Identifying COVID-19 Cases Rapidly and Remotely Using Big Data Analytics

Introduction

As the number of COVID-19 hospitalizations and fatal cases continuously increases, the pandemic challenges healthcare systems globally in many terms, such as shortages in personal protective equipment, significant increases in hospitality demands, the lack of medical capacities, and highly extensive testing among the population. Identifying the COVID-19 cases is the first step through the whole medical decision-making process that will support in forms of determining confirmed cases, generating appropriate disease control policies, allocating healthcare resources, and making clinical decisions and treatment planning. Reverse transcriptase polymerase chain reaction (RT-PCR) tests, as one of the most validated and common-used medical diagnosis for the novel coronavirus, have been applied since the pandemic. However, major challenges remain in COVID-19 tests, for example, shortages in testing capacity, time consuming, and increasing social costs. Traditional diagnostic tests, such as molecular tests and antigen tests, require people to be tested in designated places with a person-to-person contact manner, which may increase the risks of exposure to the virus. Specifically, such a COVID-19 diagnostic test produces many problems in the lockdown scenario, e.g. difficulties of allocating test capacities, challenges of managing test queues, and risks of the increasing human mobility. Despite the at-home test kits available in pharmacies, the tests results are usually available in many days after labs receiving the sample, which may not be applicable for those who need the test results immediately. Individuals do not know under what circumstances they should be tested, typically, when they may not fully understand critical issues related to availabilities, characteristics, and strategies around the COVID-19 test. Such controversies and conflicts contribute to the increased burdens on healthcare systems and delay critical preventive measures on disease control policies.

Encumbered by exponential increase of COVID-19 cases, lockdowns, and travel restrictions, the role of effective screening technologies should be considered in responding to the pandemic crisis. Cutting-edge technologies such as big data and artificial intelligence (AI) have been substantially applied as evidence-based decision support providers in epidemiological modeling and disease control management. Predictive analytics and machine learning techniques have been widely used to assist mining insightful information from various data sources, which can be transformed through knowledge discovering processes for appropriate actions. Such big data predictive analytics frameworks based on machine learning and deep learning have been developed over hybrid data sources with real-world applications for analyzing a variety of diseases, such as dengue fever, SARS, bird flu, H1N1, and MERS (Bansal et al., 2016; Chae et al., 2018; Souza et al., 2020). The same ideology can be potentially incorporated in designing the effective screening system against COVID-19, especially in the early stage of the outbreak, in assisting medical professionals to identify confirmed cases quickly. Several studies related to predictive approaches with features, such as clinical symptoms, laboratory testing results, and computer tomography (CT) scans, have been proposed in terms of predictive analytics of early symptoms (Tostmann et al., 2020), suspected case identifications with the AI-based diagnosis aid tool (Feng et al., 2020), and automated detectors through deep learning-based CT image analysis (Gozes et al., 2020). Despite the effectiveness of automatic identifications relied on such features, existing models based on clinical information from hospitalized patients are not applicable for the general population because of the limited sample size and the restricted hypothesis. Besides, some clinical data, such as CT images and laboratory test results, rely on medical diagnosis and assay that require individuals be tested onsite with medical professionals, thereby may not be adaptable to the principle of the effective screening for COVID-19 under the lockdown scenario.