Emerging technologies for the management of COVID19: A review

The outbreak of COVID19 has put a halt on life over the globe. For a while, everything was stopped except the spread of disease and mortality rate. This has become the greatest challenge of decade to deal with it. Globally, scientists and researchers were busy in finding a way to deal with this deadly pandemic. As this pandemic breaks out a huge demand for healthcare equipment, medicinal facilities has been rises and Industry 4.0 seems to be a hope during this pandemic which has potential to satisfy all these needs. In the battle, against this pandemic branches of computer science: Artificial Intelligence(AI), Internet of Things(IoT), Robotics, Machine Learning(ML) and Deep Learning(DL) played very important roles. Without the help of IoT and Robotics it would be impossible for frontline warriors to remain contactless with an infected person. Meanwhile, rapid testing, prediction of disease, sentiment analysis of population and many more would be only possible due to presence ML and DL algorithms. Undoubtedly, if this pandemichappened before the emergence of AI, IoT, ML, DL and Robotics; then the aftermath will surely be something else. This paper will highlight the contribution of these technologies in handling this pandemic from its treatment to management. This paper will give idea about the role of technologies, their affects, solutions provided by them, improvement needed in healthcare facilities, their role in managing sentiments of public during pandemic. The innovative part of this paper is that we are exploring each field of industry 4.0 and observing which plays the most important role.


Introduction
"COVID-19 " which stands for "coronavirus disease-2019 " is a respiratory problem happens due to "severe acute respiratory syndrome (SARS-CoV-2) " virus, a contagious virus of the coronaviridae family of single stranded, positive-sense RNA viruses. Like influenza, "SARS-CoV-2 " also damages the breathing system and causes symptoms such as coughing, temperature, tiredness and shortness of breath [1] . It is an infectious illness that grows readily from one to another in close contact, and scientists divide it into 5 classes: "Asymptomatic, mild, moderate, severe, and critical " [2] . The "World Health Organization (WHO) " labelled it as pandemic on 11 March 2020. This sickness infected over 100 million people as of January 28, 2021, resulting in over 2 million deaths [2] . There have been 6166,978 cases worldwide, with 372,037 deaths, and these are just the reported cases [3] . The exponential increase in the instances strains medicalameneties, and significant multimedia medical data is being analysed in search of a solution [4] . In this context, scientists, mathematicians, and all community of artificial intelligence trying hard to construct significant prediction system for COVID-19 instances in many nations [3] .
Mathematical modelling of disease transmission and propagation aids in the prediction of epidemic course and the design of mass vaccina-  In recent years, deep learning, the foundation of new AI, has been linked to drastically improved diagnostic accuracy in medical imaging for the automatic detection of pulmonary illnesses. Most basic labelling for covid-19 is patient level labelling whether positive (infected) or negative (non-infected) [13] . One of the most extensively utilised and effective approaches for identifying COVID-19 from digitised photos is the "convolutional neural network (CNN) " [14] . A fuzzy rule-based strategy based on a priority-based method is offered for providing hospital beds for "COVID-19 infected patients " in the worst situation, where the counts of beds available are much smaller than the number of pandemic infected patients [15] . Table 1 describes the applications of computer science in mitigating the pandemic.
Many people's typical working habits and routine tasks have been thrown off by pandemic, but technology innovation and intervention have resulted in best practises that can be expected long after pandemic is over. The most effective are AI, Machine Learning, Analytics, Cloud-Based Platforms, Gamification, and IoT as depicted in Fig. 1 [16] .

Review methodology
The review is categorized into various parts as shown below.
I Collection of literature: Papers, articles, thesis etc. are gathered from authentic and trustable resources like IEEExplore, Sciencedirect, Elsevier and Springer etc. II Literature Survey: Research carried out all over the world in the woke of COVID-19 to find out the mitigating role of AI, ML, DL, IoT and Robotics in the battle against this pandemic. III Identifying the role of, branches of computer sciences in the battle against this pandemic: After the literature survey, identification of the contribution of "Internet of Things (IoT) ", "Robotics ", "Artifi-cial Intelligence (AI) ", "Machine Learning (ML) " and "Deep Learning (DL) " in handling this pandemic situation. Identification of most used ML and DL algorithm. IV Discussion: Overall discussion of their role, significance, contribution and concerns. Discussion part consist of role identification played by AI, ML, DL, IoT and Robotics in the battle against this pandemic. V Further in this paper, Fig. 1 gives the idea how the combination of various technologies of Industry 4.0 contributing in this pandemic and Fig. 2 is structured view of their application and Fig. 2 represents the percentage of literature of various domain of Industry 4.0 surveyed. In this paper, Table 1 illustrates the major domain of computer science and categorize their remarkable applications, Table 2 depicts the name of some machine learning and deep learning algorithms which are widely used in handling this pandemic, Fig. 3 depicted four criteria's to evaluate the literature survey done Table 3 . Table 4 3. Related work

Internet of things
The "Internet of Medical Things (IoMT), " often known as the "healthcare IoT, " is a network of healthcare equipment and software applications that provide comprehensive healthcare services and are linked to healthcare information technology systems. IoMT applications include (1) remote patient monitoring, (2) medication order tracking, and (3) Wearables are being used to send health information to the relevant health care experts. Because of its ability to rapidly gather, analyse, and transmit health data, IoMT technologies have been recognised as having disruptive potential in the health care sector. Several researchers, Table 2 Illustrates some widely used ML and DL algorithms for "COVID-19 ".
A SMART TEMPERATURE SENSING DEVICES: These kind of thermometers are connected to the smartphone applications, which sends their readings to the company immediately. B IoT BUTTONS: These buttons were developed for quick deployment in any building, regardless of size, to send immediate notifications to management, alerting them to any sanitation or maintenance issues that could jeopardise public safety.
C TELEMEDICINE: Telemedicine is the practise of remotely monitoring patients utilising IoMT technologies. This practise, also called as "telehealth ", allows doctors to "evaluate ", "diagnose ", and "treat " patients without physically interacting with them. D SENSORS: The "Covid19 mobile defender " is a sensing device which catches virus safety and violation measures to assist authorities in managing the coronavirus's propagation and allocating scarce available means [17] . E Digital SCREENING TOOLS [18] .

Table 3
Summary of the evaluation criteria.

Criteria
Definition Function It is a supervised machine learning technique that classifies new observation into some classes on the basis of trained data.
This technique is basically used to distinghuish between covid infected and non-infected person. Prediction C 2 A machine learning technique that estimate or predict from observation. This is used to predict severity of covid, its peak in any geographical area, its outbreak and how much will affect. Management Techniques C 3 Applications of computer science used for the management of pandemic.
ML and DL used for sentiment analysis, drone cameras are used for keeping eye on lockdown, masque detection technique in crowd etc. Contactless Treatment C 4 As covid-19 is an infectious disease, we need technologies which can handle it without contact.
Robotics and various IoT devices plays very important role in tackling it without contact; like robots taking care of highly infected patient, unmanned vehicles used for sanitisation purpose.

Robotics
Robots are the finest option for this risky profession since they are innately resistant to virus infection. Autonomous robots, such as the floorcleaning bots that are already accessible as consumer products, can do more than just clean in hospitals [19] . Similar to "drone technology, " other autonomous technologies such as "robots and autonomous vehicles (AVs) " have made important advancements in the battle against the pandemic [20] .In some countries robots are used for taking care of patients as well as for cleaning and sanitising purpose, in case of severely infected patient [21]

Artificial intelligence
Amongst engineering technologies AI is one of the technology which can help in managing this deadly virus in many ways like identification of high risk patients, providing aid in infection management and examining the transmission of virus. It can also help in predicting mortality risk by examining patient's medical records. AI is assisting in many ways in the fight against this virus like; patient screening healthcare facilities and assistance, infection control suggestions etc. [22] . Artificial intelligence is a potential and significant method for diagnosing early coronavirus infections and monitoring infected patients' health. It is possible to dramatically increase treatment uniformity and decision making by designing effective algorithms. It is utilised in patient's treatment as well as in their monitoring. It can monitor the COVID-19 outbreak at multiple scales, including medical, molecular, and epidemiological applications. It is also helping in research work by analysing the accessible raw data.AI can help with the formulation of effective treatment regimens, preventative initiatives, and the development of drugs and vaccines [20] . Furthermore, AI has been widely applied in all major healthcare disciplines, either through automating processes or augmenting decision-making. "Artificial Intelligence " is a way of integrating human intelligence with machine for developing a system capable of decision making. The main function of AI would be providing necessary and exact analysis for tracking people (infected) or at risk of infection [23] .
AI and ML are increasingly being employed as a tool to increase patient diagnosis rate, crowd management and surveillance, prevention of infection [24] . Several imaging-based COVID-19 diagnosis techniques supported by AI and machine learning have been presented in the last year, along with their correlation with RT-PCR. AI techniques are used to process CT and CXR images in order to detect pneumonia-like imaging features. COVID-19 is diagnosed using "CXR images " and "deep convolutional neural networks (CNNs). " Recently, researchers sought to deliver AI-solutions based on "deep learning techniques " to distinguish between covid infected patients from healthy and other pneumonic patients [24] . There are already various AI based system for several diseases but higher growth velocity is giving tough challenges [25] . As a result, researchers have been attempting to harness the power of ML or DL in order to assist medical personnel in accurately detecting this disease [26] . ML and AI techniques has been utilised for better understanding of patient's category which helps in clinical decision making Fig. 3. Emerging technologies for COVID 19. [27] . Computed tomographic scans are one of the vitally used diagnosis tool in the battle of this pandemic and a huge applaud to AI techniques for providing rapid decision making system [28] . The combination of AI and open-source data sets results in a practical COVID-19 diagnosis solution that can be implemented in hospitals worldwide [29] . The rising number of COVID-19 hospitalizations resulted in the creation of a large medical and population structural database, which is currently accessible. These datasets enables a very significant computational methods in finding the needle in the haystack which will aid in medical decisionmaking for 'COVID-19' identification and prediction. Hence, data mining approaches can be carried out using a "supervised machine learning algorithm " to predict future values through classification and/or regression, or by unsupervised learning to cluster data [19] . With the help of AI assisted healthcare devices, only contactless treatment is possible. Furthermore, one of the primary advantages of an AI-based system is remote location self-treatment [30] . Inspite of having advance technology, drug invention and production is a big challenge, with more failure rate and less efficiency. As a result, computer-aided drug design is increasingly utilised to identify medications, reducing costs and chance of failure. CADD refers to computer-aided design tools for keeping, maintaining, analysing, and modelling molecular compounds. As a result, it contains tools for designing compounds, evaluating potential lead candidates, and researching compounds' chemical interactions and physicochemical properties [ 27 , 31 ]. Below are the domains in which AI is assisting in dealing with this deadly virus:

Machine learning and deep learning
Because ML and DL have capability of recognising and predicting patterns in large, complex datasets, they have been highlighted as a viable strategy for developing COVID-19 diagnosis solutions. In comparison to other issue domains, the number of tests on COVID-19 using ML has increased rapidly in just two months [32] . Machine and deep learning (DL) techniques to automatic image analysis have recently shown promise for tissue reconstruction, classification, regression, and segmentation utilising ultrasound data [26] . When analysing vast amounts of data for disease diagnosis, ML and DL approaches have shown impressive performance. Several methods in illness diagnostics that leverage ML and DL approaches rather than standard computer-aided systems have been developed. When there is a huge medical dataset, deep learning models are often used. To construct a prediction and detection model, it is important to automatically extract characteristics from images. The comprehension of DL approaches has been considerably diminished [25] .

Machine learning
ML, is a subgroup of artificial Intelligence, previously proved its efficacy in the drug discovery procedures during prior health catastrophes [20] . In general, ML is used to improve the structure of data that humans deal with. Input data is trained, and output data is statistically analysed using machine learning methods. The detection of infected people as well as the monitoring of a person's temperature are examples of ML applications [33] . A number of studies have indicated that employing convolutional neural networks or other deep learning algorithms on CXR images for COVID-19 identification produced satisfactory diagnosis accuracy [34] . "Machine learning algorithms, " in instances, are used to link a patient's data parameters to the administration of a certain drug using AI technology. This type of correlation can be used to predict how a medicine will affect a certain group of patients. Doctors and medical suppliers can be better prepared for the consequences if they are aware of these factors ahead of time. Various "supervised machine learning techniques " use labelled data and features to build an automatic detection model [20] . From a set of characteristics that included Gross Domestic Product, sex, socio-political group, medical facilities, homeless, type of lockdown, population density, airport activity, and age groups, supervised machine learning (ML) algorithms were utilised to determine the major determinants driving COVID-19 infection and death numbers [35] . "Population density, testing numbers, and airport traffic " are the most prominent characteristics, followed by older age groups (over 40, notably 60 + ) [36] . In his studies, Dr. Jayavrinda Vrindavanam et al. use cough audio samples to differentiate between normal cough and covid-19 cough. The samples are then classified using a machine learning algorithm. This method is used to avoid making contact with the patient [19] . However, this procedure does not yield accurate results. Using proper "stacking algorithms, " differentiation between covid-19 patients and normal patients are possible not only this but we can also determine that a patient needed to be admit in ICU or in general ward [37] . The machine learning time series models are built using what we've learned from the spatial distribution of infections over time [38] .

Deep learning
Deep learning (DL) is a subset of AI, influenced by the human brain's structure [39] . DL has been a popular method for creating networks that can mimic higher-order systems and perform like humans [40] . DL is a multilayer artificial neural network that was created to increase the performance of neural networks. The greater the number of layers, the greater the accuracy [33] . It enables artificial intelligence to be trained to predict outcomes based on a set of data. To train artificial intelligence, both supervised and unsupervised learning can be used. The term deep learning refers to artificial neural networks. The human brain inspired artificial neural networks. It is made up of neurons, just like the human brain. The difference between them is the amount and speed with which they learn. To put it another way, in order to train, artificial neural networks require both a data source and computing power. The proper features are chosen to determine the quality of machine learning algorithms. Pre-processing, size reduction, feature selection, and other transactions are performed [41] .
Deep learning, also known as Convolutional Neural Networks, has received a lot of attention and praise for its effectiveness (CNN). CNN is a novel type of "neural network " which mimics the basic structure of the brain, including "neurons " and their connections across "intermediate layers, learnable weights associated with each link, activation function, and bias ". A "neuron " gets information from previous links, multiplies it by its "weights ", applies an "activation function " to the "weighted values, " and reacts with a "new value, " which is then passed on to the next layer of "neurons ". Neurons for input layer are similar to the pixels for CT scanner. Numerous intermediate layers of neurons, known as "feature maps, " exist behind the input layer of neurons. Because they use filters to convolve the outputs of the input layers, feature maps are also known as convolutional layers. Convolution is a feature extraction approach that passes through multiple layers of feature maps, filtering out the uninteresting and leaving the relevant features in each layer. The "feature maps " are collectively referred to as "kernels, " and the designer determines their size at random. The selected output has become concise after a feature map has been 'convoluted and sub-sampled' [19] . Each neural network design learns specific patterns from other neural networks since deep CNNs are stochastic. The ensemble technique increases feature extraction as well as improving accuracy [42] . For medical diagnosis, massive DL models incorporating "Convolutional Neural Networks (CNN) " were applied. Deep Methods such as "Stacked Auto-Encoder (SAE) ", "Deep Belief Network (DBN), " and "Deep Boltzmann Machine (DBM) "along with vector inputs are responsible for this [43] . The COVID19 disease damages the lungs of humans, which can be seen on a lung X-ray [44] . To forecast the Pneumonia case from chest X-rays, an effective CNN strategy was applied employing a convolutional neural network (CNN) [45] .

Transfer learning
Transfer learning is a deep learning technique that employs a deep convolutional neural network that has been trained to perform one task to perform another. The original model's parameters are fine-tuned for the second task [46] . Transfer learning (TL) has simplified the process of rapidly retraining neural networks on selected datasets with high accuracy [39] . DTLs such as "VGG, ResNet, and DenseNets " are now becoming an essential process in image/video detection and diagnosis for the time being. DL is used in the diagnosis of medical x-rays and computed tomography. DL improves the medical diagnosis system (MDS) to achieve excellent outcomes, and implements a relevant real-time medical diagnosis system [47] . Ensemble learning incorporates several transfer learning models, including "EfficientNet, GoogLeNet, and Xception-Net. " Some models can classify patients as having "COVID-19 ( + ), pneumonia ( + ), tuberculosis ( + ), or being healthy " [48] . Because it does not require a huge annotated dataset for training, the transfer learning method is quicker and easier to deploy [41] .

Sentiment analysis
The lockdown resulted in the closure of many businesses, economic downturns, and suicides amongst the common civilians. Additionally, persons who rely on multimedia devices to pass the time during a lockdown may experience severe psychological repercussions such as loneliness and sadness. As a result, there is a need to examine the psychology of the human mind in such a situation [49] . The public sentiment gleaned from numerous reflexions (hashtags, comments, tweets, and Twitter postings) is accurately measured, ensuring that various policy decisions and communications are taken into account [50] . The application exhibits premonition in the improvement of terror sentiment finally as panemicreaches its peak globally by utilising extensive textual analysis with the assistance of essential text data visualisation [51] . Text mining is defined as "the process of extracting useful information from unstructured textual data by identifying and exploring interesting patterns." Text mining is not only more useful than data mining, but also significantly more sophisticated, as it uses software that combines components of database systems, artificial intelligence, machine learning, and quantitative statistics to filter huge amounts of unstructured data. Following the filtering of this data, interesting and important patterns emerge, which can be examined and used [52] .
Because a great deal of misinformation about pandemic is disseminated to the public via various technological platforms, it is necessary to identify misinformation and mis-informants and then provide accurate information [5] . Based on text features, the problem of "misinformation detection " is classed as a supervised learning problem. Shallow or deep learning-based methods for detecting deceptive posts can be used, with at least two training and testing steps. These methods aim to build a binary classification that uses a variety of characteristics and auxiliary qualities during the training process to evaluate whether a post is deceptive or not [53] . To categorise social media information, several natural language processing approaches were applied. We use supervised learning methods like "support vector machines (SVM), naive Bayes (NB), and random forest (RF) " to learn the types of unlabelled data based on labelled data [54] . Sentiment analysis can be performed using hate speeches, polarity and ranking coefficients, as well as spammy and nonspammy nodes [55] .
The following are the most notable DL and ML algorithms used for COVID-19 detection and prediction:

Mobile applications
Based on responses from the mobile app, AI is utilised to substitute human expert judgement in calculating risk level. The software scans the user's information for suspected coronavirus infection, like some indications are (feeling feverish, headache, dry cough, breathing problem, and exhaustion), length and extremity of symptoms, travelling data, employment and home details, and population stats [19] . The "Covid19 mobile defender " is an example of a sensing device which keeps watch on virus defensive measures and violation of rules done by people during lockdown to help authorities in managing the coronavirus' propagation and allocating scarce resources [45] . Mobile applications, on the other hand, are always a threat to the user's privacy. Malicious actors are quick to exploit the pandemic to launch cyber-attacks [70] .

Discussion
Early prediction is crucial in the case of an epidemic in order to control the outbreak [79] . Based on the projection, government agencies and public health organisations can plan accordingly [80] . If an infected individual is not promptly detected and treated, he or she may become a carrier of the virus, unwittingly transmitting it to healthy persons. As a result, early detection of infected individuals would aid in quarantining them, which would aid in limiting the spread of pandemic [73] . Therefore, it would be very necessary to have some method to detect it with higher pace and accuracy. During this period of pandemic, Internet of Things (IoT), Robotics, Artificial Intelligence, Machine Learning and Deep Learning plays a vital role. Especially subset of AI; Machine Learning (ML) and Deep Learning (DL) proved to be a boon in handling this pandemic situation. Whether, they are used for classification of COVID-19 infected person, prediction of outbreak, prediction of its peak, severity of this disease, prediction of the intensive care unit (ICU) requirement. These algorithms are also used for sentiment analysis. Gathering data from social media platforms like Twitter or Facebook, and classifying them as positive, negative and neutral responses. It also helps in busting fake news and rumours. Through image classification, masque detection is also applied (recognising whether a person has wearing masque or not). All researchers use these algorithms for classification and prediction purpose. Most of them follows the approach of neural network concept, like: Convolution Neural Network (CNN), Deep Neural Network (DNN)or Transfer Learning. Internet of Things (IoT) are used in handling situation through using devices like; smart wearables for tracing isolated patient, keeping record of patient (where they go and to whom they meet), oximeter for checking oxygen level of patient and many more. In the same way Robotics technology are used for contactless treatment of patient, drones are used to keep watch on lockdown imposition, robots are also used for sanitisation and cleaning purpose, at some places unmanned vehicles are used for sanitisation. With the combination of these fields various mobile application are developed to handle pandemic. These apps have functionalities like vaccination slot booking, isolated patient tracking and many more. However, they always not happened to be a boon many a time mobile apps but the privacy of user at stake and not trustable by all users. After relief in pandemic many people experienced data breaches through many apps which they have downloaded for pandemic management purpose.
In countries with huge populations like India; rapid testing, proper imposition of lockdown, proper isolation could be only possible with the help of these (Internet of Things (IoT), Robotics, Artificial Intelligence, Machine Learning and Deep Learning). It is also observed that people with healthy diet have survived this virus, more than people with improper diet [67] .
Indeed, three well-known AI flaws can lead to failure: a lack of strong AI, the inability to execute without domain knowledge, and the requirement for adequate features and propagation. In addition, ML and DL methodologies should be scrutinised for identifying not only most recent elucidation but also for future improvements and research op-portunities, while neglecting moral considerations (like "trust and privacy ") which now limit Artificial Intelligence implementation for the humankind [81] . COVID-19 testing labs and kits are still in short supply around the world [82] .

Conclusion
After the literature survey, we come to the result that, emergence of computer engineering into the battle against this pandemic proved to be a master-stroke. Whether it would be about treatment, advancement in rapid testing or management, it plays a vital role in all fields. Especially, a tribute to Machine learning and Deep learning that makes many predictions and classification possible. However; we can't ignore its limitations and threats. Like, through machine learning and deep learning only image tomography is possible. There are lot of things remains unsolved. Another major issue is the privacy concern. Installation of mobile apps for tracking infected patients put the privacy at stake and didn't warm welcomed by everyone. Since, the outbreak of pandemic research work over its prediction, classification, peak and severity using machine learning and deep learning algorithms, are still going on. Rapid testing, manufacturing PPE kits and vaccines in bulks becomes easier using AI backed techniques. Various types of sensors and wearables invented for contactless treatment. Still research is going on for detecting disease with 100% accuracy. Still research is going on vaccines for higher efficiency. However, all these technologies united are not able to defeat this pandemic properly. As this deadly virus is changing its variant rapidly and mutating itself more and more exponentially; giving a tough time to these all technologies. Here, we can see one more thing that after the disastrous wave of covid-19 world became more dependent on technology and people became more techno-friendly. As, the situation of pandemic compelled world to go on online mode. People, now keen to do their work from home. The online delivery at their doorstep becomes more popular after this pandemic. People became familiar with new devices like rapid testing kit, automatic temperature detecting device, oximeteter etc. day by day. So, conclusively we can say that this pandemic made people more dependent on technologies. However, we can't deny this fact that these technologies created a major class difference. Privileged class and those who can afford these technologies; enjoys life even during pandemic and marganilised class suffers a lot. People suffered from malnutrition and those who have unhealthy food habits died more than people who do healthy diets. However, when this pandemic will be completely over then only real and most accurate aftermath can be calculated. When this deadly virus stop mutating more researches can be done over that which mutant will be most effectively handled or whether they all are handled similarly.