Forecasting of Novel Corona Virus Disease (Covid-19) Using LSTM and XG Boosting Algorithms Risk factors for COVID-19 and rheumatic disease flare in a US cohort of Latino patients

ABSTRACT

Summary The viruses are called as enteric viruses developed using ingestion termed as fecal oral transmission and is relicated using intestinal tract. Enteric viruses are genus Enterovirus phrased as Caliciviridae, Picoornaviridae, Coronaviride, Astroviridae, Orthoreovirus, genera Rotavirus, and Reoviridae phrase as Adenovirida and Reoviridae. Coronaviruses belong to the Coronaviridae family. It belongs to one of the Ribo-Nucleic Acid (RNA) families of the order Nidovirales, the others are pathegens of birds and insects of Arteriviridae and the Roniviridae families. The coronaviruses consists of single stranded RNA genome of 30kb in length in size. An epidemic of novel corona virus called as SARS-CoV-2 irritates the COVID- 19 disease is reported recently. It is enveloped, plus stranded RNA viruses with extra ordinarily large genomes and helical nucleocapsids. During the pandemic situations, it is necessary to predict Covid cases in advance to take the preventive measures and thus saving the human life and other living beings. To predict the count of Covid-19 in advance and to improve the accuracy, this chapter proposes Machine and Deep learning algorithms such as Long Short Term Memory (LSTM), eXtreme Gradient Boost (XG Boost) algorithms and polynomial regression for forecasting. The real time dataset is taken from Kaggle which contains around 36,000 samples. The sample is taken from around 187 countries from the world and the dataset contains the details which included from the month of January to May, 2020. The algorithm is tested using test dataset and the performance is evaluated through the performance metrics. Abstract Objectives Latino patients are overrepresented among cases of coronavirus disease 2019 (COVID-19) and are at an increased risk for severe disease. Prevalence of COVID-19 in Latinos with rheumatic diseases are poorly reported. The purpose of this study was to characterize COVID-19 clinical features and outcomes in Latino patients with rheumatic diseases. Methods This is a retrospective study of Latino patients with rheumatic diseases from an existing observational cohort in the Washington, DC area. Patients seen between April 1 to October 15, 2020 were analyzed in this study. We reviewed demographics, body mass index (BMI), comorbidities, and immunomodulatory therapies. An exploratory Classification and Regression Tree (CART) analysis along with logistic regression (LR) analyses were performed to identify risk factors for COVID-19 and rheumatic disease flare. Results Out of 178 patients, 32 (18%) were identified with COVID-19 and the incidence rate of infection was found to be three-fold higher than the general Latino population. No patients required ICU level care. CART analysis and multivariable LR analysis identified BMI>30.35 as a risk factor for COVID-19 [P=0.004, OR=3.37, 95%CI (1.5-7.7)]. COVID-19 positivity was a risk factor for rheumatic disease flare [P=0.02, OR=4.57, 95%CI (1.2-17.4)]. Conclusion Latino patients with rheumatic diseases had a higher rate of COVID-19 compared with the general Latino population. Obesity was identified as a risk factor for COVID-19 and COVID-19 itself was found to be a risk factor for rheumatic disease flare. Latino patients with risk factors should be followed closely, especially post-COVID-19 in anticipation of disease flare.