Deep Learning Applications and Perspectives: COVID 19

An anticipated outbreak of corona virus (COVID-19) brings out the severe acute respiratory syndrome and corona-virus2(SARS-COV-2) attacked in Wuhan city, china beginning of February 2020. World health organization (WHO) announced the outburst of pandemic disease through public health emergency of international concern. A variety of control measures has been taken by the government to control the disease. The investigation can be done about the pathogen and current epidemic. The world healthcare system has a major concern for new infectious diseases like covid-19 and needs new technological support. Deep learning in Artificial intelligence (AI) helps the world by safeguard the people from pandemic disease. Our aim is to investigate the AI based deep learning algorithm to analyze, prevent and prepare to defend against covid-19 and similar infectious disease.


Introduction
According to WHO, the Covid-19, December 1, 2020, the confirmed cases of covid-19 is 62,844,837 and death 1465144 with countries, areas, or territories with cases. The COVID-19 -19 pandemic was reported in Wuhan, China, and has spread over 180 countries including India. The WHO published the first disease outbreak news on 4 January, it mainly includes the risk assessment measures, states of patients in China and public health response in the community pneumonia cases in Wuhan. Later after much joint mission which includes health experts from Germany, Canada, Russia, Singapore. The reports are taken and preventive measures are suggested to health workers and the public. In March 2020, the alarming level of spread and severity are started to assess the severity of pandemic disease and can predict the next outbreak location WHO launched the solitary trial called as an International medical trial, that focuses to produce robust data, around the globe to perceive the most successful treatment for covid-19. Today, the global health crisis, healthcare is seeking advanced technology to analyze, monitor, control real-time, and control the outspread of dangerous disease Covid-19. Deep learning is an emerging technology which can track the escalation and growth of infectious virus and control the spread of disease in. By using deep learning technology, the patient mortality risk can be analyzed, the technology is also used to predict the next outbreak location.

Main Application of Deep Learning Towards Covid-19
Prevention of pandemic infections The quick analysis can be done by AI to find the irregular symptom which alerts the healthcare authorities and individuals [1]

Treatment tracking
IoT-based deep learning technology is an efficient system for autonomous monitoring and analyses of the escalating nature of the virus [2][3][4]. The CNN structure has been developed which helps to extract the disease feature and suggest the solution for the cases. The suggested CNN structure has a convolution layer that collects the input features and filters it. The neural network has a pooling and fully connected layer. The pooling layer is used to reduce the density for comp performance were fully connected is a neural network. combing the above two the CNN mode is created and the task can be adjusted by internal parameters.
Here object identification and classification are a major task.

Contact tracing
The contact tracing is a 6-step procedure 1.Basic information collection 2. Collect the infectious period 3.List their contacts 4.Isolate the infectious case with instructions 5.Contact tracing: Intimate the case closed contact about symptoms and ask them to in quarantine 6.Regular check-ins of the case and close contact can be done until quarantine ends. By collecting the above information. Deep learning methods help to identify the infection level of virus, their contact clusters and upcoming hotspots for tracing, monitoring and controlling the disease Forecast the cases and mortality AI-driven models can trace foresee the virus infections and identification of hotspots from the statistics given in social media, Risk of infection. Moreover, the positive cases of this disease and death can be predicted. The deep learning technology can help the people by forecasting the about data, this leads the social awareness about the pandemic disease Build out the vaccines and drugs Deep learning acts as a self-learning platform that contains historical data, pragmatic information, and pharmaceutical data. Depending upon the molecular structure of the drug, the drug design initiatives are taken.
The neural network-based predictive model is used to find the drug which contains the collection of the drug-like small molecules [5,6]. Approximately 1.6 million drug-like molecules are fed into the training dataset from the CHEML dataset which used the SMILES Simplified molecular Input line-entry system format which triggers the model to identify the needed features to make dry like tiny molecules (TCS)

Prediction of Next outbreak location
The next outbreak location of covid-19 can be predicted using data analytic techniques which consist of two methods statistical and learning methods. predictive analysis deals with pattern discovery and finding the upcoming happening where statistical analysis involves gathering and examine the data sample in a set of information from where the samples are taken linear regression, predictive modeling, Statistical analysis, and logistic model are the core capability of predictive analysis. The pillars of predictive analysis are Models and algorithms. The models are termed as time series, forecast, clustering, and classification Time series used for short term predictions which use the historical data to analyze where the input parameter is time Forecast models analyze the past numerical data points and predict future value. Performance metrics or new values are derived from the details. Clustering model organizes data into diverse groups which rely on general parameters Ex: COVID-19 symptoms cases are classified into levels Level 1 contains most common symptoms and Level 2 has the abrupt massive growth in the number of patients during the infections, the healthcare professionals and workers face extreme workload. The healthcare system has been attracted by technological innovation, this leads to an increase in therapeutic procedures and innovations in the medical field. This innovation assists the healthcare professionals in analyzing the disease's nature and providing treatment immediately this uses high technical innovations to assist them to feed more confidence while decreasing the physical workloads. Level 1: Serious symptoms MCS: Fever, tiredness, dry cough LCS: Sore throat, aches & pains diarrhea, headache SS: Difficulty in breathing, chest pain, loss of speech Load minimization of healthcare professionals The abrupt massive increase in the number of patients during the pandemic infections, the healthcare professionals and workers face extreme workload. The healthcare system has been attracted by technological innovation, this leads to an increase in therapy procedures and innovations in the medical field. This innovation assists the healthcare professionals in analyzing the disease's nature and providing treatment immediately this uses high technical innovations to assist them to feed more confidence while decreasing the physical workloads [7][8][9][10].

Deep Learning and Covid-19
This section focuses on the introduction of deep learning methods that support the existing techniques which deal with the COVID-19 in the healthcare system. The effectiveness of the strategies around the world analyzes and publishes the latest updates about COVID-19. The enhanced Artificial Neural network (ANN) based methods support to improvise medical treatment and disease diagnosis. The Comparison of traditional and modern AI-based methods is presented in Figure 1. The various steps are used to overcome covid-19.
The first step is data mining which includes data preparation, data understanding, and big data. The clinical data collection is also an important part which includes clinical reports, images, text records, and other information that is understood by the machine. The primary objective of data understanding includes data volume and data variables in which raw data is converted. Physician contribution is most important to observe the machine learning methods.

Classification model
In the classification model, data into different categories with historical data and future prediction Ex: Prediction of disease Whether the person has a disease or not. The prediction algorithms for pandemic disease in deep learning need structure and it can be linear or nonlinear fashion. To identify or recognize the entity they need analysis in deep learning. The algorithms are widely used in healthcare application for predicting the disease. For classification and regression, the Random forest algorithm can be used gradient boosted algorithm is used for classification. The diverse applications of ANN in identifying the disease and finding the symptoms in layered processes are shown in Figure 2. This process is exclusively designed for identifying COVID-19 problems using the image processing technique. The initial layer is an input layer designed for maintaining a database. In the front end, the main computer is used as a high-speed channel. The database is closely coupled to the central processing unit and loosely coupled to the network. The database sent an enormous amount of data to the mainframe system. The second layer is called a selection layer which is composed of an ANN-based selector that has the enhanced image processing techniques. The physician overlooks and confirms the decision made by this layer and suggest suitable images for the next layer. Diverse image processing techniques like Magnetic resonance imaging, microscopic imaging, x-ray imaging, and tomography are the image processing techniques are used. For pathological examination, the optical microscope is used as a dominant tool. The disease detection is done by PET scan and it tests for identifying the body functional mechanisms. The fourth layer is the optimization layer, used for image improvisation. To classify the covid-19 and viral pneumonia, the DL technology is used. The Layer5 is reserved for disease diagnosis using the ANN method. The conventional neural network (CNN) is used to achieve the goals. This network type is used for non-linear modeling and medical image for the disease diagnosis process [11][12][13].

Conclusion
The Machine learning technique in Artificial intelligence is a prediction tool in the medical field to detect the early disease infections due to the coronavirus and assist the physician to monitor the physical condition of the infected person. It identifies machines, develop an algorithm, or optimize the medical data with high accuracy. This AI-based system for covid-19 is yet to be achieved. The deep learning in AI assists in giving proper treatment, suggest prevention methodology and supports vaccine & drug development. the disease state using image processing techniques and makes a decision using an AI algorithm. These emerging platforms assist physicians to train.