Article Preview
Top1. Introduction
The sudden outbreak of COVID-19 worldwide has caused a lot of deaths following crashing economies and the standstill of everyday life. The virus severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is thought to spread from person to person who comes in contact with an infected person (mostly within 6 feet). With hundreds of recent pieces of evidence, it could also be transferred among different species (Kar & Leszczynski, 2020). It spreads through respiratory droplets produced by sneezing, coughing of an infected person. Unfortunately, symptoms are not apparent immediately after the virus enters the human body in most cases (Ojha, Kar, Krishna, Roy, & Leszczynski, 2020). Symptoms of COVID-19 as suggested by CDC (Center for Disease Control and Prevention) include fever or chills, cough, shortness of breath or difficulty breathing, fatigue, muscle or body aches, headache, the recent loss of taste or smell, sore throat, congestion or runny nose, nausea or vomiting, diarrhea which are usually hard to differentiate from that of common viruses and seasonal flu. As of March 23, 2021, deaths that occurred due to the virus are more than 2.7 million, and over 124 million people have got affected around the world. More than 29 million cases are registered in the USA, and close to 0.54 million people died as per CDC records (“CDC COVID Data Tracker,” n.d.). According to the FDA, three different types of tests are conducted, namely molecular test, antigen test, and antibody test (“Coronavirus Testing Basics | FDA,” n.d.). Molecular and Antigen tests confirm the active coronavirus infection, whereas the antibody tests such as serological tests suggest past coronavirus infection (“Coronavirus Testing Basics | FDA,” n.d.).
Among these suggested tests for the diagnosis of COVID-19, reverse transcription Polymerase Chain Reaction (rt-PCR) of molecular tests-based methodologies are shown to have high accuracy and sensitivity and do not need to be repeated (Banerjee et al., 2020). But, due to rapidly increasing cases day by day, it is hard to conduct such a wide range of tests with the available resources. To overcome this problem, we propose to use, at a preliminary level, machine learning (ML) models from blood sample test values that are both predictive and accurate in detecting the coronavirus infection. The ML methods are a form of artificial intelligence (A.I.) methods and are concerned with developing algorithms and techniques that allow computers to learn, train, and are used in various cases to predict the results. There is a wide range of uses of ML methods in clinical conclusions, in the field of bioinformatics and cheminformatics, used in securities exchange investigation, computer vision, video games, etc. In recent years, researchers and scientists found reasonable solutions for various diseases through ML models with tremendous accuracies.
This research considered essential features that could offer considerable insights into the data analysis aspects for testing and diagnosing COVID infections. It could drastically reduce the amount of time needed for data analysis. We have proposed here applying multiple ML tools to filter large amounts of blood test data to diagnose coronavirus infection in the upcoming days. Based on blood test parameters modeling of available data, we can predict future COVID-19 infection using our proposed method, which might save millions of dollars without the necessity of performing molecular level tests and antigen tests. Researchers have previously used ML techniques like artificial neural networks, random forest, and generalized linear model net for data analysis on whole blood counts with Eosinophils, Leukocytes, Rbc, MPV, Basophils Platelets, RBC. D.W., Lymphocytes, Hemoglobin, Hematocrit, MCHC, MCH, MCV, and Monocytes with 598 records from the data they collected from The Hospital Israelita Albert Einstein in the state of Sao Paulo, Brazil (Banerjee et al., 2020).