Neural network analysis of clinical variables predicts escalated care in COVID-19 patients: a retrospective study

Predicting omicron pneumonia severity and outcome: a single-center study in Hangzhou, China

Background

In December 2022, there was a large Omicron epidemic in Hangzhou, China. Many people were diagnosed with Omicron pneumonia with variable symptom severity and outcome. Computed tomography (CT) imaging has been proven to be an important tool for COVID-19 pneumonia screening and quantification. We hypothesized that CT-based machine learning algorithms can predict disease severity and outcome in Omicron pneumonia, and we compared its performance with the pneumonia severity index (PSI)-related clinical and biological features.

Methods

Our study included 238 patients with the Omicron variant who have been admitted to our hospital in China from 15 December 2022 to 16 January 2023 (the first wave after the dynamic zero-COVID strategy stopped). All patients had a positive real-time polymerase chain reaction (PCR) or lateral flow antigen test for SARS-CoV-2 after vaccination and no previous SARS-CoV-2 infections. We recorded patient baseline information pertaining to demographics, comorbid conditions, vital signs, and available laboratory data. All CT images were processed with a commercial artificial intelligence (AI) algorithm to obtain the volume and percentage of consolidation and infiltration related to Omicron pneumonia. The support vector machine (SVM) model was used to predict the disease severity and outcome.

Results

The receiver operating characteristic (ROC) area under the curve (AUC) of the machine learning classifier using PSI-related features was 0.85 (accuracy = 87.40%, p < 0.001) for predicting severity while that using CT-based features was only 0.70 (accuracy = 76.47%, p = 0.014). If combined, the AUC was not increased, showing 0.84 (accuracy = 84.03%, p < 0.001). Trained on outcome prediction, the classifier reached the AUC of 0.85 using PSI-related features (accuracy = 85.29%, p < 0.001), which was higher than using CT-based features (AUC = 0.67, accuracy = 75.21%, p < 0.001). If combined, the integrated model showed a slightly higher AUC of 0.86 (accuracy = 86.13%, p < 0.001). Oxygen saturation, IL-6, and CT infiltration showed great importance in both predicting severity and outcome.

Conclusion

Our study provided a comprehensive analysis and comparison between baseline chest CT and clinical assessment in disease severity and outcome prediction in Omicron pneumonia. The predictive model accurately predicts the severity and outcome of Omicron infection. Oxygen saturation, IL-6, and infiltration in chest CT were found to be important biomarkers. This approach has the potential to provide frontline physicians with an objective tool to manage Omicron patients more effectively in time-sensitive, stressful, and potentially resource-constrained environments.

Longitudinal medical subspecialty follow-up of critically and non-critically ill hospitalized COVID-19 survivors up to 24 months after discharge

Medical specialty usage of COVID-19 survivors after hospital discharge is poorly understood. This study investigated medical specialty usage at 1-12 and 13-24 months post-hospital discharge in critically ill and non-critically ill COVID-19 survivors. This retrospective study followed ICU (N = 89) and non-ICU (N = 205) COVID-19 survivors who returned for follow-up within the Stony Brook Health System post-hospital discharge. Follow-up data including survival, hospital readmission, ongoing symptoms, medical specialty care use, and ICU status were examined 1-12 and 13-24 months after COVID-19 discharge. "New" (not previously seen) medical specialty usage was also identified. Essentially all (98%) patients survived. Hospital readmission was 34%, but functional status scores at discharge were not associated with hospital readmission. Many patients reported ongoing [neuromuscular (50%) respiratory (39%), chronic fatigue (35%), cardiovascular (30%), gastrointestinal (28%), neurocognitive (22%), genitourinary (22%), and mood-related (13%)] symptoms at least once 1-24 months after discharge. Common specialty follow-ups included cardiology (25%), vascular medicine (17%), urology (17%), neurology (16%), and pulmonology (14%), with some associated with pre-existing comorbidities and with COVID-19. Common new specialty visits were vascular medicine (11%), pulmonology (11%), and neurology (9%). ICU patients had more symptoms and follow-ups compared to the non-ICU patients. This study reported high incidence of persistent symptoms and medical specialty care needs in hospitalized COVID-19 survivors 1-24 months post-discharge. Some specialty care needs were COVID-19 related or exacerbated by COVID-19 disease while others were associated with pre-existing medical conditions. Longer follow-up studies of COVID-19 survivor medical care needs are necessary.

Deep learning of longitudinal chest X-ray and clinical variables predicts duration on ventilator and mortality in COVID-19 patients

Objectives

To use deep learning of serial portable chest X-ray (pCXR) and clinical variables to predict mortality and duration on invasive mechanical ventilation (IMV) for Coronavirus disease 2019 (COVID-19) patients.

Methods

This is a retrospective study. Serial pCXR and serial clinical variables were analyzed for data from day 1, day 5, day 1–3, day 3–5, or day 1–5 on IMV (110 IMV survivors and 76 IMV non-survivors). The outcome variables were duration on IMV and mortality. With fivefold cross-validation, the performance of the proposed deep learning system was evaluated by receiver operating characteristic (ROC) analysis and correlation analysis.

Results

Predictive models using 5-consecutive-day data outperformed those using 3-consecutive-day and 1-day data. Prediction using data closer to the outcome was generally better (i.e., day 5 data performed better than day 1 data, and day 3–5 data performed better than day 1–3 data). Prediction performance was generally better for the combined pCXR and non-imaging clinical data than either alone. The combined pCXR and non-imaging data of 5 consecutive days predicted mortality with an accuracy of 85 ± 3.5% (95% confidence interval (CI)) and an area under the curve (AUC) of 0.87 ± 0.05 (95% CI) and predicted the duration needed to be on IMV to within 2.56 ± 0.21 (95% CI) days on the validation dataset.

Conclusions

Deep learning of longitudinal pCXR and clinical data have the potential to accurately predict mortality and duration on IMV in COVID-19 patients. Longitudinal pCXR could have prognostic value if these findings can be validated in a large, multi-institutional cohort.

Fine-Tuned Siamese Network with Modified Enhanced Super-Resolution GAN Plus Based on Low-Quality Chest X-ray Images for COVID-19 Identification

Coronavirus disease has rapidly spread globally since early January of 2020. With millions of deaths, it is essential for an automated system to be utilized to aid in the clinical diagnosis and reduce time consumption for image analysis. This article presents a generative adversarial network (GAN)-based deep learning application for precisely regaining high-resolution (HR) CXR images from low-resolution (LR) CXR correspondents for COVID-19 identification. Respectively, using the building blocks of GAN, we introduce a modified enhanced super-resolution generative adversarial network plus (MESRGAN+) to implement a connected nonlinear mapping collected from noise-contaminated low-resolution input images to produce deblurred and denoised HR images. As opposed to the latest trends of network complexity and computational costs, we incorporate an enhanced VGG19 fine-tuned twin network with the wavelet pooling strategy in order to extract distinct features for COVID-19 identification. We demonstrate our proposed model on a publicly available dataset of 11,920 samples of chest X-ray images, with 2980 cases of COVID-19 CXR, healthy, viral and bacterial cases. Our proposed model performs efficiently both on the binary and four-class classification. The proposed method achieves accuracy of 98.8%, precision of 98.6%, sensitivity of 97.5%, specificity of 98.9%, an F1 score of 97.8% and ROC AUC of 98.8% for the multi-class task, while, for the binary class, the model achieves accuracy of 99.7%, precision of 98.9%, sensitivity of 98.7%, specificity of 99.3%, an F1 score of 98.2% and ROC AUC of 99.7%. Our method obtains state-of-the-art (SOTA) performance, according to the experimental results, which is helpful for COVID-19 screening. This new conceptual framework is proposed to play an influential role in addressing the issues facing COVID-19 examination and other diseases.

Longitudinal prediction of hospital-acquired acute kidney injury in COVID-19: a two-center study

BACKGROUND

To investigate the temporal characteristics of clinical variables of hospital-acquired acute kidney injury (AKI) in COVID-19 patients and to longitudinally predict AKI onset.

METHODS

There were 308 hospital-acquired AKI and 721 non-AKI (NAKI) COVID-19 patients from Stony Brook Hospital (New York, USA) data, and 72 hospital-acquired AKI and 303 NAKI COVID-19 patients from Tongji Hospital (Wuhan, China). Demographic, comorbidities, and longitudinal (3 days before and 3 days after AKI onset) clinical variables were used to compute odds ratios for and longitudinally predict hospital-acquired AKI onset.

RESULTS

COVID-19 patients with AKI were more likely to die than NAKI patients (31.5% vs 6.9%, adjusted p < 0.001, OR = 4.67 [95% CI 3.1, 7.0], Stony Brook data). AKI developed on average 3.3 days after hospitalization. Procalcitonin was elevated prior to AKI onset (p < 0.05), peaked, and remained elevated (p < 0.05). Alanine aminotransferase, aspartate transaminase, ferritin, and lactate dehydrogenase peaked the same time as creatinine, whereas D-dimer and brain natriuretic peptide peaked a day later. C-reactive protein, white blood cell and lymphocyte showed group differences - 2 days prior (p < 0.05). Top predictors were creatinine, procalcitonin, white blood cells, lactate dehydrogenase, and lymphocytes. They predicted AKI onset with areas under curves (AUCs) of 0.78, 0.66, and 0.56 at 0, - 1, and - 2 days prior, respectively. When tested on the Tongji Hospital data, the AUCs were 0.80, 0.79, and 0.77, respectively.

CONCLUSIONS

Time-locked longitudinal data provide insight into AKI progression. Commonly clinical variables reasonably predict AKI onset a few days prior. This work may lead to earlier recognition of AKI and treatment to improve clinical outcomes.

Kynurenine and Hemoglobin as Sex-Specific Variables in COVID-19 Patients: A Machine Learning and Genetic Algorithms Approach

Differences in clinical manifestations, immune response, metabolic alterations, and outcomes (including disease severity and mortality) between men and women with COVID-19 have been reported since the pandemic outbreak, making it necessary to implement sex-specific biomarkers for disease diagnosis and treatment. This study aimed to identify sex-associated differences in COVID-19 patients by means of a genetic algorithm (GALGO) and machine learning, employing support vector machine (SVM) and logistic regression (LR) for the data analysis. Both algorithms identified kynurenine and hemoglobin as the most important variables to distinguish between men and women with COVID-19. LR and SVM identified C10:1, cough, and lysoPC a 14:0 to discriminate between men with COVID-19 from men without, with LR being the best model. In the case of women with COVID-19 vs. women without, SVM had a higher performance, and both models identified a higher number of variables, including 10:2, lysoPC a C26:0, lysoPC a C28:0, alpha-ketoglutaric acid, lactic acid, cough, fever, anosmia, and dysgeusia. Our results demonstrate that differences in sexes have implications in the diagnosis and outcome of the disease. Further, genetic and machine learning algorithms are useful tools to predict sex-associated differences in COVID-19.

Characterizing non-critically ill COVID-19 survivors with and without in-hospital rehabilitation

This study investigated pre-COVID-19 admission dependency, discharge assistive equipment, discharge medical follow-up recommendation, and functional status at hospital discharge of non-critically ill COVID-19 survivors, stratified by those with (N = 155) and without (N = 162) in-hospital rehabilitation. “Mental Status”, intensive-care-unit (ICU) Mobility, and modified Barthel Index scores were assessed at hospital discharge. Relative to the non-rehabilitation patients, rehabilitation patients were older, had more comorbidities, worse pre-admission dependency, were discharged with more assistive equipment and supplemental oxygen, spent more days in the hospital, and had more hospital-acquired acute kidney injury, acute respiratory failure, and more follow-up referrals (p < 0.05 for all). Cardiology, vascular medicine, urology, and endocrinology were amongst the top referrals. Functional scores of many non-critically ill COVID-19 survivors were abnormal at discharge (p < 0.05) and were associated with pre-admission dependency (p < 0.05). Some functional scores were negatively correlated with age, hypertension, coronary artery disease, chronic kidney disease, psychiatric disease, anemia, and neurological disorders (p < 0.05). In-hospital rehabilitation providing restorative therapies and assisting discharge planning were challenging in COVID-19 circumstances. Knowledge of the functional status, discharge assistive equipment, and follow-up medical recommendations at discharge could enable appropriate and timely post-discharge care. Follow-up studies of COVID-19 survivors are warranted as many will likely have significant post-acute COVID-19 sequela.

Predicting Mechanical Ventilation and Mortality in COVID-19 Using Radiomics and Deep Learning on Chest Radiographs: A Multi-Institutional Study

In this study, we aimed to predict mechanical ventilation requirement and mortality using computational modeling of chest radiographs (CXRs) for coronavirus disease 2019 (COVID-19) patients. This two-center, retrospective study analyzed 530 deidentified CXRs from 515 COVID-19 patients treated at Stony Brook University Hospital and Newark Beth Israel Medical Center between March and August 2020. Linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), and random forest (RF) machine learning classifiers to predict mechanical ventilation requirement and mortality were trained and evaluated using radiomic features extracted from patients' CXRs. Deep learning (DL) approaches were also explored for the clinical outcome prediction task and a novel radiomic embedding framework was introduced. All results are compared against radiologist grading of CXRs (zone-wise expert severity scores). Radiomic classification models had mean area under the receiver operating characteristic curve (mAUCs) of 0.78 ± 0.05 (sensitivity = 0.72 ± 0.07, specificity = 0.72 ± 0.06) and 0.78 ± 0.06 (sensitivity = 0.70 ± 0.09, specificity = 0.73 ± 0.09), compared with expert scores mAUCs of 0.75 ± 0.02 (sensitivity = 0.67 ± 0.08, specificity = 0.69 ± 0.07) and 0.79 ± 0.05 (sensitivity = 0.69 ± 0.08, specificity = 0.76 ± 0.08) for mechanical ventilation requirement and mortality prediction, respectively. Classifiers using both expert severity scores and radiomic features for mechanical ventilation (mAUC = 0.79 ± 0.04, sensitivity = 0.71 ± 0.06, specificity = 0.71 ± 0.08) and mortality (mAUC = 0.83 ± 0.04, sensitivity = 0.79 ± 0.07, specificity = 0.74 ± 0.09) demonstrated improvement over either artificial intelligence or radiologist interpretation alone. Our results also suggest instances in which the inclusion of radiomic features in DL improves model predictions over DL alone. The models proposed in this study and the prognostic information they provide might aid physician decision making and efficient resource allocation during the COVID-19 pandemic.

Validation of Neutrophil-to-Lymphocyte Ratio Cut-off Value Associated with High In-Hospital Mortality in COVID-19 Patients

Introduction

The neutrophil-to-lymphocyte ratio (NLR) could be a predictive factor of severe COVID-19. However, most relevant studies are retrospective, and the optimal NLR cut-off point has not been determined. The objective of our research was identification and validation of the best NLR cut-off value on admission that could predict high in-hospital mortality in COVID-19 patients.

Methods

Medical files of all patients admitted for COVID-19 pneumonia in our dedicated COVID-units between March and April 2020 (derivation cohort) and between October and December 2020 (validation cohort) were reviewed.

Results

Two hundred ninety-nine patients were included in the study (198 in the derivation and 101 in the validation cohort, respectively). Youden’s J statistic in the derivation cohort determined the optimal cut-off value for the performance of NLR at admission to predict mortality in hospitalized patients with COVID-19. The NLR cut-off value of 5.94 had a sensitivity of 62% and specificity of 64%. In ROC curve analysis, the AUC was 0.665 [95% CI 0.530–0.801, p= 0.025]. In the validation cohort, the best predictive cut-off value of NLR was 6.4, which corresponded to a sensitivity of 63% and a specificity of 64% with AUC 0.766 [95% CI 0.651–0.881, p <0.001]. When the NLR cut-off value of 5.94 was applied in the validation cohort, there was no significant difference in death and survival in comparison with the derivation NLR cut-off. Net reclassification improvement (NRI) analysis showed no significant classification change in outcome between both NLR cut-off values (NRI:0.012, p=0.31).

Conclusion

In prospective analysis, an NLR value of 5.94 predicted high in-hospital mortality upon admission in patients hospitalized for COVID-19 pneumonia.

Machine Learning Approaches to Identify Patient Comorbidities and Symptoms That Increased Risk of Mortality in COVID-19

Providing appropriate care for people suffering from COVID-19, the disease caused by the pandemic SARS-CoV-2 virus, is a significant global challenge. Many individuals who become infected may have pre-existing conditions that may interact with COVID-19 to increase symptom severity and mortality risk. COVID-19 patient comorbidities are likely to be informative regarding the individual risk of severe illness and mortality. Determining the degree to which comorbidities are associated with severe symptoms and mortality would thus greatly assist in COVID-19 care planning and provision. To assess this we performed a meta-analysis of published global literature, and machine learning predictive analysis using an aggregated COVID-19 global dataset. Our meta-analysis suggested that chronic obstructive pulmonary disease (COPD), cerebrovascular disease (CEVD), cardiovascular disease (CVD), type 2 diabetes, malignancy, and hypertension as most significantly associated with COVID-19 severity in the current published literature. Machine learning classification using novel aggregated cohort data similarly found COPD, CVD, CKD, type 2 diabetes, malignancy, and hypertension, as well as asthma, as the most significant features for classifying those deceased versus those who survived COVID-19. While age and gender were the most significant predictors of mortality, in terms of symptom-comorbidity combinations, it was observed that Pneumonia-Hypertension, Pneumonia-Diabetes, and Acute Respiratory Distress Syndrome (ARDS)-Hypertension showed the most significant associations with COVID-19 mortality. These results highlight the patient cohorts most likely to be at risk of COVID-19-related severe morbidity and mortality, which have implications for prioritization of hospital resources.