ABSTRACT
The lack of modern technology in healthcare has led to the death of thousands of lives worldwide due to COVID19 since its outbreak. The Internet of Things (IoT) along with other technologies like Machine Learning can revolutionize the traditional healthcare system. Instead of reactive healthcare systems, IoT technology combined with machine learning and edge computing can deliver proactive and preventive healthcare services. In this study, a novel healthcare edge-assisted framework has been proposed to detect and prognosticate the COVID-19 suspects in the initial phases to stop the transmission of coronavirus infection. The proposed framework is based on edge computing to provide personalized healthcare facilities with minimal latency, short response time, and optimal energy consumption. In this paper, the COVID-19 primary novel dataset has been used for experimental purposes employing various classification-based machine learning models. The proposed models were validated using k cross-validation to ensure the consistency of models. Based on the experimental results, our proposed models have recorded good accuracies with highest of 97.767% by Support Vector Machine. According to the findings of experiments, the proposed conceptual model will aid in the early detection and prediction of COVID-19 suspects, as well as continuous monitoring of the patient in order to provide emergency care in case of medical volatile situation.
ABSTRACT
Purpose: The rapid spread of severe acute respiratory syndrome coronavirus-2 infection resulted in an exponential increase in hospitalizations and mortality. We aimed to explore the determinants of mortality and formulate a score that can predict mortality in patients hospitalized due to coronavirus disease 2019 (COVID-19). Materials and Methods: In this retrospective study, 1024 COVID-19 patients hospitalized between March 2020 and October 2020 were included. Patient demographics, underlying comorbid illnesses, clinical features, vital signs at admission, disease severity, and laboratory parameters, were collected from hospital medical records and analyzed to derive risk factors for in-hospital mortality and formulate a mortality prediction score. Results: The median age of the study population was 56 years (interquartile range [IQR], 45-65) and was significantly higher in nonsurvivors than in survivors (62 [IQR 55-70] vs. 52 [IQR 40-65];P = 0.001). Hypertension and diabetes were the most common associated comorbid illnesses seen in 50.5% (n = 518) and 29.1% (n = 299) of patients, respectively. The presence of altered level of consciousness (C), azotemia with serum creatinine >1.5 mg/dl (A), respiratory rate >25/min (R), interleukin-6 >25 pg/ml (I), and age >= 65 years were independent predictors of mortality. A six-point COVID-19 mortality prediction score, "CARI-65," was developed using variables predicting mortality in multivariate regression analysis. The CARI-65 score >= 3 had a sensitivity and specificity of 87.1% and 57.3%, respectively, and positive and negative predictive values of 42.52% and 92.45%, respectively, in predicting mortality. Conclusion: This study demonstrated various demographic, clinical, and laboratory parameters that predict mortality in hospitalized COVID-19 patients. We also proposed a simple risk stratification score to predict mortality in hospitalized COVID-19 patients, so that effective triaging of patients can be done to utilize health-care resources efficiently.