Development and External Validation of a Deep Learning Algorithm for Prognostication of Cardiovascular Outcomes

Background and Objectives

We aim to explore the additional discriminative accuracy of a deep learning (DL) algorithm using repeated-measures data for identifying people at high risk for cardiovascular disease (CVD), compared to Cox hazard regression.

Methods

Two CVD prediction models were developed from National Health Insurance Service-Health Screening Cohort (NHIS-HEALS): a Cox regression model and a DL model. Performance of each model was assessed in the internal and 2 external validation cohorts in Koreans (National Health Insurance Service-National Sample Cohort; NHIS-NSC) and in Europeans (Rotterdam Study). A total of 412,030 adults in the NHIS-HEALS; 178,875 adults in the NHIS-NSC; and the 4,296 adults in Rotterdam Study were included.

Results

Mean ages was 52 years (46% women) and there were 25,777 events (6.3%) in NHIS-HEALS during the follow-up. In internal validation, the DL approach demonstrated a C-statistic of 0.896 (95% confidence interval, 0.886–0.907) in men and 0.921 (0.908–0.934) in women and improved reclassification compared with Cox regression (net reclassification index [NRI], 24.8% in men, 29.0% in women). In external validation with NHIS-NSC, DL demonstrated a C-statistic of 0.868 (0.860–0.876) in men and 0.889 (0.876–0.898) in women, and improved reclassification compared with Cox regression (NRI, 24.9% in men, 26.2% in women). In external validation applied to the Rotterdam Study, DL demonstrated a C-statistic of 0.860 (0.824–0.897) in men and 0.867 (0.830–0.903) in women, and improved reclassification compared with Cox regression (NRI, 36.9% in men, 31.8% in women).

Conclusions

A DL algorithm exhibited greater discriminative accuracy than Cox model approaches.

Trial Registration

ClinicalTrials.gov Identifier: NCT02931500

Colon Transit Time May Predict Inadequate Bowel Preparation in Patients With Chronic Constipation

Background/Aims

We evaluated whether colonic transit time (CTT) can predict the degree of bowel preparation in patients with chronic constipation undergoing scheduled colonoscopy in order to assist in the development of better bowel preparation strategies for these patients.

Methods

We analyzed the records of 160 patients with chronic constipation from March 2007 to November 2012. We enrolled patients who had undergone a CTT test followed by colonoscopy. We defined patients with a CTT ≥30 hours as the slow transit time (STT) group, and patients with a CTT <30 hours as the normal transit time (NTT) group. Boston Bowel Preparation Scale (BBPS) scores were compared between the STT and NTT groups.

Results

Of 160 patients with chronic constipation, 82 (51%) were included in the STT group and 78 (49%) were included in the NTT group. Patients with a BBPS score of <6 were more prevalent in the STT group than in the NTT group (31.7% vs. 10.3%, P=0.001). Multivariate analysis showed that slow CTT was an independent predictor of inadequate bowel preparation (odds ratio, 0.261; 95% confidence interval, 0.107-0.634; P=0.003). The best CTT cut-off value for predicting inadequate bowel preparation in patients with chronic constipation was 37 hours, as determined by receiver operator characteristic (ROC) curve analysis (area under the ROC curve: 0.676, specificity: 0.735, sensitivity: 0.643).

Conclusions

Patients with chronic constipation and a CTT >30 hours were at risk for inadequate bowel preparation. CTT measured prior to colonoscopy could be useful for developing individualized strategies for bowel preparation in patients with slow CTT, as these patients are likely to have inadequate bowel preparation.

Relationship between tetrahydrobiopterin and portal hypertension in patients with chronic liver disease

Tetrahydrobiopterin (BH4) is an essential cofactor in NO synthesis by endothelial nitric oxide synthase (eNOS) enzymes. It has been previously suggested that reduced intrahepatic BH4 results in a decrease in intrahepatic NO and contributes to increased hepatic vascular resistance and portal pressure in animal models of cirrhosis. The main aim of the present study was to evaluate the relationship between BH4 and portal hypertension (PHT). One hundred ninety-three consecutive patients with chronic liver disease were included in the study. Liver biopsy, measurement of BH4 and hepatic venous pressure gradient (HVPG) were performed. Hepatic fibrosis was classified using the Laennec fibrosis scoring system. BH4 levels were determined in homogenized liver tissues of patients using a high performance liquid chromatography (HPLC) system. Statistical analysis was performed to evaluate the relationship between BH4 and HVPG, grade of hepatic fibrosis, clinical stage of cirrhosis, Child-Pugh class. A positive relationship between HVPG and hepatic fibrosis grade, clinical stage of cirrhosis and Child-Pugh class was observed. However, the BH4 level showed no significant correlation with HVPG or clinical features of cirrhosis. BH4 concentration in liver tissue has little relation to the severity of portal hypertension in patients with chronic liver disease.

The usefulness of non-invasive liver stiffness measurements in predicting clinically significant portal hypertension in cirrhotic patients: Korean data

BACKGROUND/AIMS

Liver stiffness measurement (LSM) has been proposed as a non-invasive method for estimating the severity of fibrosis and the complications of cirrhosis. Measurement of the hepatic venous pressure gradient (HVPG) is the gold standard for assessing the presence of portal hypertension, but its invasiveness limits its clinical application. In this study we evaluated the relationship between LSM and HVPG, and the predictive value of LSM for clinically significant portal hypertension (CSPH) and severe portal hypertension in cirrhosis.

METHODS

LSM was performed with transient elastography in 59 consecutive cirrhotic patients who underwent hemodynamic HVPG investigations. CSPH and severe portal hypertension were defined as HVPG ≥10 and ≥12 mmHg, respectively. Linear regression analysis was performed to evaluate the relationship between LSM and HVPG. Diagnostic values were analyzed based on receiver operating characteristic (ROC) curves.

RESULTS

A strong positive correlation between LSM and HVPG was observed in the overall population (r(2)=0.496, P<0.0001). The area under the ROC curve (AUROC) for the prediction of CSPH (HVPG ≥10 mmHg) was 0.851, and the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for an LSM cutoff value of 21.95 kPa were 82.5%, 73.7%, 86.8%, and 66.7%, respectively. The AUROC at prediction of severe portal hypertension (HVPG ≥12 mmHg) was 0.877, and the sensitivity, specificity, PPV, and NPV at LSM cutoff value of 24.25 kPa were 82.9%, 70.8%, 80.6%, and 73.9%, respectively.

CONCLUSIONS

LSM exhibited a significant correlation with HVPG in patients with cirrhosis. LSM could be a non-invasive method for predicting CSPH and severe portal hypertension in Korean patients with liver cirrhosis.

Complete endoscopic resection of very early stage gastric plasmacytoma

Gastric plasmacytomas are very rare, and most are not detected until the disease has progressed to an advanced stage. However, there have been recent reports of cases of early-stage gastric plasmacytoma, in which neoplastic cells are confined to the mucosa or submucosa. Here we report a case of a very early stage gastric plasmacytoma that was confined to the lamina propria of the gastric mucosa. The lesion was successfully and completely removed by endoscopic submucosal dissection, and the surveillance endoscopy showed no recurrence during the follow-up of 40 months. This report appears to be the first documented case of complete endoscopic removal of a primary gastric plasmacytoma.