Performance, Trust, or both? COVID-19 Diagnosis and Prognosis using Deep Ensemble Transfer Learning on X-ray Imagesg±

The COVID-19 pandemic still affects most parts of the world today. Despite a lot of research on diagnosis, prognosis, and treatment, a big challenge today is the limited number of expert radiologists who provide diagnosis and prognosis on X-Ray images. Thus, to make the diagnosis of COVID-19 accessible and quicker, several researchers have proposed deep-learning-based Artificial Intelligence (AI) models. While most of these proposed machine and deep learning models work in theory, they may not find acceptance among the medical community for clinical use due to weak statistical validation. For this article, radiologists' views were considered to understand the correlation between the theoretical findings and real-life observations. The article explores Convolutional Neural Network (CNN) classification models to build a four-class viz. "COVID-19", "Lung Opacity", "Pneumonia", and "Normal"classifiers, which also provide the uncertainty measure associated with each class. The authors also employ various pre-processing techniques to enhance the X-Ray images for specific features. To address the issues of over-fitting while training, as well as to address the class imbalance problem in our dataset, we use Monte Carlo dropout and Focal Loss respectively. Finally, we provide a comparative analysis of the following classification models - ResNet-18, VGG-19, ResNet-152, MobileNet-V2, Inception-V3, and EfficientNet-V2, where we match the state-of-the-art results on the Open Benchmark Chest X-ray datasets, with a sensitivity of 0.9954, specificity of 0.9886, the precision of 0.9880, F1-score of 0.9851, accuracy of 0.9816, and receiver operating characteristic (ROC) of the area under the curve (AUC) of 0.9781 (ROC-AUC score). © 2022 ACM.

A New Coronavirus Estimation Global Score for Predicting Mortality During Hospitalization in Patients with COVID-19.

Objective: Coronavirus disease 2019 (COVID-19) exists as a pandemic. Mortality during hospitalization is multifactorial, and there is urgent need for a risk stratification model to predict in-hospital death among COVID-19 patients. Here we aimed to construct a risk score system for early identification of COVID-19 patients at high probability of dying during in-hospital treatment. Methods: In this retrospective analysis, a total of 821 confirmed COVID-19 patients from 3 centers were assigned to developmental (n = 411, between January 14, 2020 and February 11, 2020) and validation (n = 410, between February 14, 2020 and March 13, 2020) groups. Based on demographic, symptomatic, and laboratory variables, a new Coronavirus estimation global (CORE-G) score for prediction of in-hospital death was established from the developmental group, and its performance was then evaluated in the validation group. Results: The CORE-G score consisted of 18 variables (5 demographics, 2 symptoms, and 11 laboratory measurements) with a sum of 69.5 points. Goodness-of-fit tests indicated that the model performed well in the developmental group (H = 3.210, P = 0.880), and it was well validated in the validation group (H = 6.948, P = 0.542). The areas under the receiver operating characteristic curves were 0.955 in the developmental group (sensitivity, 94.1%; specificity, 83.4%) and 0.937 in the validation group (sensitivity, 87.2%; specificity, 84.2%). The mortality rate was not significantly different between the developmental (n = 85,20.7%) and validation (n = 94, 22.9%, P = 0.608) groups. Conclusions: The CORE-G score provides an estimate of the risk of in-hospital death. This is the first step toward the clinical use of the CORE-G score for predicting outcome in COVID-19 patients.

Identification of the susceptibility genes for COVID-19 in lung adenocarcinoma with global data and biological computation methods.

INTRODUCTION: The risk of infection with COVID-19 is high in lung adenocarcinoma (LUAD) patients, and there is a dearth of studies on the molecular mechanism underlying the high susceptibility of LUAD patients to COVID-19 from the perspective of the global differential expression landscape. OBJECTIVES: To fill the research void on the molecular mechanism underlying the high susceptibility of LUAD patients to COVID-19 from the perspective of the global differential expression landscape. METHODS: Herein, we identified genes, specifically the differentially expressed genes (DEGs), correlated with the susceptibility of LUAD patients to COVID-19. These were obtained by calculating standard mean deviation (SMD) values for 49 SARS-CoV-2-infected LUAD samples and 24 non-affected LUAD samples, as well as 3931 LUAD samples and 3027 non-cancer lung samples from 40 pooled RNA-seq and microarray datasets. Hub susceptibility genes significantly related to COVID-19 were further selected by weighted gene co-expression network analysis. Then, the hub genes were further analyzed via an examination of their clinical significance in multiple datasets, a correlation analysis of the immune cell infiltration level, and their interactions with the interactome sets of the A549 cell line. RESULTS: A total of 257 susceptibility genes were identified, and these genes were associated with RNA splicing, mitochondrial functions, and proteasomes. Ten genes, MEA1, MRPL24, PPIH, EBNA1BP2, MRTO4, RABEPK, TRMT112, PFDN2, PFDN6, and NDUFS3, were confirmed to be the hub susceptibility genes for COVID-19 in LUAD patients, and the hub susceptibility genes were significantly correlated with the infiltration of multiple immune cells. CONCLUSION: In conclusion, the susceptibility genes for COVID-19 in LUAD patients discovered in this study may increase our understanding of the high risk of COVID-19 in LUAD patients.

Eleven routine clinical features predict COVID-19 severity uncovered by machine learning of longitudinal measurements.

Severity prediction of COVID-19 remains one of the major clinical challenges for the ongoing pandemic. Here, we have recruited a 144 COVID-19 patient cohort, resulting in a data matrix containing 3,065 readings for 124 types of measurements over 52 days. A machine learning model was established to predict the disease progression based on the cohort consisting of training, validation, and internal test sets. A panel of eleven routine clinical factors constructed a classifier for COVID-19 severity prediction, achieving accuracy of over 98% in the discovery set. Validation of the model in an independent cohort containing 25 patients achieved accuracy of 80%. The overall sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were 0.70, 0.99, 0.93, and 0.93, respectively. Our model captured predictive dynamics of lactate dehydrogenase (LDH) and creatine kinase (CK) while their levels were in the normal range. This model is accessible at https://www.guomics.com/covidAI/ for research purpose.

Are population size and diverse climatic conditions the driving factors for next COVID-19 pandemic epicenter in India?

Although a nationwide lockdown was imposed in India amid COVID-19 outbreak since March 24, 2020, the COVID-19 infection is increasing day-by-day. Till June 10, 2021 India has recorded 29,182,072 COVID cases and 359,695 deaths. A number of factors help to influence COVID-19 transmission rate and prevalence. Accordingly, the present study intended to integrate the climatic parameters, namely ambient air temperature (AT) and relative humidity (H) with population mass (PM) to determine their influence for rapid transmission of COVID-19 in India. The sensibility of AT, H and PM parameters on COVID-19 transmission was investigated based on receiver operating characteristics (ROC) classification model. The results depicted that AT and H models have very low sensibility (i.e., lower area under curve value 0.26 and 0.37, respectively compared with AUC value 0.5) to induce virus transmission and discrimination between infected people and healthy ones. Contrarily, PM model is highly sensitive (AUC value is 0.912, greater than AUC value 0.5) towards COVID-19 transmission and discrimination between infected people and healthy ones and approximate population of 2.25 million must impose like social distancing, personal hygiene, etc. as strategic management policy. Therefore, it is predicted, India could be the next epicenter of COVID-19 outbreak because of its over population.

Personalized Prediction of Hospital Mortality in COVID-19-Positive Patients.

OBJECTIVE: To develop predictive models for in-hospital mortality and length of stay (LOS) for coronavirus disease 2019 (COVID-19)-positive patients. PATIENTS AND METHODS: We performed a multicenter retrospective cohort study of hospitalized COVID-19-positive patients. A total of 764 patients admitted to 14 different hospitals within the Cleveland Clinic from March 9, 2020, to May 20, 2020, who had reverse transcriptase-polymerase chain reaction-proven coronavirus infection were included. We used LightGBM, a machine learning algorithm, to predict in-hospital mortality at different time points (after 7, 14, and 30 days of hospitalization) and in-hospital LOS. Our final cohort was composed of 764 patients admitted to 14 different hospitals within our system. RESULTS: The median LOS was 5 (range, 1-44) days for patients admitted to the regular nursing floor and 10 (range, 1-38) days for patients admitted to the intensive care unit. Patients who died during hospitalization were older, initially admitted to the intensive care unit, and more likely to be white and have worse organ dysfunction compared with patients who survived their hospitalization. Using the 10 most important variables only, the final model's area under the receiver operating characteristics curve was 0.86 for 7-day, 0.88 for 14-day, and 0.85 for 30-day mortality in the validation cohort. CONCLUSION: We developed a decision tool that can provide explainable and patient-specific prediction of in-hospital mortality and LOS for COVID-19-positive patients. The model can aid health care systems in bed allocation and distribution of vital resources.

Risk of in-hospital death associated with Covid-19 lung consolidations on chest computed tomography - A novel translational approach using a radiation oncology contour software.

PURPOSE: To determine whether the percentage of lung involvement at the initial chest computed tomography (CT) is related to the subsequent risk of in-hospital death in patients with coronavirus disease-2019 (Covid-19). MATERIALS AND METHODS: Using a cohort of 154 laboratory-confirmed Covid-19 pneumonia cases that underwent chest CT between February and April 2020, we performed a volumetric analysis of the lung opacities. The impact of relative lung involvement on outcomes was evaluated using multivariate logistic regression. The primary endpoint was the in-hospital mortality rate. The secondary endpoint was major adverse hospitalization events (intensive care unit admission, use of mechanical ventilation, or death). RESULTS: The median age of the patients was 65 years: 50.6 % were male, and 36.4 % had a history of smoking. The median relative lung involvement was 28.8 % (interquartile range 9.5-50.3). The overall in-hospital mortality rate was 16.2 %. Thirty-six (26.3 %) patients were intubated. After adjusting for significant clinical factors, there was a 3.6 % increase in the chance of in-hospital mortality (OR 1.036; 95 % confidence interval, 1.010-1.063; P = 0.007) and a 2.5 % increase in major adverse hospital events (OR 1.025; 95 % confidence interval, 1.009-1.042; P = 0.002) per percentage unit of lung involvement. Advanced age (P = 0.013), DNR/DNI status at admission (P < 0.001) and smoking (P = 0.008) also increased in-hospital mortality. Older (P = 0.032) and male patients (P = 0.026) had an increased probability of major adverse hospitalization events. CONCLUSIONS: Among patients hospitalized with Covid-19, more lung consolidation on chest CT increases the risk of in-hospital death, independently of confounding clinical factors.