Development and validation of nomogram to predict severe illness requiring intensive care follow up in hospitalized COVID-19 cases

Prior knowledge-guided vision-transformer-based unsupervised domain adaptation for intubation prediction in lung disease at one week

Data-driven approaches have achieved great success in various medical image analysis tasks. However, fully-supervised data-driven approaches require unprecedentedly large amounts of labeled data and often suffer from poor generalization to unseen new data due to domain shifts. Various unsupervised domain adaptation (UDA) methods have been actively explored to solve these problems. Anatomical and spatial priors in medical imaging are common and have been incorporated into data-driven approaches to ease the need for labeled data as well as to achieve better generalization and interpretation. Inspired by the effectiveness of recent transformer-based methods in medical image analysis, the adaptability of transformer-based models has been investigated. How to incorporate prior knowledge for transformer-based UDA models remains under-explored. In this paper, we introduce a prior knowledge-guided and transformer-based unsupervised domain adaptation (PUDA) pipeline. It regularizes the vision transformer attention heads using anatomical and spatial prior information that is shared by both the source and target domain, which provides additional insight into the similarity between the underlying data distribution across domains. Besides the global alignment of class tokens, it assigns local weights to guide the token distribution alignment via adversarial training. We evaluate our proposed method on a clinical outcome prediction task, where Computed Tomography (CT) and Chest X-ray (CXR) data are collected and used to predict the intubation status of patients in a week. Abnormal lesions are regarded as anatomical and spatial prior information for this task and are annotated in the source domain scans. Extensive experiments show the effectiveness of the proposed PUDA method.

Establishment of nomogram prediction model of contrast-enhanced ultrasound and Gd-EOB-DTPA-enhanced MRI for vessels encapsulating tumor clusters pattern of hepatocellular carcinoma

To establish clinical prediction models of vessels encapsulating tumor clusters (VETC) pattern using preoperative contrast-enhanced ultrasound (CEUS) and gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid magnetic resonance imaging (EOB-MRI) in patients with hepatocellular carcinoma (HCC). A total of 111 resected HCC lesions from 101 patients were included. Preoperative imaging features of CEUS and EOB-MRI, postoperative recurrence, and survival information were collected from medical records. The best subset regression and multivariable Cox regression were used to select variables to establish the prediction model. The VETC-positive group had a statistically lower survival rate than the VETC-negative group. The selected variables were peritumoral enhancement in the arterial phase (AP), hepatobiliary phase (HBP) on EOB-MRI, intratumoral branching enhancement in the AP of CEUS, intratumoral hypoenhancement in the portal phase of CEUS, incomplete capsule, and tumor size. A nomogram was developed. High and low nomogram scores with a cutoff value of 168 points showed different recurrence-free survival rates and overall survival rates. The area under the curve (AUC) and accuracy were 0.804 and 0.820, respectively, indicating good discrimination. Decision curve analysis showed a good clinical net benefit (threshold probability > 5%), while the Hosmer-Lemeshow test yielded excellent calibration (P = 0.6759). The AUC of the nomogram model combining EOB-MRI and CEUS was higher than that of the models with EOB-MRI factors only (0.767) and CEUS factors only (0.7). The nomogram verified by bootstrapping showed AUC and calibration curves similar to those of the nomogram model. The Prediction model based on CEUS and EOB-MRI is effective for preoperative noninvasive diagnosis of VETC.

Impact of Inflammation-Related Genes on COVID-19: Prospective Study at Turkish Cohort

The pandemic coronavirus disease 2019 (COVID-19) has caused a high mortality rate and poses a significant threat to the population. The disease may progress with mild symptoms or may cause the need for intensive care, depending on many factors. In this study, it was aimed to determine if there is a tendency due to genetic factors in COVID-19 patients. Ninety-four of 188 patients with mild clinical and 94 with severe clinical symptoms were included in the study. The targeted panel including coagulopathy (F2, F5), viral invasion (ACE2), and inflammation (CXCL8, IFNAR2, IFNL4, IL10, IL2, IL6, IRF7, TLR3, TLR7, TNF) related genes was performed sequenced by the next generation sequencing (NGS). The variants found were classified and univariate analyses were performed to select candidate variables for logistic model. Risk factors and variants were compared. It was revealed that the presence of 2 or more risk factors caused the disease to progress severely (p < 0.001). Heterozygous IRF7:c.1357-23dup variant had a 2.5 times higher risk for mild disease compared to severe disease. Other variants were found to be more significant in mild disease. Since polymorphic variants were not evaluated in the literature, the findings of our study could not be compared with the literature. However, as variants that may be effective in the severity of infections may differ according to ethnicity. This study has the feature of being a guide for subsequent studies to be carried out especially in Turkish population. Clinical course of the COVID-19 is likely to depend on a variety of risk factors, including age, sex, clinical status, immunology and genetic factors.

Establishment and Validation of the Risk Nomogram of Poor Prognosis in Patients with Severe Pulmonary Infection Complicated with Respiratory Failure

Objective

To investigate the prognosis of patients with severe pulmonary infection combined with respiratory failure and analyze the influencing factors of prognosis.

Methods

The clinical data of 218 patients with severe pneumonia complicated with respiratory failure were retrospectively analyzed. The risk factors were analyzed by univariate and multivariate logistic regression analyses. The risk nomogram and Bootstrap self-sampling method were used for internal inspection. Calibration curves and receiver operating characteristic (ROC) curve were drawn to assess the predictive ability of the model.

Results

Among 218 patients, 118 (54.13%) cases had a good prognosis and 100 (45.87%) cases had a poor prognosis. Multivariate logistic regression analysis showed that the number of complicated basic diseases ≥5, APACHE II score >20, MODS score >10, PSI score >90, and multi-drug resistant bacterial infection were independent risk factors affecting the prognosis (P<0.05), and the level of Alb was an independent protective factor (P<0.05). The consistency index (C-index) was 0.775, and the Hosmer Lemeshow goodness-of-fit test showed that the model was not significant (P>0.05). The area under the curve (AUC) was 0.813 (95% CI: 0.778~0.895), with the sensitivity of 83.20%, and the specificity of 77.00%.

Conclusion

The risk nomograph model had good discrimination and accuracy in predicting the prognosis of patients with severe pulmonary infection combined with respiratory failure, which may provide a basis for early identification and intervention of patients at clinical risk and improve the prognosis.

Prognostic models in COVID-19 infection that predict severity: a systematic review

Current evidence on COVID-19 prognostic models is inconsistent and clinical applicability remains controversial. We performed a systematic review to summarize and critically appraise the available studies that have developed, assessed and/or validated prognostic models of COVID-19 predicting health outcomes. We searched six bibliographic databases to identify published articles that investigated univariable and multivariable prognostic models predicting adverse outcomes in adult COVID-19 patients, including intensive care unit (ICU) admission, intubation, high-flow nasal therapy (HFNT), extracorporeal membrane oxygenation (ECMO) and mortality. We identified and assessed 314 eligible articles from more than 40 countries, with 152 of these studies presenting mortality, 66 progression to severe or critical illness, 35 mortality and ICU admission combined, 17 ICU admission only, while the remaining 44 studies reported prediction models for mechanical ventilation (MV) or a combination of multiple outcomes. The sample size of included studies varied from 11 to 7,704,171 participants, with a mean age ranging from 18 to 93 years. There were 353 prognostic models investigated, with area under the curve (AUC) ranging from 0.44 to 0.99. A great proportion of studies (61.5%, 193 out of 314) performed internal or external validation or replication. In 312 (99.4%) studies, prognostic models were reported to be at high risk of bias due to uncertainties and challenges surrounding methodological rigor, sampling, handling of missing data, failure to deal with overfitting and heterogeneous definitions of COVID-19 and severity outcomes. While several clinical prognostic models for COVID-19 have been described in the literature, they are limited in generalizability and/or applicability due to deficiencies in addressing fundamental statistical and methodological concerns. Future large, multi-centric and well-designed prognostic prospective studies are needed to clarify remaining uncertainties.

A prediction model for massive hemorrhage in trauma: a retrospective observational study

Background

Massive hemorrhage is the main cause of preventable death after trauma. This study aimed to establish prediction models for early diagnosis of massive hemorrhage in trauma.

Methods

Using the trauma database of Chinese PLA General Hospital, two logistic regression (LR) models were fit to predict the risk of massive hemorrhage in trauma. Sixty-two potential predictive variables, including clinical symptoms, vital signs, laboratory tests, and imaging results, were included in this study. Variable selection was done using the least absolute shrinkage and selection operator (LASSO) method. The first model was constructed based on LASSO feature selection results. The second model was constructed based on the first vital sign recordings of trauma patients after admission. Finally, a web calculator was developed for clinical use.

Results

A total of 2353 patients were included in this study. There were 377 (16.02%) patients with massive hemorrhage. The selected predictive variables were heart rate (OR: 1.01; 95% CI: 1.01–1.02; P<0.001), pulse pressure (OR: 0.99; 95% CI: 0.98–0.99; P = 0.004), base excess (OR: 0.90; 95% CI: 0.87–0.93; P<0.001), hemoglobin (OR: 0.95; 95% CI: 0.95–0.96; P<0.001), displaced pelvic fracture (OR: 2.13; 95% CI: 1.48–3.06; P<0.001), and a positive computed tomography scan or positive focused assessment with sonography for trauma (OR: 1.62; 95% CI: 1.21–2.18; P = 0.001). Model 1, which was developed based on LASSO feature selection results and LR, displayed excellent discrimination (AUC: 0.894; 95% CI: 0.875–0.912), good calibration (P = 0.405), and clinical utility. In addition, the predictive power of model 1 was better than that of model 2 (AUC: 0.718; 95% CI: 0.679–0.757). Model 1 was deployed as a public web tool (http://82.156.217.249:8080/).

Conclusions

Our study developed and validated prediction models to assist medical staff in the early diagnosis of massive hemorrhage in trauma. An open web calculator was developed to facilitate the practical application of the research results.

Predictors of the Outcomes Following the Tocilizumab Treatment for Severe COVID-19

Background: Tocilizumab is used in severe COVID-19 yet has significant rates of treatment failure.

Objectives: This retrospective study aimed to identify early predictors of the response to tocilizumab therapy.

Methods: Biochemical and clinical characteristics of adult patients who received tocilizumab for severe COVID-19 pneumonia were retrospectively examined. A multivariable logistic regression model was constructed to identify factors that could predict the failure of tocilizumab therapy. A predictive nomogram was also created using the selected model.

Results: Out of 101 eligible patients, 30 had treatment failure, and 71 survived on a 28-day follow-up. The partial pressure of oxygen to fraction of inspired oxygen ratio (PFR) on the day of tocilizumab administration (100 vs 80.5), lactate dehydrogenase (LDH; 668 vs 507 U/L), neutrophil-to-lymphocyte ratio (NL ratio; 24.7 vs 10), and creatine kinase myocardial band (CKMB; 30.9 vs 22.7 U/L) were significantly different among the non-survivors and survivors, respectively. A logistic regression model was created, identifying LDH, NL ratio, pro-brain natriuretic peptide (ProBNP), and PFR on the day of tocilizumab administration as best predictors of mortality with an optimism-corrected area under the receiver operator characteristics (ROC) curve of 0.82. The model-implied odds ratios for mortality were 1.89 (95% CI 1.13-3.15) for every 100 U/L rise in serum LDH, 2.29 (95% CI 2.2-4.39) for every 10 unit rise in NL ratio, 1.23 (95% CI 0.95-1.58) for every 100 pg/ml increase in ProBNP, and 0.36 (95% CI 0.13-0.95) for every mmHg rise in PFR at intervention.

Conclusion: This study identified NL ratio, LDH, CKMB, and PFR at intervention as important markers of risk of treatment failure following the tocilizumab therapy. A multivariable logistic regression model including LDH, NL ratio, ProBNP, and PFR at intervention best predicted the risk of mortality in patients with severe COVID-19 pneumonia treated with tocilizumab.

Predictive Risk Factors at Admission and a “Burning Point” During Hospitalization Serve as Sequential Alerts for Critical Illness in Patients With COVID-19

Background

We intended to establish a novel critical illness prediction system combining baseline risk factors with dynamic laboratory tests for patients with coronavirus disease 2019 (COVID-19).

Methods

We evaluated patients with COVID-19 admitted to Wuhan West Union Hospital between 12 January and 25 February 2020. The data of patients were collected, and the illness severity was assessed.

Results

Among 1,150 enrolled patients, 296 (25.7%) patients developed into critical illness. A baseline nomogram model consists of seven variables including age [odds ratio (OR), 1.028; 95% confidence interval (CI), 1.004–1.052], sequential organ failure assessment (SOFA) score (OR, 4.367; 95% CI, 3.230–5.903), neutrophil-to-lymphocyte ratio (NLR; OR, 1.094; 95% CI, 1.024–1.168), D-dimer (OR, 1.476; 95% CI, 1.107–1.968), lactate dehydrogenase (LDH; OR, 1.004; 95% CI, 1.001–1.006), international normalised ratio (INR; OR, 1.027; 95% CI, 0.999–1.055), and pneumonia area interpreted from computed tomography (CT) images (medium vs. small [OR, 4.358; 95% CI, 2.188–8.678], and large vs. small [OR, 9.567; 95% CI, 3.982–22.986]) were established to predict the risk for critical illness at admission. The differentiating power of this nomogram scoring system was perfect with an area under the curve (AUC) of 0.960 (95% CI, 0.941–0.972) in the training set and an AUC of 0.958 (95% CI, 0.936–0.980) in the testing set. In addition, a linear mixed model (LMM) based on dynamic change of seven variables consisting of SOFA score (value, 2; increase per day [I/d], +0.49), NLR (value, 10.61; I/d, +2.07), C-reactive protein (CRP; value, 46.9 mg/L; I/d, +4.95), glucose (value, 7.83 mmol/L; I/d, +0.2), D-dimer (value, 6.08 μg/L; I/d, +0.28), LDH (value, 461 U/L; I/d, +13.95), and blood urea nitrogen (BUN value, 6.51 mmol/L; I/d, +0.55) were established to assist in predicting occurrence time of critical illness onset during hospitalization.

Conclusion

The two-checkpoint system could assist in accurately and dynamically predicting critical illness and timely adjusting the treatment regimen for patients with COVID-19.

Novel biomarker-based score (SAD-60) for predicting mortality in patients with COVID-19 pneumonia: a multicenter retrospective cohort of 1013 patients

Background: The aim was to explore a novel risk score to predict mortality in hospitalized patients with COVID-19 pneumonia. Methods: This was a retrospective, multicenter study. Results: A total of 1013 patients with COVID-19 were included. The mean age was 60.5 ± 14.4 years, and 581 (57.4%) patients were male. In-hospital death occurred in 124 (12.2%) patients. Multivariate analysis revealed peripheral capillary oxygen saturation (SpO2), albumin, D-dimer and age as independent predictors. The mortality score model was given the acronym SAD-60, representing SpO2, Albumin, D-dimer, age ≥60 years. The SAD-60 score (0.776) had the highest area under the curve compared with CURB-65 (0.753), NEWS2 (0.686) and qSOFA (0.628) scores. Conclusion: The SAD-60 score has a promising predictive capacity for mortality in hospitalized patients with COVID-19.