An Approach to Identifying Pre-Existing Chronic Conditions for Corona Virus Positive Patients Using Regression

Introduction

COVID- 19 was first detected in Wuhan, China in the month of December 2019, (Salehi, Baglat, & Gupta, 2020) from where it spread to all across the world. WHO declared it as a pandemic in March 2020 There are many symptoms, viral origin, transmission, identification and diagnosis for potential treatment (Valencia, 2020). Until now, we have not been able to find a better and confirmed cure for the corona virus treatment. (Muhammad, Islam, Sharif, & Ayon, 2020) Although every country has taken lots of measures and educated their citizens to take of them and their family including lock down in various stages. Stay home, work from home, no outings have become new normal now along with use of mask to cover nose and mouth and sanitize/wash your hand regularly. This lockdown has changed the pattern of season, effected country as a nation in terms of economy crisis, effected individuals in terms of their expenses and income imbalance resulting into frustration, depressing life. However, its not only negative impact has positive impact also. People have spent quality time with their family, rejuvenated their hobbies etc. (Chatterjee, Sujath, & Hassanien, 2020). The symptoms data have been taken from Kaggle website to anticipate the symptoms responsible for positive results. Some of the fields taken into considerations can be pneumonia, asthma, hypertension, cardiovascular, tobacco, diabetes etc. Moreover, it is important to study the effect of such diseases on patients whether covid positive or not. This will also help us to understand the impact of various diseases on the patients with respect to covid.

Artificial Intelligence (AI) is considered to be of great help in predicting the impact of all such symptoms/diseases in the patient.in computer edge world today Machine Learning is a solution to almost all crucial problem where we need to work with real time situation. AI can help us to predict and analyse covid situation in a better way without much working on the field. Once ML/AI had been used to predict the result story telling or data analysis can help us better understand the data and the relation of every filed with each other. (Prakash, Imambi, Mohammed Ismail, Kumar, & Pavan, 2020) Machine Learning and Data Analysis helps us to understand the impact of corona virus on people-based age, gender, pre medical history etc. Based on these results we can predict the mortality rate/ risk (Pourhomayoun & Shakibu, 2020), recommend precautions, avoid coming in contact with infected person etc. it can help to predict epidemic spread situation using mathematical models. Along with Deep Learning, ML is used to exponential behaviour and predicting future reachability of COVID-19. Polynomial Regression, Root Mean Square Error are taken into consideration for evaluating losses due to the virus. (Singh, Sonbhadra, & Agarwal, 2020) (Salehi, Baglat, & Gupta, 2020)

Data Mining Techniques, Machine Learning and AI Techniques and various algorithms such as, Linear Regression, Logistic Regression, Support Vector Machine, Decision Tree, Random Forest, Naïve Bayes, KNN etc. are becoming helpful in predicting lots of possibilities to manage spread and diagnosis of corona virus. (Muhammad, Islam, Sharif, & Ayon, 2020) Machine Learning Model performance are evaluated using various metrics like: Kappa, Accuracy, F1 – Score, Precision, Recall, ROC, AOC, Principal Component Analysis for feature extraction, Chi Square test to check overall significance etc. (Gawriljuk, et al., 2020)

Here we have taken database named “covid.csv” from Kaggle that has various variables as id, sex, patient_type, entry_date, intubed, age, pregnancy, diabetes, tobacco etc. These variable/columns have values as 1- yes, 2- No and 97-99 as no data available. We have pre-processed the data and eliminated many variables not relevant for the research and the rows found irrelevant had been deleted/omitted.