Identification of Symptoms Prognostic of COVID-19 Severity: Multivariate Data Analysis of a Case Series in Henan Province

113

106

Background Patients with COVID-19 in the intensive care unit (ICU) have a high mortality rate, and methods to assess patients’ prognosis early and administer precise treatment are of great significance. Objective The aim of this study was to use machine learning to construct a model for the analysis of risk factors and prediction of mortality among ICU patients with COVID-19. Methods In this study, 123 patients with COVID-19 in the ICU of Vulcan Hill Hospital were retrospectively selected from the database, and the data were randomly divided into a training data set (n=98) and test data set (n=25) with a 4:1 ratio. Significance tests, correlation analysis, and factor analysis were used to screen 100 potential risk factors individually. Conventional logistic regression methods and four machine learning algorithms were used to construct the risk prediction model for the prognosis of patients with COVID-19 in the ICU. The performance of these machine learning models was measured by the area under the receiver operating characteristic curve (AUC). Interpretation and evaluation of the risk prediction model were performed using calibration curves, SHapley Additive exPlanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME), etc, to ensure its stability and reliability. The outcome was based on the ICU deaths recorded from the database. Results Layer-by-layer screening of 100 potential risk factors finally revealed 8 important risk factors that were included in the risk prediction model: lymphocyte percentage, prothrombin time, lactate dehydrogenase, total bilirubin, eosinophil percentage, creatinine, neutrophil percentage, and albumin level. Finally, an eXtreme Gradient Boosting (XGBoost) model established with the 8 important risk factors showed the best recognition ability in the training set of 5-fold cross validation (AUC=0.86) and the verification queue (AUC=0.92). The calibration curve showed that the risk predicted by the model was in good agreement with the actual risk. In addition, using the SHAP and LIME algorithms, feature interpretation and sample prediction interpretation algorithms of the XGBoost black box model were implemented. Additionally, the model was translated into a web-based risk calculator that is freely available for public usage. Conclusions The 8-factor XGBoost model predicts risk of death in ICU patients with COVID-19 well; it initially demonstrates stability and can be used effectively to predict COVID-19 prognosis in ICU patients.

Section: Introductionmentioning

confidence: 99%

Prognostic Assessment of COVID-19 in the Intensive Care Unit by Machine Learning Methods: Model Development and Validation

Pan¹,

Li²,

Xiao³

et al. 2020

113

106

“…COVID-19 is predominately characterized by systemic symptoms, such as fever and fatigue, and respiratory symptoms, such as cough, expectoration, and a runny nose [ 3 , 4 ]. In a recent analysis examining the relationship between symptoms of COVID-19 and disease severity, fatigue and expectoration were found to be the most critical, positive prognostic symptoms of COVID-19 severity, whereas a runny nose and nausea were favorable prognostic factors [ 4 ]. Transmission of SARS-CoV-2 occurs primarily via direct contact with the respiratory droplets of infected individuals, or via indirect contact with virally contaminated objects [ 5 ].…”

Section: Introductionmentioning

confidence: 99%

Perceptions, Knowledge, and Behaviors Related to COVID-19 Among Social Media Users: Cross-Sectional Study

Ali¹,

Whitebridge²,

Jamal³

et al. 2020

Background Social media is one of the most rapid and impactful ways of obtaining and delivering information in the modern era. Objective The aim of this study was to rapidly obtain information on public perceptions, knowledge, and behaviors related to COVID-19 in order to identify deficiencies in key areas of public education. Methods Using a cross-sectional study design, a survey web link was posted on the social media and messaging platforms Instagram, Twitter, and WhatsApp by the study investigators. Participants, aged ≥18 years, filled out the survey on a voluntary basis. The main outcomes measured were knowledge of COVID-19 symptoms, protective measures against COVID-19, and source(s) of information about COVID-19. Subgroup analyses were conducted to determine the effects of age, gender, underlying illness, and working or studying in the health care industry on the perceived likelihood of acquiring COVID-19 and getting vaccinated. Results A total of 5677 subjects completed the survey over the course of 1 week. “Fever or chills” (n=4973, 87.6%) and “shortness of breath” (n=4695, 82.7%) were identified as the main symptoms of COVID-19. Washing and sanitizing hands (n=4990, 87.9%) and avoiding public places and crowds (n=4865, 85.7%) were identified as the protective measures most frequently used against COVID-19. Social media was the most utilized source for information on the disease (n=4740, 83.5%), followed by the World Health Organization (n=2844, 50.1%). Subgroup analysis revealed that younger subjects (<35 years), males, and those working or studying in health care reported a higher perceived likelihood of acquiring COVID-19, whereas older subjects, females, and those working or studying in non–health care areas reported a lower perceived likelihood of acquiring COVID-19. Similar trends were observed for vaccination against COVID-19, with older subjects, females, and those working or studying in non–health care sectors reporting a lower likelihood of vaccinating against COVID-19. Conclusions Our results are indicative of a relatively well-informed cohort implementing appropriate protective measures. However, key knowledge deficiencies exist with regards to vaccination against COVID-19, which future efforts should aim at correcting.

“…Despite the explosive growth of the number of studies on COVID-19 [ 2 - 4 ], several key epidemiological parameters of the disease remain to be clarified, among which are incubation period and serial interval. The mean or median incubation period and serial interval estimated by previous studies were mostly 4-5 days [ 5 - 16 ].…”

Section: Introductionmentioning

confidence: 99%

Epidemiological Parameters of COVID-19: Case Series Study

Ma¹,

Zhang²,

Zeng³

et al. 2020

Background The estimates of several key epidemiological parameters of the COVID-19 pandemic are often based on small sample sizes or are inaccurate for various reasons. Objective The aim of this study is to obtain more robust estimates of the incubation period, serial interval, frequency of presymptomatic transmission, and basic reproduction number (R0) of COVID-19 based on a large case series. Methods We systematically retrieved and screened 20,658 reports of laboratory-confirmed COVID-19 cases released by the health authorities of China, Japan, and Singapore. In addition, 9942 publications were retrieved from PubMed and China National Knowledge Infrastructure (CNKI) through April 8, 2020. To be eligible, a report had to contain individual data that allowed for accurate estimation of at least one parameter. Widely used models such as gamma distributions were fitted to the data sets and the results with the best-fitting values were presented. Results In total, 1591 cases were included for the final analysis. The mean incubation period (n=687) and mean serial interval (n=1015 pairs) were estimated to be 7.04 (SD 4.27) days and 6.49 (SD 4.90) days, respectively. In 40 cases (5.82%), the incubation period was longer than 14 days. In 32 infector-infectee pairs (3.15%), infectees’ symptom onsets occurred before those of infectors. Presymptomatic transmission occurred in 129 of 296 infector-infectee pairs (43.58%). R0 was estimated to be 1.85 (95% CI 1.37-2.60). Conclusions This study provides robust estimates of several epidemiological parameters of COVID-19. The findings support the current practice of 14-day quarantine of persons with potential exposure, but also suggest the need for additional measures. Presymptomatic transmission together with the asymptomatic transmission reported by previous studies highlight the importance of adequate testing, strict quarantine, and social distancing.