able to solve complex systems by learning from simple representations. DL has two main features: 1) the ability to learn the right representations shows one feature of DL and 2) DL allows the system to learn the data from a deep manner, multiple layers are used sequentially to learn increasingly meaningful representations.

AI offers a powerful tool to fight against the COVID-19 pandemic. For example, the scientists in developed a DL model to identify existing and commercial drugs for the “drug-repurposing” (also known as drug repositioning), i.e., finding a rapid drug strategy using existing drugs that can be immediately applied to the infected patients. This study is motivated by the fact that newly developed drugs usually take years to be successfully tested before coming to the market. Although findings in this study are currently not clinically approved, they still open new ways to fight the COVID-19 disease. The work in proposed using the deep generative model for drug discovery (which is defined as the process of identifying new medicines). The COVID-19 protease structures generated from the DL model in this work would be further used for computer modelling and simulations for the purpose of obtaining new molecular entity compounds against the COVID-19 coronavirus. Utilizing DL for computed tomography (CT) image processing, the authors in showed that their proposed DL model when training with 499 CT volumes and testing on 131 CT volumes, can achieve an accuracy of 0.901 with a positive predictive value of 0.840 and a negative predictive value of 0.982. This study offers a fast approach to identify the infected COVID-19 patient, which may provide great helps in timely quarantine and medical treatment. The last example is the use of AI (i.e. an auto encoder based method) in real-time predicting dynamics (e.g. the number of infected cases, ending time and trajectory of the COVID-19 pandemic) of the COVID-19 outbreak in China. The high accuracy of the AI-based approach proposed in is helpful in monitoring the COVID-19 outbreak and improving the health and policy strategies. Along with the applications mentioned above, the involvement of giant techs is needed because researchers, doctors and scientists can be effectively supported to expedite the research and development of COVID-19 virus. Recently, IBM announced that they are now providing a cloud-based research resource that has been trained on a COVID-19 open dataset (CORD-19), which is a collection of research articles related to COVID-19. Moreover, IBM has adopted their proposed AI technology for drug discovery, from which 3000 novel COVID-19 molecules have been obtained, officially reported. Another support is from the White House Office of Science and Technology Policy, the U.S. Department of Energy and IBM with the development of COVID- 19 HPC Consortium, which is open for research proposals concerning COVID-19. Another example is the Coronavirus International Research Team (COV-IRT), a group of scientists who are developing vaccines and therapeutic solutions against COVID-19.

1) Definition and Characteristics

The rapid development of the Internet of Things (IoT) results in a massive explosion of data generated from ubiquitous wearable devices and sensors. The unprecedented increase of data volumes associated with advances of analytic techniques empowered from AI has led to the emergence of a big data era. Big data has been employed in a wide range of industrial application domains, including healthcare where electronic healthcare records (EHRs) are exploited by using intelligent analytics for facilitating medical services. For example, health big data potentially supports patient health analysis, diagnosis assistance, and drug manufacturing. Big data can be generated from a number of sources which may include online social graphs, mobile devices (i.e. smartphones), IoT devices (i.e. sensors), and public data in various formats such as text or video. In the context of COVID-19, big data refers to the patient care data such as physician notes, X-Ray reports, case history, list of doctors and nurses, and information of outbreak areas. In general, big data is the information asset characterized by such a high volume, velocity and variety to acquire specific technology and analytical methods for its transformation into useful information to serve the end users, i.e, big data in digital twin technologies. The three characteristics of big data are summarized as follows.

● Volume: This feature shows the huge amount of data that can range from terabytes to exabytes. According to a Cisco’s forecast, the data traffic is expected to reach 930 exabytes by 2020, a seven-fold growth from 2017.

● Variety: It refers to the diversity and heterogeneity of big data. For example, big data in healthcare can be produced from healthcare users (i.e. doctors, patients), medical IoT devices, and healthcare organizations. Data can be formatted in text, images, videos with structured or un-structured dataset types.

● Velocity: It expresses the data generation rate that can be calculated in time or frequency domain. In fact, in industrial applications like healthcare, data generated from devices is always updated in real-time, which is of significant importance for time-sensitive applications such as health monitoring or diagnosis.

Big data has been proved its capability to support fighting infectious diseases like COVID-19. Big data potentially provide a number of promising solutions to help combat COVID-19 epidemic. By combining with AI analytics, big data helps us to understand the COVID-19 in terms of outbreak tracking, virus structure, disease treatment, and vaccine manufacturing. For example, big data associated with intelligent AI-based tools can build complex simulation models using coronavirus data streams for outbreak estimation. This would aid health agencies in monitoring the coronavirus spread and preparing better preventive measurements. Models from big data also supports future prediction of COVID-19 epidemic by its data aggregation ability to leverage large amounts of data for early detection. Moreover, the big data analytics from a variety of real-world sources including infected patients can help implement large- scale COVID-19 investigations to develop comprehensive treatment solutions with high reliability. This would also help healthcare providers to understand the virus development for better response to the various treatment and diagnoses.

Based on the above analysis, we want to highlight that big data analytics is the process of collecting and analyzing the large volume of data sets to discover useful hidden patterns and other information, e.g., COVID-19 data discovering. Moreover, AI (and explainable AI) aims to apply logic and reasoning to build human intelligence that can mimic the function of a machine for learning, classifying, and estimating possible outcomes, e.g., COVID-19 symptom classifications. The potential applications of each technology in fighting the COVID-19 pandemic will be explained and discussed in the following sections through a number of practical use cases.