Resolving Covid-19 with Blockchain and AI: a Systematic Review

The overarching goal of this review article is to offer a comprehensive overview of the research on the application of blockchain technology and artificial intelligence in the context of COVID-19 epidemiology. This includes systematic mapping as the chosen research methodology. Before delving into the rationale for employing blockchain and AI to address COVID-19, it is imperative to acknowledge the limitations of current healthcare systems. The second step involves locating all relevant scientific literature pertaining to the investigation’s subject matter. Papers addressing the elements of blockchain technology and AI technology relevant to combatting the COVID-19 epidemic have been categorized to support efforts to combat the pandemic. In order to filter and select the most pertinent technical literature, we employed search methods focused on blockchain, machine learning, and deep learning. Our searches were conducted in scientific databases, with a focus on peer-reviewed, high-quality papers presented at conferences, workshops, symposiums, and other research-related events, as well as those published in books and journals. We utilized scientific databases such as IEEE Xplore, Digital Library, Springer Link, and Science Direct to retrieve relevant papers. The fourth and final step involves title-based screening of each linked paper, during which subpar content, unavailable texts, and non-English-language papers were excluded while selecting the most relevant publications.

2. Background

This research study explores promising research avenues at the intersection of blockchain and artificial intelligence (AI) for managing the COVID-19 pandemic, offering a comprehensive guide for researchers.

2.1. Corona Pandemic

The COVID-19 pandemic originated in December 2019 in Wuhan, a city with a population of 11 million and the capital of China’s Hubei Province. The new coronavirus is believed to have originated in bats, supported by various pieces of evidence, although the exact intermediate species remains uncertain. COVID-19 primarily affects the respiratory system, leading to symptoms ranging from mild, such as a runny nose or cough, to severe cases that impede breathing. Common symptoms include fever, cough, and fatigue. While the majority experience mild symptoms, older individuals and those with preexisting conditions are more vulnerable. The World Health Organization reported 959,116 fatalities and 30,949,804 confirmed cases of COVID-19 worldwide. The pandemic has rapidly spread, particularly in the US, Southeast Asia, and Europe, significantly disrupting daily life and economic growth. Cities have been under lockdown, people’s movements restricted, and businesses forced to pause operations, with lasting repercussions on the global economy (Ashby et al., 2020). This viral outbreak has emerged as one of the most significant threats to financial markets and the global economy.

2.2. Blockchain and AI Technologies

The foundational technology of Bitcoin is commonly referred to as blockchain. Decentralization is a fundamental aspect of blockchain, wherein data is distributed across a network of nodes rather than residing in a single location. This decentralized structure imparts exceptional robustness and security by eliminating single points of failure. The transparency of blockchain is crucial for its overall operation, and consensus mechanisms facilitate this transparency. Consensus involves a set of rules that ensure agreement among all participants regarding the current state of the distributed blockchain ledger. Blockchains can be categorized as either public (permissionless) or private (permissioned) based on the chosen classification (Kadadha et al., 2020). Public blockchains allow anyone to conduct transactions, interact with other users, and participate in the consensus process. Bitcoin and Ethereum represent notable applications of public blockchains Conversely, private blockchains are restricted to a specific community within a large organization, where participants must adhere to an authorization process. Every blockchain comprises three fundamental building blocks: the block, consensus algorithms, and distributed ledger (database). Figure 2 illustrates the operation of a blockchain. Any type of record, including scientific data, can be executed and stored on the blockchain. A blockchain is formed by linking numerous blocks, each consisting of multiple transactions. An interconnected chain is created by including a header section on each block containing the hash of the preceding block. The primary advantage of blockchain lies in its ability to maintain the chronological order of data, including COVID-19-related information. These data points are securely stored in chronological order on the blockchain.

Blockchain’s proof of work (PoW) and the underlying consensus mechanism ensure the network’s security and transparency, without the involvement of intermediaries. The distributed ledger is resistant to tampering because each record possesses a unique cryptographic signature independent of a specific date. In terms of consensus algorithms, the most widely used method for validating blocks across the blockchain should not be controlled by any single entity, allowing all users with equal rights to govern each block, thereby preventing security issues such as double-spending attacks. This is achieved through consensus building. From a blockchain perspective, the consensus process directly provides assurance regarding the settlement of every data dimension within the blockchain among the involved entities

The primary objective of mining is to serve as the leading node in the verification of a block. This process involves pitting nodes with the highest computational power against each other. Successful miners are rewarded with a positive number of coins for their efforts. Over time, as blockchain technology has evolved, various new consensus algorithms have been introduced, including Proof of Stake (PoS). Innovative technology, such as smart contracts, has gained significance across multiple domains, including healthcare (Fetzer et al., 2021). Smart contracts are programmable utilities within the blockchain ecosystem. Nick Szabo coined the term “smart contract” to describe it as a “digital transaction protocol that executes the terms of a contract.” The core objectives of smart contract design include fulfilling common contractual requirements (such as payment terms, liens, confidentiality, or enforcement), reducing exceptions, both malicious and unintentional, and diminishing the need for trusted intermediaries. Each conditional script is associated with a distinct blockchain address, and an official transaction is initiated when it enters a smart contract. Because smart contracts operate independently of external states, they offer a high level of transparency to the blockchain community. The outcomes of smart contract execution by community nodes are transparently recorded on the blockchain.

Blockchain technology has the capability to interact with the Internet of Things (IoT) in the context of healthcare (Wang et al., 2018). IoT devices, like sensors and gateways, can collect real-time sensory data from patients and securely transmit it to a blockchain for sharing and archiving. Blockchain technology enables decentralized communication between IoT users and devices, eliminating the need for central authorities to facilitate communication among healthcare professionals, including doctors, nurses, and patients. In cases of erratic IoT activity, such as sensor data collection, community members (comprising blockchain nodes) can monitor and approve transactional events and manage community operations. This approach could update the public ledger accessible to all network users, even though the role of central authorities is currently diminished (Saleh et al., 2019). Blockchain has demonstrated promise in securing medical records, simplifying record management, and directly storing scientific data. It has also proven effective in biomedical applications, making it a viable solution to address healthcare challenges stemming from the COVID-19 pandemic. Modern COVID-19 data analysis, prediction, and drug/vaccine discovery have benefited significantly from the application of artificial intelligence (AI) techniques. Recent research primarily underscores the use of two core AI methods: machine learning (ML) and deep learning (DL). ML forms part of AI and focuses on understanding the structure of data and incorporating it into models that individuals can communicate and use (Rangone et al., 2021). The goal of ML is to allow computers to produce values within a given range based on statistical evaluation methods rather than relying on records as training inputs. ML can establish patterns from data to automate decision-making using inputs from facts alone. For instance, several early studies employed temperature-monitoring-based techniques, such as facial and body temperature recognition with smart caps, to identify individuals with abnormal temperatures and those who may be symptomatic. An American AI company utilizes intelligent graphs powered by ML to monitor disease activity across China, enabling the development of an alert system that informs users if an infected individual has visited their vicinity. This response simplifies the identification of infected individuals and provides them with access to medical resources. AI-driven solutions have been effectively applied in innovative COVID-19 containment methods.

In deep learning, increasing the number of layers in neural networks enhances their capacity to model numerous neurons in the brain’s network, which spans various layers. A fundamental deep neural network consists of three layers: an input layer for gathering data samples, a hidden layer for processing, and an output layer for forecasting outcomes. The depth of the hidden layers in this context signifies the depth of feature learning. Supervised or unsupervised learning techniques are used to produce the desired outcome, adjusting the weights between perceptron based on labeled or unlabeled statistical patterns. DL, a prominent field within AI, finds application in speech recognition, computer vision, image processing, and object detection for healthcare purposes, with COVID-19 infection prediction and the development of related drugs and vaccines as particular areas of focus. Several AI research teams have harnessed DL-powered solutions to predict COVID-19 infections and distribute the necessary treatments and vaccines. The public literature encompasses a wide array of AI-based techniques for COVID-19 detection, analysis, and prediction. Due to the pandemic, it has become more challenging to collect adequate data for the use of AI algorithms, and concerns about user privacy have grown due to data centers sharing public information for COVID-19 data analysis. In this context, federated learning (FL) has emerged as a promising AI strategy to develop cost-effective COVID-19 scientific applications with advanced security safeguards. Federated learning is a distributed AI approach that enables the training of powerful AI models by aggregating local updates from various hospitals and research facilities without necessitating direct access to local information. As it mitigates the risk of personal data exposure, the gamble of personal data leakage during this pandemic is likely reduced. The quality and accuracy of COVID-19 data instruction would also be notably enhanced, as FL trains AI models using a vast array of data and computational resources from numerous COVID-19 data sources. Centralized AI systems, which have limited data and computational power, cannot achieve this. To effectively control the COVID-19 epidemic, it is essential to monitor virus transmission and assess the likelihood of an outbreak. AI can leverage vast data sets on relevant subjects, including case occurrences, deaths, demographics, and environmental factors, to predict future COVID-19 occurrences and outbreak scale. Moreover, it facilitates the development of assessment techniques used to predict potential outbreaks, which is highly beneficial for governments seeking to formulate the most effective COVID-19 countermeasures. To establish the structure of COVID-19 and develop potential remedies to keep individuals healthy, it is imperative to model COVID-19 infections. The task of constructing a data-learning and evaluation mechanism to accurately predict the real COVID-19 infection pattern would be exceedingly challenging without the aid of machines. AI-driven algorithms on computers that mimic AI technology have the capability to assess how an individual’s constituent proteins will fold or deform based on their genome, drawing from data. This information is then utilized to ascertain the shape of receptors, thereby facilitating the development of effective COVID-19 drugs. Effective COVID-19 drug discovery may become more straightforward due to this development.

3. Relevant Work