Early Prediction of Mortality for Septic Patients Visiting Emergency Room Based on Explainable Machine Learning: A Real-World Multicenter Study.

BACKGROUND: Worldwide, sepsis is the leading cause of death in hospitals. If mortality rates in patients with sepsis can be predicted early, medical resources can be allocated efficiently. We constructed machine learning (ML) models to predict the mortality of patients with sepsis in a hospital emergency department. METHODS: This study prospectively collected nationwide data from an ongoing multicenter cohort of patients with sepsis identified in the emergency department. Patients were enrolled from 19 hospitals between September 2019 and December 2020. For acquired data from 3,657 survivors and 1,455 deaths, six ML models (logistic regression, support vector machine, random forest, extreme gradient boosting [XGBoost], light gradient boosting machine, and categorical boosting [CatBoost]) were constructed using fivefold cross-validation to predict mortality. Through these models, 44 clinical variables measured on the day of admission were compared with six sequential organ failure assessment (SOFA) components (PaO2/FIO2 [PF], platelets (PLT), bilirubin, cardiovascular, Glasgow Coma Scale score, and creatinine). The confidence interval (CI) was obtained by performing 10,000 repeated measurements via random sampling of the test dataset. All results were explained and interpreted using Shapley's additive explanations (SHAP). RESULTS: Of the 5,112 participants, CatBoost exhibited the highest area under the curve (AUC) of 0.800 (95% CI, 0.756-0.840) using clinical variables. Using the SOFA components for the same patient, XGBoost exhibited the highest AUC of 0.678 (95% CI, 0.626-0.730). As interpreted by SHAP, albumin, lactate, blood urea nitrogen, and international normalization ratio were determined to significantly affect the results. Additionally, PF and PLTs in the SOFA component significantly influenced the prediction results. CONCLUSION: Newly established ML-based models achieved good prediction of mortality in patients with sepsis. Using several clinical variables acquired at the baseline can provide more accurate results for early predictions than using SOFA components. Additionally, the impact of each variable was identified.

The clinical availability of facial nerve enhancement in temporal bone MRI for the patients of idiopathic acute peripheral facial palsy.

PURPOSE: This study is to evaluate the duration of facial nerve enhancement in gadolinium-enhanced temporal bone MRI after the onset of acute facial palsy. METHODS: Gd-enhanced MRI imagines were examined in 13 patients with idiopathic acute facial palsy within 14 days after the onset. The degree of facial nerve function was measured according to the House-Brackmann (H-B) grading system at their first visit at outpatient clinic. The follow-up MRI was taken about 16.5 months (7-24 months) after onset of disease. The degree of facial nerve enhancement was measured with signal intensity (SI) which was quantitatively analyzed using the region-of-interest (ROI) measurements for each segment of the facial nerve. SI was statistically analyzed by comparing SI values of contralateral site and ipsilateral site using the paired t test with SPSS program. RESULTS: The gadolinium enhancement was statistically increased at labyrinthine segment and geniculate ganglion area of facial nerve at initial temporal bone MRI. The gadolinium enhancement was statistically decreased at all the segments of facial nerve except tympanic segment (p < 0.05) at follow-up MRI. CONCLUSIONS: The facial nerve enhancement in Gd-enhanced MRI images prolonged more than 21 months of the onset. The newly developed pathologic lesions of acute facial palsy especially occur at the site of labyrinthine and geniculate ganglion.

In-Advance Prediction of Pressure Ulcers via Deep-Learning-Based Robust Missing Value Imputation on Real-Time Intensive Care Variables.

Pressure ulcers (PUs) are a prevalent skin disease affecting patients with impaired mobility and in high-risk groups. These ulcers increase patients' suffering, medical expenses, and burden on medical staff. This study introduces a clinical decision support system and verifies it for predicting real-time PU occurrences within the intensive care unit (ICU) by using MIMIC-IV and in-house ICU data. We develop various machine learning (ML) and deep learning (DL) models for predicting PU occurrences in real time using the MIMIC-IV and validate using the MIMIC-IV and Kangwon National University Hospital (KNUH) dataset. To address the challenge of missing values in time series, we propose a novel recurrent neural network model, GRU-D++. This model outperformed other experimental models by achieving the area under the receiver operating characteristic curve (AUROC) of 0.945 for the on-time prediction and AUROC of 0.912 for 48h in-advance prediction. Furthermore, in the external validation with the KNUH dataset, the fine-tuned GRU-D++ model demonstrated superior performances, achieving an AUROC of 0.898 for on-time prediction and an AUROC of 0.897 for 48h in-advance prediction. The proposed GRU-D++, designed to consider temporal information and missing values, stands out for its predictive accuracy. Our findings suggest that this model can significantly alleviate the workload of medical staff and prevent the worsening of patient conditions by enabling timely interventions for PUs in the ICU.

Reappraisal of sepsis-3 and CLIF-SOFA as predictors of mortality in patients with cirrhosis and infection presenting to the emergency department: A multicenter study.

BACKGROUND/AIMS: Sepsis-3 criteria and quick Sequential Organ Failure Assessment (qSOFA) have been advocated to be used in defining sepsis in the general population. We aimed to compare the Sepsis-3 criteria and Chronic Liver Failure-SOFA (CLIF-SOFA) scores as predictors of in-hospital mortality in cirrhotic patients admitted to the emergency department (ED) for infections. METHODS: A total of 1,622 cirrhosis patients admitted at the ED for infections were assessed retrospectively. We analyzed their demographic, laboratory, and microbiological data upon diagnosis of the infection. The primary endpoint was inhospital mortality rate. The predictive performances of baseline CLIF-SOFA, Sepsis-3, and qSOFA scores for in-hospital mortality were evaluated. RESULTS: The CLIF-SOFA score proved to be significantly better in predicting in-hospital mortality (area under the receiver operating characteristic curve [AUROC], 0.80; 95% confidence interval [CI], 0.78-0.82) than the Sepsis-3 (AUROC, 0.75; 95% CI, 0.72-0.77, P<0.001) and qSOFA (AUROC, 0.67; 95% CI, 0.64-0.70; P<0.001) score. The CLIF-SOFA, CLIF-C-AD scores, Sepsis-3 criteria, septic shock, and qSOFA positivity were significantly associated with in-hospital mortality (adjusted hazard ratio [aHR], 1.24; 95% CI, 1.19-1.28; aHR, 1.13; 95% CI, 1.09-1.17; aHR, 1.19; 95% CI, 1.15-1.24; aHR, 1.88; 95% CI, 1.42-2.48; aHR, 2.06; 95% CI, 1.55-2.72; respectively; all P<0.001). For CLIF-SOFA scores ≥6, in-hospital mortality was >10%; this is the cutoff point for the definition of sepsis. CONCLUSION: Among cirrhosis patients presenting with infections at the ED, CLIF-SOFA scores showed a better predictive performance for mortality than both Sepsis-3 criteria and qSOFA scores, and can be a useful tool of risk stratification in cirrhotic patients requiring timely intervention for infection.

Word Embedding Reveals Cyfra 21-1 as a Biomarker for Chronic Obstructive Pulmonary Disease.

BACKGROUND: Although patients with chronic obstructive pulmonary disease (COPD) experience high morbidity and mortality worldwide, few biomarkers are available for COPD. Here, we analyzed potential biomarkers for the diagnosis of COPD by using word embedding. METHODS: To determine which biomarkers are likely to be associated with COPD, we selected respiratory disease-related biomarkers. Degrees of similarity between the 26 selected biomarkers and COPD were measured by word embedding. And we infer the similarity with COPD through the word embedding model trained in the large-capacity medical corpus, and search for biomarkers with high similarity among them. We used Word2Vec, Canonical Correlation Analysis, and Global Vector for word embedding. We evaluated the associations of selected biomarkers with COPD parameters in a cohort of patients with COPD. RESULTS: Cytokeratin 19 fragment (Cyfra 21-1) was selected because of its high similarity and its significant correlation with the COPD phenotype. Serum Cyfra 21-1 levels were determined in patients with COPD and controls (4.3 ± 5.9 vs. 3.9 ± 3.6 ng/mL, P = 0.611). The emphysema index was significantly correlated with the serum Cyfra 21-1 level (correlation coefficient = 0.219, P = 0.015). CONCLUSION: Word embedding may be used for the discovery of biomarkers for COPD and Cyfra 21-1 may be used as a biomarker for emphysema. Additional studies are needed to validate Cyfra 21-1 as a biomarker for COPD.

Correlation between Telomere Length and Chronic Obstructive Pulmonary Disease-Related Phenotypes: Results from the Chronic Obstructive Pulmonary Disease in Dusty Areas (CODA) Cohort.

BACKGROUND: Chronic obstructive pulmonary disease (COPD) is a common chronic respiratory disease with increased prevalence in the elderly. Telomeres are repetitive DNA sequences found at the end of the chromosome, which progressively shorten as cells divide. Telomere length is known to be a molecular marker of aging. This study aimed to assess the relationship between telomere length and the risk of COPD, lung function, respiratory symptoms, and emphysema index in Chronic Obstructive Pulmonary Disease in Dusty Areas (CODA) cohort. METHODS: We extracted DNA from the peripheral blood samples of 446 participants, including 285 COPD patients and 161 control participants. We measured absolute telomere length using quantitative real-time polymerase chain reaction. All participants underwent spirometry and quantitative computed tomography scan. Questionnaires assessing respiratory symptoms and the COPD Assessment Test was filled by all the participants. RESULTS: The mean age of participants at the baseline visit was 72.5±7.1 years. Males accounted for 72% (321 participants) of the all participants. The mean telomere length was lower in the COPD group compared to the non-COPD group (COPD, 16.81±13.90 kb; non-COPD, 21.97±14.43 kb). In COPD patients, 112 (75.7%) were distributed as tertile 1 (shortest), 91 (61.1%) as tertile 2 and 82 (55%) as tertile 3 (longest). We did not find significant associations between telomere length and lung function, exacerbation, airway wall thickness, and emphysema index after adjusting for sex, age, and smoking status. CONCLUSION: In this study, the relationship between various COPD phenotypes and telomere length was analyzed, but no significant statistical associations were shown.

Air Pollution and Incidence of Lung Cancer by Histological Type in Korean Adults: A Korean National Health Insurance Service Health Examinee Cohort Study.

: Studies have reported associations between long-term exposure to ambient air pollution and lung cancer. However, there have been inconsistent reports of such associations with lung cancer by histological type. Thus, the aim of this study was to assess the association of long-term exposure to particulate matter with an aerodynamic diameter up to 10 µm (PM10) and nitrogen dioxide (NO2) with lung cancer incidence by histological subtype in South Korea. This population-based cohort study included 6,567,909 cancer-free subjects from the Korean National Health Insurance Service (NHIS) database for 2006-2007. We linked population data to Korea Central Cancer Registry data to confirm lung cancer incidence for 2006-2013. Individual exposures to PM10 and NO2 were assessed as five-year average concentrations predicted at subjects' district-specific home addresses for 2002-2007. We divided these exposures into two categories based on the 75th percentile. Cox proportional hazards models were used to estimate hazard ratios (HRs) of lung cancer incidence for the upper 25% exposure compared to the low 75% by histological subtypes at diagnosis after adjusting for potential confounders. A total of 27,518 lung cancer were found between 2006 to 2013. The incidence of lung cancer was higher in males, smokers, drinkers and subjects with chronic obstructive pulmonary disease. Overall, we did not find an increased risk of lung cancer with higher exposure to PM10 or NO2. However, high exposure to PM10 was associated with increased risk of adenocarcinoma in comparison with lower exposure in males and current smokers (HR, 1.14; 95% CI, 1.03-1.25). This study showed that long-term air pollution exposures were associated with an elevated risk of lung adenocarcinoma in male smokers in Korea.

Identification of Serial DNA Methylation Changes in the Blood Samples of Patients with Lung Cancer.

BACKGROUND: The development of lung cancer results from the interaction between genetic mutations and dynamic epigenetic alterations, although the exact mechanisms are not completely understood. Changes in DNA methylation may be a promising biomarker for early detection and prognosis of lung cancer. We evaluated the serial changes in genome-wide DNA methylation patterns in blood samples of lung cancer patients. METHODS: Blood samples were obtained for three consecutive years from three patients (2 years before, 1 year before, and after lung cancer detection) and from three control subjects (without lung cancer). We used the MethylationEPIC BeadChip method, which covers the 850,000 bp cytosine-phosphate-guanine (CpG) site, to conduct an epigenome-wide analysis. Significant differentially methylated regions (DMRs) were identified using p-values <0.05 in a correlation test identifying serial methylation changes and serial increase or decrease in ß value above 0.1 for three consecutive years. RESULTS: We found three significant CpG sites with differentially methylated ß values and 7,105 CpG sites with significant correlation from control patients without lung cancer. However, there were no significant DMRs. In contrast, we found 11 significant CpG sites with differentially methylated ß values and 10,562 CpG sites with significant correlation from patients with lung cancer. There were two significant DMRs: cg21126229 (RNF212) and cg27098574 (BCAR1). CONCLUSION: This study revealed DNA methylation changes that might be implicated in lung cancer development. The DNA methylation changes may be the possible candidate target regions for the early detection and prevention of lung cancer.