Area Under ROC Curve Research Articles

The prognostic value of HEART score in patients with known heart failure evaluated at the emergency department for acute chest pain

Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main funding source(s): National Heart Lung and Blood Institute (United States). Background Chest pain is a common and concerning symptom in emergency care, often necessities further risk stratification for adverse cardiovascular events on short-term follow up. Patients with known heart failure (HF) constitutes a unique sub-population that would greatly benefit from short-term risk stratification to guide appropriate clinical interventions and optimize resource allocation. HEART (History, ECG, Age, Risk factors, Troponin) is a well-validated risk score that has been shown to outperform other clinical risk scores in patients with chest pain. It remains unknown if HEART score can be safely used in patients with known HF presenting to the emergency department with chest pain. Purpose We sought to evaluate the prognostic value of HEART score in patients with known HF compared to their counterparts. Method This was a prospective observational cohort study of consecutive adult patients evaluated at the emergency department for non-traumatic chest pain. HEART score was computed from charts by independent physician reviewers. The primary study outcome was major adverse cardiac events (MACE) defined as acute coronary syndrome, invasive revascularization, reinfarction, death, cardiac arrest, ventricular tachyarrhythmia, or acute exacerbation within 30 days of indexed admission. Outcomes were adjudicated by two independent reviewers (Kappa = 0.908). Classification performance between groups was compared using area under receiver operator characteristic curve (AUROC), sensitivity, and negative predictive value (NPV). Results Our cohort included 4,132 patients (age 59±16, 47% female, 33% Black). Overall, 921 patients (22.3%) experienced 30-day MACE. The classification performance of HEART score in those without HF (n = 3,329, 81%) was AUROC of 0.831 (95% CI 0.813-0.848), compared to only 0.633 (95% CI 0.592-0.673) in those with known HF (n = 803, 19%) (z score = 8.771, p < 0.001, Figure). The sensitivity and NPV of HEART score among those without HF were 87.4% and 95.3%, compared to 81.2% and 73.3%, respectively, among those with known HF. Conclusion Our findings suggest that the prognostic value of HEART score for predicting 30-day MACE is inferior and suboptimal in patients with known HF evaluated for chest pain at the emergency department compared to their counterparts. These findings emphasize the need for more tailored risk assessment and management strategies in this specific patient population. Further research is warranted to better understand the limitations and potential enhancements of the HEART score in patients with HF.Classification Performance of HEART score

Diabetes Prediction Using Machine Learning

Diabetes, characterized by high blood sugar levels over time, poses challenges for accurate prediction due to limited labelled data and data quality issues like outliers and missing values. To address these challenges, our research introduces a robust prediction framework. This framework integrates techniques such as outlier rejection, filling missing values, standardizing data, selecting relevant features, and employing K-fold cross-validation. We utilize a range of machine learning algorithms including k-nearest Neighbour, Decision Trees, Random Forest, AdaBoost, Naive Bayes, XGBoost, and Multilayer Perceptron (MLP). Additionally, we propose a weighted ensembling technique to enhance prediction accuracy. The weights for ensembling are determined based on the performance of each classifier using the Area Under ROC Curve (AUC) metric. We optimize model performance through hyperparameter tuning using grid search. Our experiments are conducted using the Pima Indian Diabetes Dataset under uniform conditions. The ensembling classifier we propose achieves superior performance with a sensitivity of 0.789, specificity of 0.934, false omission rate of 0.092, diagnostic odds ratio of 66.234, and AUC of 0.950. This outperforms existing methods by 2.00% in terms of AUC. Our framework surpasses other approaches discussed in the literature and shows promise for improving diabetes prediction. We have released our source code for diabetes prediction to the public.

Prognostic factors and novel prediction models for overall survival of patients with submandibular gland cancer: A population-based retrospective cohort study

BackgroundAccurately predicting the survival rate of submandibular gland cancer (SGC) is of significant importance for guiding treatment decision-making and improving patient outcomes. This study was aimed to identify the independent prognostic factors of overall survival (OS) in SGC patients, and develop novel prediction models to aid clinicians in predicting the survival probability. Materials and methodsPatients diagnosed with primary SGC after the year 2010 were extracted from SEER database and then randomly allocated into training and test samples in a 7:3 ratio. Uni- and multi-variable COX analyses were employed using the training sample to ascertain independent prognostic factors for OS. Subsequently, graphic and online dynamic nomograms were established basing on the independent prognostic factors. We utilized C-index, calibration curve, receiver operating characteristic (ROC) curve, and area under ROC curve (AUC) value to evaluate the discrimination capacity and the consistency between predicted and actual survival. ResultsA total of 527 SGC patients were included (369 assigned to training group and 158 assigned to test group). The multivariable COX analysis showed that age, sex, marital status, tumor histology, summary stage, metastases to bone, and tumor size were independently associated with OS. Novel graphical and online dynamic (URL: https://yangxg1209.shinyapps.io/overall_survival_submandibular_gland_tumor/) nomograms were established. The C-indices (training: 0.77, 95%CI 0.71–0.84; test: 0.77, 95%CI 0.68–0.85) indicate favorable discrimination ability of the model, and the calibration curves demonstrated favorable consistency between the predicted and actual survival rates. ConclusionsOur study identified the independent prognostic factors influencing OS in patients with SGC, and successfully established and validated novel nomograms, which provide accurate prediction of survival rates and allows for personalized risk assessment.

Human circular RNA hsa_circ_0000231 clinical diagnostic effectiveness as a new tumor marker in gastric cancer.

Owing to the subtlety of initial symptoms associated with gastric cancer (GC), the majority of patients are diagnosed at later stages. Given the absence of reliable diagnostic markers, it is imperative to identify novel markers that exhibit high sensitivity and specificity. Circular RNA, a non-coding RNA, plays an important role in tumorigenesis and development and is well expressed in body fluids. In this study, we aimed to identify hsa_circ_0000231 as a new biomarker for the diagnosis of GC and to assess its clinical diagnostic value in serum. The stability and correctness of hsa_circ_0000231 was determined by agarose gel electrophoresis, Rnase R assay and Sanger sequencing. Real-time quantitative polymerase chain reaction (qRT-PCR) was designed to discover the expression level of hsa_circ_0000231 and whether it has dynamic serum monitoring capability. The correlation between hsa_circ_0000231 and clinicopathological parameters was analyzed by collecting clinical and pathological data from GC patients. In addition, diagnostic efficacy was assessed by constructing receiver operating characteristic curves (ROC). Hsa_circ_0000231 exhibits a stable and consistently expressed structure. In GC serum, cells, and tissues, it demonstrates reduced expression levels. Elevated expression levels observed postoperatively suggest its potential for dynamic monitoring. Additionally its expression level correlates with TNM staging and neuro/vascular differentiation. The area under ROC curve (AUC) for hsa_circ_0000231 is 0.781, indicating its superior diagnostic value compared to CEA, CA19-9, and CA72-4. The combination of these four indicators enhances diagnostic accuracy, with an AUC of 0.833. The stable expression of hsa_circ_0000231 in the serum of gastric cancer patients holds promise as a novel biomarker for both the diagnosis and dynamic monitoring of GC.

Impact of Pretransplant C-Reactive Protein, Neutrophiles, Platelets, and Albumin Levels on Recipient Survival After Simultaneous Pancreas and Kidney Transplantation

BackgroundLong-lasting diabetes mellitus type 1 and end-stage renal disease induce severe metabolic and immunological deterioration. Pretransplant C-reactive protein (CRP) and albumin (ALB) levels have impact on kidney transplantation. We evaluated impact of preoperative CRP, ALB, neutrophiles (NEU) and platelets (PLT) count, on one- and five-year recipients survival after simultaneous pancreas and kidney transplantation (SPK). MethodsAmong 103 SPK recipients, parameters: CRP (mean: 4.5±4.97mg/L); NEU (mean: 5.12 ±2.13∗103/mm3); PLT (mean: 244±84∗103/mm3); ALB (mean 4.5±0.75 g/dL) were obtained before transplantation. Cox regression, uni-, multivariate analysis for one- and five-year survivals were performed with 95% Confidence Intervals (CI) and area under ROC curve (AUC) was assessed. ResultsIn Cox regression, ALB <3,65g/dL had significant impact for one-year and five-year survivors with hazard ratios: 8 (CI:1.5-38.28, p<0.05) and 3.13 (CI:1.45-6.73, p<0.05), respectively. In univariate analysis, we found significantly decreased one-year survival if: PLT<180∗103/mm3, ALB <3.65g/dL, NEU>5.8∗103/mm3 and CRP>2.25mg/L with odds ratios (OR): 6.75 (CI: 2.12-21.15); 4.05(CI:1.3-12.09); 2.97 (CI: 1.02-8.64) and 5.51 (CI:1.67-18.19), respectively. Independent factors for five-year survival were: CRP, ALB and PLT with OR: 4.72(CI:1.67–13.29); 3.31(CI:1.18–9.25) and 4.2(CI:1.39–12.68), respectively. In multivariate analysis, we built two models for one-year survival. Model 1 (ALB+PLT) with OR: 3.12 (CI:0.97-10.07) and 5.55 (CI:1.67-18.4); and model 2 (CRP+PLT) with OR: 5.51 (CI:1.5-17.3) and 4.3 (CI:1.2-15.06), respectively. AUC for models 1 and 2 were: 0.74 and 0.759, respectively. ConclusionsNEU, PLT, ALB and CRP levels assessed before transplantation are independent factors for one- and five-year SPK recipients survival.

A Neoteric Feature Extraction Technique to Predict the Survival of Gastric Cancer Patients.

At the time of cancer diagnosis, it is crucial to accurately classify malignant gastric tumors and the possibility that patients will survive. This study aims to investigate the feasibility of identifying and applying a new feature extraction technique to predict the survival of gastric cancer patients. A retrospective dataset including the computed tomography (CT) images of 135 patients was assembled. Among them, 68 patients survived longer than three years. Several sets of radiomics features were extracted and were incorporated into a machine learning model, and their classification performance was characterized. To improve the classification performance, we further extracted another 27 texture and roughness parameters with 2484 superficial and spatial features to propose a new feature pool. This new feature set was added into the machine learning model and its performance was analyzed. To determine the best model for our experiment, Random Forest (RF) classifier, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Naïve Bayes (NB) (four of the most popular machine learning models) were utilized. The models were trained and tested using the five-fold cross-validation method. Using the area under ROC curve (AUC) as an evaluation index, the model that was generated using the new feature pool yields AUC = 0.98 ± 0.01, which was significantly higher than the models created using the traditional radiomics feature set (p < 0.04). RF classifier performed better than the other machine learning models. This study demonstrated that although radiomics features produced good classification performance, creating new feature sets significantly improved the model performance.

Prediction of neoplastic gallbladder polyps in patients with different age level based on preoperative ultrasound: a multi-center retrospective real-world study

BackgroundThe prevalence of neoplastic polyps in gallbladder polyps (GPs) increases sharply with age and is associated with gallbladder carcinoma (GBC). This study aims to predict neoplastic polyps and provide appropriate treatment strategies based on preoperative ultrasound features in patients with different age level.MethodsAccording to the age classification of WHO, 1523 patients with GPs who underwent cholecystectomy from January 2015 to December 2019 at 11 tertiary hospitals in China were divided into young adults group (n=622), middle-aged group (n=665) and elderly group (n=236). Linear scoring models were established based on independent risk variables screened by the Logistic regression model in different age groups. The area under ROC (AUC) to evaluate the predictive ability of linear scoring models, long- and short- diameter of GPs.ResultsIndependent risk factors for neoplastic polyps included the number of polyps, polyp size (long diameter), and fundus in the young adults and elderly groups, while the number of polyps, polyp size (long diameter), and polyp size (short diameter) in the middle-aged groups. In different age groups, the AUCs of its linear scoring model were higher than the AUCs of the long- and short- diameter of GPs for differentiating neoplastic and non-neoplastic polyps (all P<0.05), and Hosmer-Lemeshow goodness of fit test showed that the prediction accuracy of the linear scoring models was higher than the long- and short- diameter of GPs (all P>0.05).ConclusionThe linear scoring models of the young adults, middle-aged and elderly groups can effectively distinguish neoplastic polyps from non-neoplastic polyps based on preoperative ultrasound features.

Home-Monitoring Vision Tests to Detect Active Neovascular Age-Related Macular Degeneration

Most neovascular age-related macular degeneration (nAMD) treatments involve long-term follow-up of disease activity. Home-monitoring would reduce the burden on patients and their caregivers and release clinic capacity. To evaluate 3 vision home-monitoring tests for patients to use to detect active nAMD compared with diagnosing active nAMD at hospital follow-up during the after-treatment monitoring phase. This was a diagnostic test accuracy study wherein the reference standard was detection of active nAMD by an ophthalmologist at hospital follow-up. The 3 home-monitoring tests evaluated included the following: (1) the KeepSight Journal (KSJ [International Macular and Retinal Foundation]), which contains paper-based near-vision tests presented as word puzzles, (2) the MyVisionTrack (mVT [Genentech]) vision-monitoring mobile app, viewed on an Apple mobile operating system-based device, and (3) the MultiBit (MBT [Visumetrics]) app, viewed on an Apple mobile operating system-based device. Participants were asked to test weekly; mVT and MBT scores were transmitted automatically, and KSJ scores were returned to the research office every 6 months. Raw scores between hospital follow-ups were summarized as averages. Patients were recruited from 6 UK hospital eye clinics and were 50 years and older with at least 1 eye first treated for active nAMD for at least 6 months or longer to a maximum of 42 months before approach. Participants were stratified by time since starting treatment. Study data were analyzed from May to September 2021. The KSJ, mVT, and MBT were compared with the reference standard (in-hospital ophthalmologist examination). Estimated area under receiver operating characteristic curve (AUROC). The study had 90% power to detect a difference of 0.06, or 80% power to detect a difference of 0.05, if the AUROC for 2 tests was 0.75. A total of 297 patients (mean [SD] age, 74.9 [6.6] years; 174 female [58.6%]) were included in the study. At least 1 hospital follow-up was available for 312 study eyes in 259 participants (1549 complete visits). Median (IQR) home-monitoring testing frequency was 3 (1-4) times per month. Estimated AUROC was less than 0.6 for all home-monitoring tests, and only the KSJ summary score was associated with lesion activity (odds ratio, 3.48; 95% CI, 1.09-11.13; P = .04). Results suggest that no home-monitoring vision test evaluated provided satisfactory diagnostic accuracy to identify active nAMD diagnosed in hospital eye service follow-up clinics. Implementing any of these evaluated tests, with ophthalmologists only reviewing test positives, would mean most active lesions were missed, risking unnecessary sight loss.

Magnetic resonance elastography (MRE) outperforms acoustic force radiation impulse (ARFI) in predicting oesophageal varices in obese NAFLD cirrhosis.

Liver stiffness measurement (LSM) by transient elastography has been shown to underperform in high-risk varices (HRVs) prediction in obese non-alcoholic fatty liver disease (NAFLD) compensated cirrhosis (CC). LSM by magnetic resonance elastography (MRE) and acoustic force radiation impulse (ARFI) has been shown to be useful in prediction of oesophageal varices (EVs), but has limited evidence in obese NAFLD-CC. Obese patients with NAFLD-CC who underwent MRE and ARFI for LSM and endoscopy for screening of varices were enrolled. Performance of MRE and ARFI for predicting EVs or HRVs was evaluated using area under receiver operating characteristics (AUROC) curves and regression analyses were performed for predictor variables. One hundred eight patients [mean age 54.7 ± 9.6years, median BMI, 28.5 (26.4-30.0) kg/m2. 72.2% diabetics, 45.4% hypertensive] were enrolled. Fifty-two (48.1%) had no varices, while 29 (26.8%) and 27 (25%) had low-risk varices (LRVs) and HRVs, respectively. MRE-LSM was higher in patients with LRVs (p = 0.01) or HRVs (p = 0.001) against those without varices. ARFI-LSM did not differ significantly between those without and with LRVs or HRVs (p > 0.05 for all). There was a low correlation between ARFI-LSM and MRE-LSM in the overall cohort (r = 0.19). Only platelet count (PC) [0.98 (0.97-0.99)] and MRE-LSM [1.8 (1.26-2.79)] were predictors of HRVs. At a cut-off of 4.75, MRE showed a sensitivity of 96.3%. Model combining MRE-LSM with PC had a diagnostic AUROC of 0.77 and 0.76 for EVs and HRVs. In obese NAFLD-CC, MRE-LSM is significantly higher in patients with varices. MRE combined with PC predicts EVs and HRVs with better accuracy than ARFI.

Des-gamma-carboxy prothrombin and alpha-fetoprotein levels as biomarkers for hepatocellular carcinoma and their correlation with radiological characteristics.

Alpha-fetoprotein (AFP), a commonly used biomarker for hepatocellular carcinoma (HCC), is normal in up to one-third of patients. To evaluate the diagnostic performance of des-gamma-carboxy-prothrombin (DCP) alone and in combination with AFP. In this study, 202 patients with radiologically proven HCC were enrolled, and their DCP and AFP levels were evaluated for their diagnostic performance. The mean age of the enrolled patients was 58.5 years; 72.0% were male. DCP was elevated in 86.6% (n = 175) of all patients, 100.0% (n = 74) of patients with portal vein thrombus, and 87.4% (n = 111) of patients with multicentric HCC. AFP was elevated in 64.3% (n = 130) of all the patients, 74% (n = 55) of the patients with portal vein thrombus, and 71.6% (n = 91) of the patients with multicentric HCC (P = 0.030, 0.001, and 0.015, respectively). In tumors less than 2 cm in size (n = 46), DCP was increased in 32 (69.5%) patients, and AFP was increased in 25 (54.3%) patients (P = 0.801). There was good pairing between DCP and AFP for HCCs of 2 cm size or larger (P < 0.001); however, the pairing among tumors < 2 cm size was not significant (P = 0.210). In 69 of the patients (34.1%), only one of the tumor markers was positive; DCP was elevated alone in 57/202 (28.2%) of all patients, and AFP alone was elevated in 12/202 (5.9%) of the patients. The areas under receiver operating characteristic curves (AUROC) for tumors > 2 cm was 0.74 for DCP and 0.59 for AFP; combining both markers resulted in an AUROC of 0.73. For tumors < 2 cm, the AUROC was 0.25 for DCP and 0.40 for AFP. DCP, as an individual marker, had a better diagnostic performance in many cases of HCC. Hence, DCP may replace AFP as the primary HCC biomarker.

Evaluation of multiple autoantibodies to tumor-associated antigens as potential diagnostic markers for pancreatic cancer

The aim of the study is to evaluate the clinical significance of autoantibodies to GNA11, MSLN, GNAS, CEBPA, MDM2, p16, Sui1, Calnuc, PTEN in Pancreatic Cancer (PC). A total of 33 sera from PC patients and 45 sera from Normal Controls (NC) were enrolled to measure nine autoantibodies by Enzyme-Linked Immunosorbent Assay (ELISA). Multiple parameters in individual or combination forms were applied for evaluation. The level of all autoantibodies except anti-GNAS and anti-Calnuc was significantly higher in PC patients than that in normal controls. Significantly increased frequencies were found for four autoantibodies to MSLN, p16, PTEN and Sui1 at 75.8%, 66.7%, 30.3% and 27.3% compared to normal controls (4.9%). The ability of these autoantibodies to distinguish PC patients from normal controls reflected by the area under ROC curve (AUC) ranged from 0.666 to 0.884, with anti-p16 (0.884 of AUC) and anti-MSLN (0.857 of AUC) showing the strongest diagnostic performance. The combination of anti-MSLN and anti-p16 could improve the diagnostic sensitivity. In addition, two autoantibodies to MSLN and Sui1 in PC patients decreased after chemotherapy. Four autoantibodies to MSLN, p16, Sui1, PTEN showed potential diagnostic markers for pancreatic cancer, with anti-MSLN and anti-p16 in best performance. The combination of these two autoantibodies showed the most economical and practical potential. The significance of the reduction of autoantibodies to MSLN and Sui1 in PC patients who were subjected to chemotherapy needs to be further explored.

Development and internal validation of a dynamic fall risk prediction and monitoring tool in aged care using routinely collected electronic health data: a landmarking approach.

Falls pose a significant challenge in residential aged care facilities (RACFs). Existing falls prediction tools perform poorly and fail to capture evolving risk factors. We aimed to develop and internally validate dynamic fall risk prediction models and create point-based scoring systems for residents with and without dementia. A longitudinal cohort study using electronic data from 27 RACFs in Sydney, Australia. The study included 5492 permanent residents, with a 70%-30% split for training and validation. The outcome measure was the incidence of falls. We tracked residents for 60 months, using monthly landmarks with 1-month prediction windows. We employed landmarking dynamic prediction for model development, a time-dependent area under receiver operating characteristics curve (AUROCC) for model evaluations, and a regression coefficient approach to create point-based scoring systems. The model identified 15 independent predictors of falls in dementia and 12 in nondementia cohorts. Falls history was the key predictor of subsequent falls in both dementia (HR 4.75, 95% CI, 4.45-5.06) and nondementia cohorts (HR 4.20, 95% CI, 3.87-4.57). The AUROCC across landmarks ranged from 0.67 to 0.87 for dementia and from 0.66 to 0.86 for nondementia cohorts but generally remained between 0.75 and 0.85 in both cohorts. The total point risk score ranged from -2 to 57 for dementia and 0 to 52 for nondementia cohorts. Our novel risk prediction models and scoring systems provide timely person-centered information for continuous monitoring of fall risk in RACFs. Embedding these tools within electronic health records could facilitate the implementation of targeted proactive interventions to prevent falls.

Optimizing building material selection: A machine learning approach for efficient concrete compressive strength forecasting

Price changes in construction materials have a significant impact on building construction projects. Such price variations occur at random and at varying rates over time. A system that can estimate the magnitude and quantity of the change in material prices with reasonable accuracy is required. The primary goal is to create a machine-learning model that can predict the type of building material chosen based on environmental factors. The compressive strength of concrete is critical in defining its mechanical qualities. Long laboratory testing is needed to determine the compressive strength of concrete. The capacity of powerful machine learning algorithms to forecast concrete compressive strength speeds up these lengthy experimental methods while also lowering expenses. This study provides abilities to precisely anticipate and categorize numerous qualities and traits of distinct materials. The framework includes a broad dataset that details materials, composition, and performance characteristics. Machine learning algorithms such as logistic regression (LR), decision trees (DT), and random forests (RF) train models on the training data. The models are hyper-parameter tweaked and feature developed to achieve the most outstanding performance. The k-fold method is used throughout the training and assessment phase to guarantee robustness and reduce bias. The F1 score and Receiver Operating Characteristic-Area Under Curve (ROC-AUC) curve are two performance measures used to measure how accurate and predictive the trained models are. The study findings provide insights into the qualities of the materials, facilitating improved material selection, quality assurance, and decision-making in the building sector. In the analyses, the best accuracy value was 99.92%, and the precision value was 88.83% using the LR algorithm. As a result, it was determined that the LR algorithm had the least execution 57.826 ms, and is thus the most suitable for use in concrete compressive strength estimation.

A new predictive scoring model for globus pharyngeus in patients with gastroesophageal reflux disease.

Globus pharyngeus is a clinical condition, wherein, a patient senses a lump or a foreign body in the throat with a tightening or choking feeling. A strong association between globus pharyngeus and gastroesophageal reflux disease (GERD) was reported. Therefore, we sought to investigate the predictive factors of globus pharyngeus in patients with established GERD and fit a predictive scoring model for globus pharyngeus. In this case-control study, 143 patients having globus pharyngeus along with GERD ( case ) and 109 patients having globus pharyngeus without GERD ( control ) were enrolled. Data comprising demographics, comorbidities, and psychosocial stress levels were recorded. The predictive factors of globus pharyngeus in patients with GERD were unraveled, and a predictive scoring model was fit for globus pharyngeus. Proton pump inhibitor usage in the case group was significantly higher compared to controls (63.63% vs 24.78%, P < 0.001), and differences in Hiatus Hernia and Stress levels were highly significant between the two groups ( P < 0.001). Multivariate logistic regression revealed that variates, Hiatus Hernia, psychosocial stress, and age were highly significant ( P < 0.001) independent predictors of globus pharyngeus. Using the regression coefficients of all the independent predictor variables, a predictive scoring model was fitted, which yielded an area under receiver operating characteristic (AUROC) curve of 78.9. Hiatus hernia, psychosocial stress, and age are significant independent predictors of globus pharyngeus in GERD, and our predictive scoring model may help in identifying patients at higher odds of experiencing globus pharyngeus and modulate treatment accordingly.

Role of radiomics in staging liver fibrosis: a meta-analysis

BackgroundFibrosis has important pathoetiological and prognostic roles in chronic liver disease. This study evaluates the role of radiomics in staging liver fibrosis.MethodAfter literature search in electronic databases (Embase, Ovid, Science Direct, Springer, and Web of Science), studies were selected by following precise eligibility criteria. The quality of included studies was assessed, and meta-analyses were performed to achieve pooled estimates of area under receiver-operator curve (AUROC), accuracy, sensitivity, and specificity of radiomics in staging liver fibrosis compared to histopathology.ResultsFifteen studies (3718 patients; age 47 years [95% confidence interval (CI): 42, 53]; 69% [95% CI: 65, 73] males) were included. AUROC values of radiomics for detecting significant fibrosis (F2-4), advanced fibrosis (F3-4), and cirrhosis (F4) were 0.91 [95%CI: 0.89, 0.94], 0.92 [95%CI: 0.90, 0.95], and 0.94 [95%CI: 0.93, 0.96] in training cohorts and 0.89 [95%CI: 0.83, 0.91], 0.89 [95%CI: 0.83, 0.94], and 0.93 [95%CI: 0.91, 0.95] in validation cohorts, respectively. For diagnosing significant fibrosis, advanced fibrosis, and cirrhosis the sensitivity of radiomics was 84.0% [95%CI: 76.1, 91.9], 86.9% [95%CI: 76.8, 97.0], and 92.7% [95%CI: 89.7, 95.7] in training cohorts, and 75.6% [95%CI: 67.7, 83.5], 80.0% [95%CI: 70.7, 89.3], and 92.0% [95%CI: 87.8, 96.1] in validation cohorts, respectively. Respective specificity was 88.6% [95% CI: 83.0, 94.2], 88.4% [95% CI: 81.9, 94.8], and 91.1% [95% CI: 86.8, 95.5] in training cohorts, and 86.8% [95% CI: 83.3, 90.3], 94.0% [95% CI: 89.5, 98.4], and 88.3% [95% CI: 84.4, 92.2] in validation cohorts. Limitations included use of several methods for feature selection and classification, less availability of studies evaluating a particular radiological modality, lack of a direct comparison between radiology and radiomics, and lack of external validation.ConclusionAlthough radiomics offers good diagnostic accuracy in detecting liver fibrosis, its role in clinical practice is not as clear at present due to comparability and validation constraints.

Development and Validation of Two-Step Prediction Models for Postoperative Bedridden Status in Geriatric Intertrochanteric Hip Fractures.

Patients with intertrochanteric hip fractures are at an elevated risk of becoming bedridden compared with those with intraarticular hip fractures. Accurate risk assessments can help clinicians select postoperative rehabilitation strategies to mitigate the risk of bedridden status. This study aimed to develop a two-step prediction model to predict bedridden status at 3 months postoperatively: one model (first step) for prediction at the time of admission to help dictate postoperative rehabilitation plans; and another (second step) for prediction at the time before discharge to determine appropriate discharge destinations and home rehabilitation programs. Three-hundred and eighty-four patients were retrospectively reviewed and divided into a development group (n = 291) and external validation group (n = 93). We developed a two-step prediction model to predict the three-month bedridden status of patients with intertrochanteric fractures from the development group. The first (preoperative) model incorporated four simple predictors: age, dementia, American Society of Anesthesiologists physical status classification (ASA), and pre-fracture ambulatory status. The second (predischarge) model used an additional predictor, ambulation status before discharge. Model performances were evaluated using the external validation group. The preoperative model performances were area under ROC curve (AUC) = 0.72 (95%CI 0.61-0.83) and calibration slope = 1.22 (0.40-2.23). The predischarge model performances were AUC = 0.83 (0.74-0.92) and calibration slope = 0.89 (0.51-1.35). A decision curve analysis (DCA) showed a positive net benefit across a threshold probability between 10% and 35%, with a higher positive net benefit for the predischarge model. Our prediction models demonstrated good discrimination, calibration, and net benefit gains. Using readily available predictors for prognostic prediction can assist clinicians in planning individualized postoperative rehabilitation programs, home-based rehabilitation programs, and determining appropriate discharge destinations, especially in environments with limited resources.

Profiling effects of filtering noise labels on learning performance

Deciding whether a label in a dataset is noisy or not depends on factors such as filtering levels, data patterns, relabeling policies, and pre-defined regulations. By focusing on filtering acceptance levels (FAL), a decision support model (named DSM-ENL) is proposed and implemented on real datasets that are granulated, dealing with noise label detection and profiling effects of 50 FALs (51 %–100 %) on noise rate (NoR), classification accuracy (CA), and area under ROC curve (AUC). A case used to demonstrate the DSM-ENL is illustrated, followed by experiments on 27 unsupervised and 29 supervised granulated datasets using five classifiers with 70 %–30 % training and testing criteria. The findings revealed that: (1) the best CAs among classifiers vary with FALs; (2) NoR, CA, and AUC oscillate as FAL increases; (3) lower FAL boundaries (from 0.51 to 0.60) are more likely to lead to higher CAs; and (4) correlation of coefficient among dataset characteristics, FAL, NoR, CA, and AUC are not high compared to that of CA and NoR and that of CA and AUC. The research demonstrates the value of profiling the effects of FALs applied for noise-label datasets on learning performance.

RENAL DYSFUNCTION IN PRETERM INFANTS WITH PERINATAL PATHOLOGY: RISK FACTORS, SENSITIVITY AND SPECIFICITY OF LABORATORY MARKERS OF DAMAGE

RENAL DYSFUNCTION IN PRETERM INFANTS WITH PERINATAL PATHOLOGY: RISK FACTORS, SENSITIVITY AND SPECIFICITY OF LABORATORY MARKERS OF DAMAGE

Does radiofrequency ablation procedural data improve the accuracy of identifying atrial fibrillation recurrence?

Radiofrequency ablation (RFA) using the CARTO 3D mapping system is a common approach for pulmonary vein isolation to treat atrial fibrillation (AF). Linkage between CARTO procedural data and patients' electronical health records (EHR) provides an opportunity to identify the ablation-related parameters that would predict AF recurrence. The objective of this study is to assess the incremental accuracy of RFA procedural data to predict post-ablation AF recurrence using machine learning model. Procedural data generated during RFA procedure were downloaded from CARTONET and linked to deidentified Mercy Health EHR data. Data were divided into train (70%) and test (30%) data for model development and validation. Automate machine learning (AutoML) was used to predict 1 year AF recurrence, defined as a composite of repeat ablation, electrical cardioversion, and AF hospitalization. At first, AutoML model only included Patients' demographic and clinical characteristics. Second, an AutoML model with procedural variables and demographical/clinical variables was developed. Area under receiver operating characteristic curve (AUROC) and net reclassification improvement (NRI) were used to compare model performances using test data. Among 306 patients, 67 (21.9%) patients experienced 1-year AF recurrence. AUROC increased from 0.66 to 0.78 after adding procedural data in the AutoML model based on test data. For patients with AF recurrence, NRI was 32% for model with procedural data. Nine of 10 important predictive features were CARTO procedural data. From CARTO procedural data, patients with lower contact force in right inferior site, long ablation duration, and low number of left inferior and right roof lesions had a higher risk of AF recurrence. Patients with persistent AF were more likely to have AF recurrence. The machine learning model with procedural data better predicted 1-year AF recurrence than the model without procedural data. The model could be used for identification of patients with high risk of AF recurrence post ablation.

Skin mottling score assesses peripheral tissue hypoperfusion in critically ill patients following cardiac surgery

BackgroundSkin mottling is a common manifestation of peripheral tissue hypoperfusion, and its severity can be described using the skin mottling score (SMS). This study aims to evaluate the value of the SMS in detecting peripheral tissue hypoperfusion in critically ill patients following cardiac surgery.MethodsCritically ill patients following cardiac surgery with risk factors for tissue hypoperfusion were enrolled (n = 373). Among these overall patients, we further defined a hypotension population (n = 178) and a shock population (n = 51). Hemodynamic and perfusion parameters were recorded. The primary outcome was peripheral hypoperfusion, defined as significant prolonged capillary refill time (CRT, > 3.0 s). The characteristics and hospital mortality of patients with and without skin mottling were compared. The area under receiver operating characteristic curves (AUROC) were used to assess the accuracy of SMS in detecting peripheral hypoperfusion. Besides, the relationships between SMS and conventional hemodynamic and perfusion parameters were investigated, and the factors most associated with the presence of skin mottling were identified.ResultsOf the 373-case overall population, 13 (3.5%) patients exhibited skin mottling, with SMS ranging from 1 to 5 (5, 1, 2, 2, and 3 cases, respectively). Patients with mottling had lower mean arterial pressure, higher vasopressor dose, less urine output (UO), higher CRT, lactate levels and hospital mortality (84.6% vs. 12.2%, p < 0.001). The occurrences of skin mottling were higher in hypotension population and shock population, reaching 5.6% and 15.7%, respectively. The AUROC for SMS to identify peripheral hypoperfusion was 0.64, 0.68, and 0.81 in the overall, hypotension, and shock populations, respectively. The optimal SMS threshold was 1, which corresponded to specificities of 98, 97 and 91 and sensitivities of 29, 38 and 67 in the three populations (overall, hypotension and shock). The correlation of UO, lactate, CRT and vasopressor dose with SMS was significant, among them, UO and CRT were identified as two major factors associated with the presence of skin mottling.ConclusionIn critically ill patients following cardiac surgery, SMS is a very specific yet less sensitive parameter for detecting peripheral tissue hypoperfusion.