Construction and validation of prognostic models in critically Ill patients with sepsis-associated acute kidney injury: interpretable machine learning approach.

BACKGROUND: Acute kidney injury (AKI) is a common complication in critically ill patients with sepsis and is often associated with a poor prognosis. We aimed to construct and validate an interpretable prognostic prediction model for patients with sepsis-associated AKI (S-AKI) using machine learning (ML) methods. METHODS: Data on the training cohort were collected from the Medical Information Mart for Intensive Care IV database version 2.2 to build the model, and data of patients were extracted from Hangzhou First People's Hospital Affiliated to Zhejiang University School of Medicine for external validation of model. Predictors of mortality were identified using Recursive Feature Elimination (RFE). Then, random forest, extreme gradient boosting (XGBoost), multilayer perceptron classifier, support vector classifier, and logistic regression were used to establish a prognosis prediction model for 7, 14, and 28 days after intensive care unit (ICU) admission, respectively. Prediction performance was assessed using the receiver operating characteristic (ROC) curve and decision curve analysis (DCA). SHapley Additive exPlanations (SHAP) were used to interpret the ML models. RESULTS: In total, 2599 patients with S-AKI were included in the analysis. Forty variables were selected for the model development. According to the areas under the ROC curve (AUC) and DCA results for the training cohort, XGBoost model exhibited excellent performance with F1 Score of 0.847, 0.715, 0.765 and AUC (95% CI) of 0.91 (0.90, 0.92), 0.78 (0.76, 0.80), and 0.83 (0.81, 0.85) in 7 days, 14 days and 28 days group, respectively. It also demonstrated excellent discrimination in the external validation cohort. Its AUC (95% CI) was 0.81 (0.79, 0.83), 0.75 (0.73, 0.77), 0.79 (0.77, 0.81) in 7 days, 14 days and 28 days group, respectively. SHAP-based summary plot and force plot were used to interpret the XGBoost model globally and locally. CONCLUSIONS: ML is a reliable tool for predicting the prognosis of patients with S-AKI. SHAP methods were used to explain intrinsic information of the XGBoost model, which may prove clinically useful and help clinicians tailor precise management.

Pulmonary disease and periodontal health: a meta-analysis.

OBJECTIVE: Research has shown a positive relationship between pulmonary and periodontal disease. However, the relationship remains unclear. The aim of this meta-analysis was to evaluate the associations between pulmonary disease and periodontal health. MATERIALS AND METHODS: MEDLINE, PubMed, EMBASE, Web of Science, Science Citation Index, Wanfang, and CNKI were searched for all the relevant studies of relationship between pulmonary disease and periodontal health. Weighted mean difference (WMD)/odds ratio (OR) with 95% confidence interval (CI) was used to assess the strength of the association. RESULTS: Thirty-seven studies were included in this meta-analysis. Pooled analysis showed a significant association between pulmonary and periodontal disease (adjusted OR, 1.93; 95%CI: 1.60-2.33; P < 0.05), and pooled adjusted OR was 1.64, 3.03, and 2.21 in COPD, asthma, and pneumonia, respectively. The pooled analysis also showed that patients with pulmonary disease suffered from worse periodontal health as most periodontal indexes in those patients were poorer. CONCLUSIONS: There is a strong association between pulmonary disease and periodontal health. Clinical trials analyzing the causality and pathological basis of the association of these two diseases are needed.

Prevalence, Outcomes, and Risk Factors of New-Onset Atrial Fibrillation in Critically Ill Patients.

The purpose of this article is to systematically evaluate the prevalence, outcomes, and risk factors of new-onset atrial fibrillation (AF) in critically ill patients.Medline, Embase, Science Citation Index, Wanfang, CNKI, and Wiley Online Library were thoroughly searched to identify relevant studies. Studies were assessed for methodological quality using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system. Odds ratio (OR) and weighted mean difference (WMD) with 95% confidence interval (CI) were used to assess the strength of the association. Heterogeneity, subgroup, sensitivity analyses, and publication bias were conducted.A total of 25 studies were included. The prevalence of new-onset AF ranged from 4.1% to 46%.The random-effects pooled prevalence was 10.7%. The pooled result jumped up to 35.8% in patients with septic shock. Pooled analysis showed significant associations between new-onset AF with intensive care unit (ICU) mortality and in-hospital mortality over those patients without AF (OR = 3.11; 95%CI 2.45-3.96 and OR = 1.63; 95%CI 1.27-2.08). The pooled analysis also indicated that both ICU and hospital length of stay are longer in patients with new-onset AF than those without AF (WMD = 1.87; 95%CI 0.89-2.84 and WMD = 2.73; 95%CI 0.77-4.69). Independent risk factors included increasing age, shock, sepsis, use of a pulmonary artery catheter and mechanical ventilation, fluid loading, and organ failures.New-onset AF incidence rate is high in critically ill patients. New-onset AF is associated with worse outcomes. Further studies should be done to explore how to prevent and treat new-onset AF in critically ill patients.

A meta-analysis of obstructive sleep apnea in patients with cerebrovascular disease.

BACKGROUND: There is increasing evidence that there is a direct relationship between obstructive sleep apnea (OSA) and cerebrovascular (CV) disease. This meta-analysis includes prospective cohorts and cross-sectional studies. It determines the prevalence of OSA among patients with CV disease and also looks for the risk o f cerebrovascular events among patients with OSA. METHOD: We conducted a computerized literature search in the databases of Medline, Embase, Wanfang, CNKI, Wiley Online Library, and CINAHL (date till April 2017). The random effects model was used to pool the hazard ratio (HR) and effect sizes (ES). Heterogeneity, subgroup, sensitivity analyses, and publication bias were performed. RESULT: Fifty-eight studies involving 3 million patients/participants were included. The pooled HR of 15 prospective cohort studies indicated a significant association between OSA and the risk of CV disease after adjustment confounding factors (HR 1.94; 95% CI 1.31-2.89; P = 0.001), and OSA is present in up to 58.8% of patients with CV disease after pooling the remaining 43 studies. CONCLUSIONS: OSA is very common in patients with CV disease. Similarly, patients with OSA are prone to have CV disease.

Serum calcium channel subunit α2δ-1 concentrations and outcomes in patients with acute spontaneous intracerebral hemorrhage.

BACKGROUND: Voltage-gated calcium channel subunit α2δ-1 plays an important role in acute brain injury. We attempted to investigate whether serum α2δ-1 subunit concentrations are correlated with severity and prognosis following intracerebral hemorrhage (ICH). METHODS: Serum α2δ-1 subunit concentrations were quantified in 103 ICH patients and 103 healthy controls. National Institutes of Health Stroke Scale (NIHSS) score and hematoma volume were estimated for assessing illness severity. Modified Rankin scale score of 3-6 at 90 days after stroke onset was defined as a worse outcome. RESULTS: Serum α2δ-1 subunit concentrations were markedly higher in patients than in controls (median, 875.1 vs. 209.3 pg/ml). Serum α2δ-1 subunit concentrations of patients were tightly correlated with NIHSS score (r = 0.589) and hematoma volume (r = 0.594). Serum α2δ-1 subunit concentrations ≥ 875.1 pg/ml independently discriminated development of 90-day poor outcome with odds ratio of 5.228 (95% CI, 2.201-12.418) and area under the receiver operating characteristic curve of 0.794 (95% CI, 0.703-0.867). Serum α2δ-1 subunit concentrations > 973.4 pg/ml predicted 90-day poor outcome with 64.0% sensitivity and 90.6% specificity. The prognostic predictive ability of serum α2δ-1 concentrations was equivalent to those of NIHSS score and hematoma volume (both P > 0.05), and serum α2δ-1 concentrations also significantly improved the prognostic predictive capabilities of NIHSS score and hematoma volume (both P < 0.05). CONCLUSIONS: Serum α2δ-1 subunit concentrations are intimately correlated with illness severity and are independently associated with poor 90-day outcome, substantializing serum α2δ-1 subunit as a potential prognostic biomarker for ICH.

[Risk factors of unplanned reoperation after colorectal cancer surgery: a case-control study].

OBJECTIVE: To explore the risk factors of unplanned reoperation after radical resection for colorectal cancer. METHODS: A retrospective analysis of 60 patients (within 14 to 24 days after the initial surgery) receiving unplanned reoperation after colorectal cancer surgery in the First Affiliated Hospital of Xinjiang Medical University from January 2010 to January 2014 was carried out, comparing with 120 randomly paired patients without reoperation during the same period. Univariate and multivariate Logistic regression analysis was performed to investigate the clinicopathologic characteristics of patients in both groups. RESULTS: Univariate logistic regression analysis showed that male, massive blood loss, diabetes, high BMI, hypertension and poorer tumor staging were selected as possible risk factors, and surgeon and laparoscopic surgery as conservative factors (all P<0.05). Multivariate logistic regression analysis revealed that massive blood loss (OR=12.935, 95% CI: 2.267 to 73.806, P=0.004), diabetes (OR=1.747, 95% CI: 1.098 to 2.777, P=0.018) and male (OR=1.805, 95% CI: 1.074 to 3.034, P=0.026) were the independent risk factors of unplanned reoperation after radical resection for colorectal cancer. CONCLUSION: For heavy bleeding, diabetes and male gender in patients with colorectal cancer, surgeon should pay attention to prevent the risk of postoperative unplanned reoperation.

Relationship between single-nucleotide polymorphism of matrix metalloproteinase-2 gene and colorectal cancer and gastric cancer susceptibility: a meta-analysis.

BACKGROUND: Recently, the published data on the association between matrix metalloproteinase-2 (MMP-2) (C-1306T) polymorphism and colorectal cancer (CRC) and gastric cancer (GC) (gastrointestinal cancer) risk remained controversial. The aim of this study is to investigate the relationship between the risk of CRC and GC and single-nucleotide polymorphism of MMP-2(C-1306T). METHODS: Medline, Embase, Science Citation Index, and PubMed were thoroughly searched to identify relevant studies. Odds ratio (OR) with 95% confidence interval (CI) was used to assess the strength of the association. RESULTS: We performed a meta-analysis of 14 studies including 642 cases and 692 controls for CRC and 1,936 cases and 3,490 controls for GC. The result indicates that there is significant relationship between MMP-2(C-1306T) polymorphism and CRC risk in recessive model and codominant model (TT vs CC/CT: OR: 2.39, 95% CI: 1.30-4.37, P=0.005; TT vs CC: OR: 2.36, 95% CI: 1.29-4.34, P=0.006). In subgroup analysis according to ethnicity, significant associations were found in Caucasians (TT vs CC/CT: OR: 2.87, 95% CI: 1.43-5.78, P=0.003; TT vs CC: OR: 2.86, 95% CI: 1.41-5.80, P=0.003), but we did not find significant evidence with GC in all genetic models, and in stratified analysis according to ethnicity, no significant risk was found in the subgroup too. CONCLUSION: This meta-analysis considered that the MMP-2(C-1306T) polymorphism is a risk factor for CRC susceptibility, especially in Caucasians, but it does not support any relationship to GC, and further studies are needed to explore the association.

[Expression and clinical significance of P-gp in colonic gland of chronic schistosomiasis patients].

OBJECTIVE: To study the expression of P-glycoprotein (P-gp) in patients with chronic schistosomiasis and its relationship with gender and age, and to explore its clinical significance. METHODS: The colonic mucosa specimens of 50 chronic schistosomiasis patients and 4 persons who had a family history of colon cancer but their physical examinations were normal (as a control group) were selected and the expressions of P-gp in the colonic gland of these samples were detected with the immunohistochemical staining method. RESULTS: Compared with the control group, the mean gray value of the positive particles of P-gp of colonic epithelial cells of the chronic schistosomiasis group was significantly reduced (P < 0.05), but positive unit values were significantly increased (P < 0.05), which meant the P-gp expression in colonic gland of the patients with chronic schistosomiasis was significantly increased, but this increase had no relationship with sex and age. CONCLUSION: Chronic schistosomiasis may induce the increase of P-gp expression which may be a compensatory protection mechanism due to the stimulation of schistosome eggs to the colon tissues.