Area Under The ROC Curve Research Articles

Development and validation of a screening tool for sepsis without laboratory results in the emergency department: amachine learning study.

Sepsis is a significant health burden on a global scale. Timely identification and treatment of sepsis can greatly improve patient outcomes, including survival rates. However, time-consuming laboratory results are often needed for screening sepsis. We aimed to develop a quick sepsis screening tool (qSepsis) based on patients' non-laboratory clinical data at the emergency department (ED) using machine learning (ML), and compare its performance with established clinical scores. This retrospective study included patients admitted to the ED of Zhongnan Hospital of Wuhan University (Wuhan, China) from 1/1/2015 to 5/31/2022. Patients who were under 18 years of age, had cardiopulmonary arrest upon arrival at the ED, or had missing and abnormal medical record data were excluded. The qSepsis was derived by three ML algorithms, including logistic regression (LR), random forest (RF), and extreme gradient boosting (XGB). To benchmark the existing clinical tools for assessing the risk of sepsis in the ED, qSepsis was compared with the Systemic Inflammatory Response Syndrome (SIRS), the Quick Sepsis-Related Organ Failure Assessment (qSOFA), and the Modified Early Warning Score (MEWS). The external validation was performed with the Medical Information Mart for Intensive Care IV ED database (United States), and adopted the same inclusion and exclusion criteria. The predictive power of qSepsis and other clinical scores was measured using the area under the receiver operating characteristic curve (AUROC). The primary outcome of the study was the diagnosis of sepsis in the ED based on the Sepsis 3.0 criteria, which served as the basis for developing the qSepsis tool. A total of 414,864 patients were finally included in the cohort (median ([IQR]) patient age, 43 (29, 60) years; 202,730 (48.87%) females, 212,134 (51.13%) males), and 200,089 in the external testing cohort (median (SD) patient age, 57 (39, 71) years; 107,427 (53.69%) females, 92,663 (46.31%) males). For internal testing, LR outperformed RF and XGB with an AUROC of 0.862 (95% CI, 0.855-0.869). In external testing, the AUROC decreased to 0.766 (95% CI, 0.758-0.774) for LR, 0.725 (95% CI, 0.717-0.733) for RF, and 0.735 (95% CI, 0.728-0.742) for XGB. In addition, the AUROC for the qSOFA, MEWS, and SIRS scores in external validation cohort were 0.579 (95% CI, 0.563-0.596), 0.600 (95% CI, 0.578-0.622), and 0.704 (95% CI, 0.683-0.725), respectively. The area under the precision-recall curve (AUPRC) for the qSepsis model was 0.213 (95% CI: 0.204-0.222). The AUPRC values for the other scores were as follows: SIRS, 0.071 (95% CI: 0.013-0.099); qSOFA, 0.096 (95% CI: 0.003-0.186); and MEWS, 0.083 (95% CI: 0.063-0.111). This retrospective study demonstrated that qSepsis had better predictive performance in terms of AUROC and area under the precision-recall curve (AUPRC) compared to existing assessment scores. It has the potential to be used in pre-hospital settings with limited access to laboratory tests and in the ED for quick screening of patients with sepsis. However, due to its low positive predictive value (PPV), the false alarms may increase in actual clinical practice. Transformation of Scientific and Technological Achievements Fund Project of Zhongnan Hospital of Wuhan University.

Development and validation of machine learning model for mortality after curative gastrectomy for gastric cancer.

386 Background: The risk for postoperative complications can significantly impact perioperative planning. The aim of this study is to develop and validate a model based on various machine learning techniques to predict postoperative mortality (PM) after curative radical gastrectomy for gastric cancer. Methods: Data from a nationwide survey led by the Korean Gastric Cancer Association was used for the development set, which included 12,722 patients who underwent gastrectomy in 2019 across 68 institutions in South Korea. The validation set comprised 4,887 patients from 8 institutions between 2009 and 2017. PM was defined as all-cause death within 30 days after surgery or in-hospital death beyond 30 days. Five machine learning models (CatBoost, Gradient Boosting, Light GBM, Random Forest, XGBoost) were evaluated using perioperative clinical characteristics. Model performance was assessed using the area under the receiver operation characteristic curve (AUROC) with 5-fold cross-validation. Results: The PM rates were 0.83% and 1.04% in the development and validation set, respectively. In the development set, the Random Forest model demonstrated the highest discrimination capacity with an AUROC of 0.772, followed by Gradient Boosting at 0.767, CatBoost at 0.742, XGBoost at 0.732, and LightGBM at 0.717. However, in the validation set, CaBoost showed the highest AUROC at 0.747, while the AUROC for Random Forest decreased to 0.688. Conclusions: The CatBoost model demonstrated robust and consistent performance in predicting PM risk across both development and validation datasets. To further enhance the performance of the model and ensure stable clinical application, future research should focus on developing methods for quantitative measurement of perioperative factors such as surgical techniques.

Prediction of reduced left ventricular ejection fraction using atrial fibrillation or flutter electrocardiograms: A machine-learning study.

Although the evaluation of left ventricular ejection fraction (LVEF) in patients with atrial fibrillation (AF) or atrial flutter (AFL) is crucial for appropriate medical management, the prediction of reduced LVEF (<50%) with AF/AFL electrocardiograms (ECGs) lacks evidence. This study aimed to investigate deep-learning approaches to predict reduced LVEF (<50%) in patients with AF/AFL ECGs and easily obtainable clinical information. Patients with 12-lead ECGs of AF/AFL and echocardiography were divided into those with LVEF <50% and ≥50%. A convolutional neural networks-based model customized to the study (AFibEFNet) and other deep-learning models were investigated. Electrocardiogram signals, ECG features, and clinical features (demographic information, comorbidities, blood cell counts, and blood test results) were collected for training. A hold-out test dataset was constructed using a different recruitment period. Five-fold cross-validation and calibration plots were used to evaluate performance. A total of 15,683 patients were analyzed (mean age, 70.0 ± 11.7 years; 61.2% men), with 82.2% having LVEF ≥50% and 17.8% having LVEF < 50%. Among the learning models, the AFibEFNet outperformed other models regarding area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), and F1-score. Using ECG signals alone, the AFibEFNet model predicted reduced LVEF with AUROC of 0.798 (95% confidence interval [CI], 0.767-0.829) and AUPRC of 0.508 (95% CI, 0.434-0.564). For the AFibEFNet model, additional training with ECG and clinical features significantly improved AUROC (0.816 vs. 0.798, p = 0.04) and AUPRC (0.547 vs. 0.508, p < 0.001). The AFibEFNet model primarily focused on the R-wave, QRS onset and offset, and T-wave in ECG signals. Among the patients with AF/AFL, machine learning may predict reduced LVEF with 12-lead ECGs of AF/AFL.

Assessment of sleep quality using cardiopulmonary coupling and its predictive value for delirium in ICU patients.

To assess sleep quality in intensive care unit (ICU) patients using cardiopulmonary coupling (CPC) analysis and explore its predictive value for delirium. ICU patients (n=135) were divided into the delirium group (n=44) and control group (n=91) based on the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU). CPC analysis was used to evaluate the sleep quality of all participants. Intergroup comparisons showed that total sleep time, stable sleep time, and sleep efficiency were significantly lower in the delirium group than in the control group, whereas unstable sleep time, rapid eye movement sleep time, wake-up time, stable sleep latency, and apnea-hypopnea index (AHI) were significantly higher in the delirium group than in the control group. Logistic regression analysis showed that advanced age and AHI were risk factors for delirium, whereas stable sleep time was a protective factor for delirium. Receiver operating characteristic (ROC) analysis showed that stable sleep time and stable sleep latency had a certain predictive value for delirium; the area under the ROC curve (AUC) for stable sleep time was higher (0.888 vs. 0.704). The cut-off value for stable sleep time was 0.65h, with a sensitivity of 84.1% and specificity of 81.3%. ICU patients with delirium have poorer sleep quality than patients without delirium. Stable sleep time derived from the CPC has a high predictive value for delirium and may serve as an objective indicator for its diagnosis.

Assessing the Predictive Value of the Neutrophil-to-Lymphocyte Ratio for Post-Thrombotic Syndrome following Iliofemoral Deep Venous Thrombosis

Post-thrombotic syndrome (PTS) is a common complication of deep vein thrombosis (DVT) that occurs in 20-50% of patients and results in a decreased quality of life. Even with the progressive identification of PTS risk factors, clinically useful predictors of PTS continue to be limited, unobjective, and ill-defined. The neutrophil-to-lymphocyte ratio (NLR) is an emerging prognostic biomarker used in a variety of diseases that reflects acute systemic inflammation. This pilot study aimed to evaluate the utility of the NLR at the time of iliofemoral DVT diagnosis in predicting PTS incidence in patients. A retrospective chart review was performed on patients identified with iliofemoral DVT at the University of Maryland Medical Center between 2020 and 2022. Patients with at least one follow-up visit between 3 to 6 months after initial DVT diagnosis were included. Diagnosis of PTS was determined based on Villalta Score. The Youden index with receiver operating characteristic (ROC) curve analysis was used to determine the NLR cutoff value that was predictive of PTS. A multivariable logistic regression model was then performed to assess the utility of this NLR cutoff value and other common clinical markers in predicting the presence of PTS symptoms. 418 patients with positive iliofemoral DVT venous duplex ultrasounds were screened for eligibility. 118 patients were eligible with a mean age of 53.18 ± 15.45 years. A total of 43 patients (36.44%) were found to have PTS. A NLR cutoff of 7.71 was determined with an area under the ROC curve (AUC) of 0.63 (p=0.046). When the NLR was assessed jointly with other clinical markers at the time of DVT diagnosis, NLR was a statistically significant positive predictor, measured using odds ratio (OR, 1.83; 95% CI, 1.20-2.78; p=0.005). Our study found that when stratified by a determined cutoff value, the NLR at the time of DVT diagnosis was significantly associated with the development of PTS in patients with iliofemoral DVT. This result is consistent with one prior research finding yet is novel in its specificity for iliofemoral DVTs and its acute lab collection for NLR calculation. The NLR should be further investigated as a potential inexpensive prognostic tool to aid in the improvement of treatment and prophylactic strategies for PTS.

Development and Validation of a Novel Model to Discriminate Idiosyncratic Drug-Induced Liver Injury and Autoimmune Hepatitis.

Discriminating between idiosyncratic drug-induced liver injury (DILI) and autoimmune hepatitis (AIH) is critical yet challenging. We aim to develop and validate a machine learning (ML)-based model to aid in this differentiation. This multicenter cohort study utilised a development set from Beijing Friendship Hospital, with retrospective and prospective validation sets from 10 tertiary hospitals across various regions of China spanning January 2009 to May 2023. Different ML algorithms were tested using 24 routine laboratory parameters. The Shapley Additive exPlanations (SHAP) analysis was used to evaluate the contribution of each parameter in the ML model. A total of 2554 patients (1750 for DILI and 804 for AIH) were included. Using Gradient Boost Decision Tree algorithm, five key parameters-aspartate transaminase, globulin, prealbumin, creatinine and platelet count-were selected to construct the ML model. Consequently, a web-based tool named Beijing-AID (BJ-AID) was developed (http://43.143.153.225:5000/). The BJ-AID model demonstrated excellent discrimination performance, with an area under the receiver operating characteristic curve (AUROC) of 0.94 (95% CI, 0.902-0.975) in the development set, 0.91 (95% CI, 0.900-0.928) in all external validation sets and 0.93 (95% CI, 0.889-0.974) in a prospective validation set. Notably, the BJ-AID model also effectively discriminated atypical cases, including drug-induced autoimmune-like hepatitis and AIH with the history of drug consumption, achieving an AUROC = 0.85 (95% CI, 0.742-0.949). We successfully developed and validated a machine learning-based model, BJ-AID, which exhibits a strong discrimination performance. BJ-AID can assist practitioners and hepatologists in diagnosing both typical and atypical cases of DILI and AIH. ClinicalTrials.gov identifier: NCT05532345.

Performance of artificial intelligence for diagnosing cervical intraepithelial neoplasia and cervical cancer: a systematic review and meta-analysis

Cervical cytology screening and colposcopy play crucial roles in cervical intraepithelial neoplasia (CIN) and cervical cancer prevention. Previous studies have provided evidence that artificial intelligence (AI) has remarkable diagnostic accuracy in these procedures. With this systematic review and meta-analysis, we aimed to examine the pooled accuracy, sensitivity, and specificity of AI-assisted cervical cytology screening and colposcopy for cervical intraepithelial neoplasia and cervical cancer screening. In this systematic review and meta-analysis, we searched the PubMed, Embase, and Cochrane Library databases for studies published between January 1, 1986 and August 31, 2024. Studies investigating the sensitivity and specificity of AI-assisted cervical cytology screening and colposcopy for histologically verified cervical intraepithelial neoplasia and cervical cancer and a minimum of five cases were included. The performance of AI and experienced colposcopists was assessed via the area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV) through random effect models. Additionally, subgroup analyses of multiple diagnostic performance metrics in developed and developing countries were conducted. This study was registered with PROSPERO (CRD42024534049). Seventy-seven studies met the eligibility criteria for inclusion in this study. The pooled diagnostic parameters of AI-assisted cervical cytology via Papanicolaou (Pap) smears were as follows: accuracy, 94% (95% CI 92-96); sensitivity, 95% (95% CI 91-98); specificity, 94% (95% CI 89-97); PPV, 88% (95% CI 78-96); and NPV, 95% (95% CI 89-99). The pooled accuracy, sensitivity, specificity, PPV, and NPV of AI-assisted cervical cytology via ThinPrep cytologic test (TCT) were 90% (95% CI 85-94), 97% (95% CI 95-99), 94% (95% CI 85-98), 84% (95% CI 64-98), and 96% (95% CI 94-98), respectively. Subgroup analysis revealed that, for AI-assisted cervical cytology diagnosis, certain performance indicators were superior in developed countries compared to developing countries. Compared with experienced colposcopists, AI demonstrated superior accuracy in colposcopic examinations (odds ratio (OR) 1.75; 95% CI 1.33-2.31; P<0.0001; I2=93%). These results underscore the potential and practical value of AI in preventing and enabling early diagnosis of cervical cancer. Further research should support the development of AI for cervical cancer screening, including in low- and middle-income countries with limited resources. This study was supported by the National Natural Science Foundation of China (No. 81901493) and the Shanghai Pujiang Program (No. 21PJD006).

Value of multi-parameter 123I-MIBG scintigraphy in the differential diagnosis of Parkinson’s disease

Background123I-MIBG scintigraphy plays a significant role in diagnosing Parkinson's disease (PD), with most studies primarily targeting cardiac uptake and relying on traditional ratio-based parameters for assessment. However, due to variations in scanning conditions and image processing methodologies, the clinical utility of different parameters remains a subject of debate. This study aims to evaluate the diagnostic accuracy of multi-parameter 123I-3-Iodobenzylguanidine (MIBG) scintigraphy and to identify the most reliable metrics for distinguishing PD from Parkinson-plus syndromes.ResultWe recruited 56 PD patients and 44 non-PD controls, who underwent 123I-MIBG scintigraphy and clinical evaluation. MIBG uptake and washout rates (WR) for the heart, salivary glands, and thyroid were assessed, with semi-quantitative methods used to calculate heart-to-mediastinum (H/M) and gland-to-background (P/B, S/B, T/B) ratios. Diagnostic efficacy was evaluated using receiver operating characteristic (ROC) curves. The H/M ratio and cardiac WR parameters scintigraphy exhibited excellent diagnostic efficacy for PD and PPS. The 4 h H/M ratio showed superior diagnostic performance with an area under the ROC curve (AUC) of 0.980. Background and time decay correction improved cardiac WR diagnostic performance (AUC = 0.963). Although P/B, S/B, and T/B ratios showed no significant differences for differential diagnosis, the WR values of the parotid and thyroid glands exhibited moderate diagnostic efficacy with AUCs of 0.648 and 0.638, respectively.ConclusionThe heart H/M ratio and cardiac WR in 123I-MIBG scintigraphy show high diagnostic efficacy for both diagnosing and differentiating PD. Including cardiac WR in routine assessments is recommended. The diagnostic potential of WR in the parotid and thyroid glands also holds promise for PD diagnosis. Incorporating these parameters could enhance the accuracy of PD diagnosis, particularly in clinical scenarios where concurrent cardiac disease may confound the diagnosis.

Development and validation of a risk prediction model for early postoperative high intraocular pressure after implantation of posterior chamber intraocular lens

Objective: To develop and validate a predictive model for assessing the risk of early postoperative high intraocular pressure (HIOP) following posterior chamber intraocular lens implantation. Methods: The clinical data of patients who underwent posterior chamber intraocular lens implantation at the Second Affiliated Hospital of Zhejiang University School of Medicine between May 2023 and April 2024 were retrospectively reviewed. Patients were divided into a modeling group and a validation group with a 7∶3 ratio using computerized random allocation. The modeling group was further categorized into normal intraocular pressure and HIOP groups based on the occurrence of early postoperative HIOP. Multivariate logistic regression analysis was employed to establish a risk prediction model. Receiver operating characteristic (ROC) curve analysis was conducted using R software to assess the predictive ability and accuracy of the model, along with calibration and decision curve analysis (DCA) curve plotting. Results: A total of 239 patients (478 eyes) were enrolled. The modeling group consisted of 168 patients (82 males and 86 females) (336 eyes), with an average age of (25.5±5.1) years. The validation group consisted of 71 patients (36 males and 35 females) (142 eyes), with an average age of (24.7±5.7) years. The early postoperative HIOP incidence rate after posterior chamber intraocular lens implantation was 32.4% (109/336), while it was 26.8% (38/142) in the validation group. The multivariate logistic regression analysis indicated that preoperative refractive diopter (OR=0.873, 95%CI: 0.782-0.976, P=0.017), corneal endothelial cell count (OR=0.996, 95%CI: 0.995-0.997, P<0.001), corneal thickness (OR=1.015, 95%CI: 1.006-1.024, P=0.001), and preoperative intraocular pressure (OR=1.470, 95%CI: 1.283-1.686, P<0.001) were associated factors for early HIOP. The Hosmer-Lemeshow test showed that the χ2 value was 8.444 (P=0.391), and the area under the ROC curve (AUC) was 0.888 (95%CI: 0.852-0.925). The Youden index reached its maximum value of 0.646, with a sensitivity of 0.862 and specificity of 0.784. AUC of validation group was 0.921 (95%CI: 0.870-0.972), with a maximum Youden index of 0.765, sensitivity of 0.842 and specificity of 0.923. Conclusions: The risk prediction model constructed based on preoperative refractive diopter, corneal endothelial cell count, corneal thickness, and preoperative intraocular pressure demonstrates good predictive performance for early postoperative HIOP following posterior chamber intraocular lens implantation.

Predictive model for ejection fraction improvement at one year in patients with acute ST-segment elevation myocardial infarction complicated with heart failure with reduced ejection fraction

Objective: To develop a predictive model for improvement of ejection fraction 1 year after heart failure with reduced ejection fraction (HFrEF) following acute ST-segment elevation myocardial infarction (STEMI). Methods: This nested case-control study included STEMI patients diagnosed with HFrEF from a prospective multicenter multimodality imaging cohort between August 2014 and March 2021. Based on the improvement of left ventricular ejection fraction (LVEF) at baseline and 1-year follow-up, the patients were classified into the heart failure with improved ejection fraction (HFimpEF) group and the persistent HFrEF group. The clinical data were collected, and cardiac histological changes were assessed using cardiac magnetic resonance imaging. Multivariate logistic regression analysis was performed to identify factors associated with ejection fraction improvement at one year, and a predictive model was developed and internally validated. The performance and clinical applicability of the model were evaluated using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis. Results: A total of 117 STEMI patients (102 males and 15 females) aged (61.6±11.6) years were included in the study. At the 1-year follow-up, there were 61 patients (52.1%)in the HFimpEF group,and 56 patients (47.9%) in the HFrEF group . Compared with persistent HFrEF group, patients in HFimpEF group had smaller baseline left ventricular end-systolic diameter (LVESD) [33.0 (30.0, 36.0) mm vs 35.5 (32.0, 39.0) mm], smaller infarct size [26.1% (20.3%, 36.0)% vs 40.6% (33.0%, 45.4)%], lower peak B-type natriuretic peptide (BNP) level [340.0 (190.5, 692.5) ng/L vs 636.0 (318.5, 1 188.8) ng/L], lower peak level of soluble suppression of tumorigenicity 2 (sST2) [36.7 (25.8, 60.5) μg/L vs 62.4 (30.6, 120.7) μg/L], and higher hematocrit [(43.5%±3.5%) vs (40.8%±5.6%)] (all P<0.05). Multivariate logistic regression analysis revealed that smaller baseline LVESD (OR=0.825, 95%CI: 0.745-0.914), smaller infarct size (OR=0.967, 95%CI: 0.939-0.995), peak BNP level≤400 ng/L (OR=3.062, 95%CI: 1.283-7.306), peak sST2 level≤35 μg/L (OR=2.600, 95%CI: 1.040-6.501), and higher hematocrit (OR=1.109, 95%CI: 1.030-1.193) were predictors of LVEF improvement in STEMI patients with HFrEF. The predictive model formula: logit (P)=2.619-0.034×infarcted myocardium percentage (%)+1.119×(peak BNP level≤400 ng/L)+0.956×(peak sST2 level≤35 μg/L)+0.103×hematocrit (%)-0.192×LVESC (mm) (where peak BNP level≤400 ng/L and peak sST2 level≤35 μg/L are binary variables: Yes=1, No=0). The area under the ROC curve (AUC) was 0.805 (95%CI: 0.723-0.887), indicating good predictive ability. Calibration curves and decision curve analysis indicated good model consistency and clinical utility. Conclusions: Smaller LVESD, smaller infarct size, peak BNP level≤400 ng/L, peak sST2 level≤35 μg/L and higher hematocrit are predictive factors for LVEF improvement after STEMI. The predictive model has good performance for predicting HFimpEF.

Development and validation of an EHR-based risk prediction model for geriatric patients undergoing urgent and emergency surgery

BackgroundClinical determination of patients at high risk of poor surgical outcomes is complex and may be supported by clinical tools to summarize the patient’s own personalized electronic health record (EHR) history and vitals data through predictive risk models. Since prior models were not readily available for EHR-integration, our objective was to develop and validate a risk stratification tool, named the Assessment of Geriatric Emergency Surgery (AGES) score, predicting risk of 30-day major postoperative complications in geriatric patients under consideration for urgent and emergency surgery using pre-surgical existing electronic health record (EHR) data.MethodsPatients 65-years and older undergoing urgent or emergency non-cardiac surgery within 21 hospitals 2017–2021 were used to develop the model (randomly split: 80% training, 20% test). The primary outcome was a 30-day composite outcome including several postoperative major complications and mortality; secondary outcomes included common individual complications included in the primary composite outcome (sepsis, progressive renal insufficiency or renal failure, and mortality). Patients’ EHR-based clinical history, vital signs, labs, and demographics were included in logistic regression, LASSO, decision tree, Random Forest, and XGBoost models. Area under the receiver operating characteristics curve (AUCROC) was used to compare model performance.ResultsOverall, 66,262 patients (median [IQR] age 78 [(70.9–84.0], female 53.9%, White race 68.5%) received urgent or emergency non-cardiac surgery (25.7% orthopedic cases, 21.9% general surgery cases). AUCROC ranged from 0.655 (Decision Tree) – 0.804 (XGBoost) for the primary composite outcome. XGBoost AUCROC was 0.823, 0.781, and 0.839 in predicting outcomes of sepsis, progressive renal insufficiency or renal failure, and mortality, respectively.ConclusionsWe developed a model to accurately predict major postoperative complications in geriatric patients undergoing urgent or emergency surgery using the patient’s own existing EHR data. EHR implementation of this model could efficiently support clinicians’ surgical risk assessment and perioperative decision-making discussions in this vulnerable patient population.

Computed tomography enterography-based radiomics nomograms to predict inflammatory activity for ileocolonic Crohn’s disease: a preliminary single-center retrospective study

ObjectivesThis study aims to develop and validate nomograms that utilize morphological and radiomics features derived from computed tomography enterography (CTE) to evaluate inflammatory activity in patients with ileocolonic Crohn’s disease (CD).MethodsA total of 54 CD patients (237 bowel segments) with clinically confirmed CD were retrospectively analyzed. The Simple Endoscopic Score for Crohn’s Disease (SES-CD) was used as a reference standard to quantify the degree of mucosal inflammation and assess disease severity. We extracted morphological and radiomics features in the training cohort to create a morphological model (M-score) and a radiomics model (Rad-score). A combined nomogram was generated by integrating the M-score and Rad-score. The predictive performance of each model was evaluated using receiver operating characteristic (ROC) curve analysis. Additionally, calibration curve and decision curve analysis (DCA) were employed to assess the accuracy and clinical applicability of the nomogram in the testing cohort.ResultsThe area under the ROC curve (AUC) for the nomogram, which included stenosis, comb sign, and Rad-score, was 0.834 [95% confidence interval (CI): 0.728–0.940] for distinguishing between active and remissive disease. Furthermore, the nomogram created using comb sign and Rad-score achieved a satisfactory AUC of 0.781 (95% CI: 0.611–0.951) in differentiating mild activity from moderate-to-severe activity. The calibration curve and DCA confirmed both nomograms’ accuracy and clinical utility.ConclusionsNomograms that combined CTE-based radiomics and morphological features could serve as valuable tools for assessing inflammatory activity, thereby supporting clinical decision-making in managing CD. Keypoints. 1. Radiomics features from CTE could predict the inflammatory activity of CD.

In Vivo Confocal Microscopy for Automated Detection of Meibomian Gland Dysfunction: A Study Based on Deep Convolutional Neural Networks.

The objectives of this study are to construct a deep convolutional neural network (DCNN) model to diagnose and classify meibomian gland dysfunction (MGD) based on the in vivo confocal microscope (IVCM) images and to evaluate the performance of the DCNN model and its auxiliary significance for clinical diagnosis and treatment. We extracted 6643 IVCM images from the three hospitals' IVCM database as the training set for the DCNN model and 1661 IVCM images from the other two hospitals' IVCM database as the test set to examine the performance of the model. Construction of the DCNN model was performed using DenseNet-169. The results of MGD classifications by three ophthalmologists were used to calculate the area under the receiver operating characteristic curve (AUROC), accuracy, precision, recall, true negative rate (TNR), true positive rate (TPR), and false positive rate (FPR) of the model. The deep learning (DL) was used to build the model to identify the IVCM images. Model accuracy and loss tests showed that the DCNN model had high accuracy, low loss, and no large fluctuations at an epoch of 175, indicating that DenseNet-169 could enable the dichotomization to proceed stably. The accuracy of each classification of the test set was above 90%, which was highly consistent with the ophthalmologists' diagnosis. The precision of the groups in each classification was more than 90%, or even close to 100%, except for the meibomian gland atrophy with obstruction group in the fifth classification. The recall ranged from 0.8728 to 0.9981, and the FPR was low in the screening and classification diagnoses. The application of DCNN can achieve accurate classification and diagnosis of MGD through IVCM images and has great potential during medical procedures.

Association between sarcopenia and weight-adjusted waist index in male patients with type 2 diabetes

BackgroundThe Weight-adjusted-waist index (WWI) has emerged as a predictive factor for a range of metabolic disorders. To date, the predictive value of the WWI in relation to sarcopenia in individuals with diabetics has not been extensively explored. This study aims to investigate the impact of the WWI on the prevalence of sarcopenia among patients with type 2 diabetes mellitus (T2DM).MethodIn this study, we enrolled 417 patients with T2DM from the Second Affiliated Hospital and Yuying Children’s Hospital of Wenzhou Medical University between Dec. 2023 and Apr. 2024. The relationship between the WWI and the prevalence of sarcopenia in T2DM patients was evaluated using multivariate logistic regression, subgroup analysis, restricted cubic spline (RCS) analysis, and receiver operating characteristic (ROC) curve analysis.ResultsAmong the 417 patients with T2DM, 76 (18.22%) were identified as having sarcopenia. The prevalence of sarcopenia across the WWI quartile categories, from the first to the fourth quartile, was 8.65%, 8.57%, 20.19% and 8.65% respectively. Multivariate logistic regression analysis revealed that, after adjusting for covariates, a higher WWI was an independent risk factor for sarcopenia in male T2DM patients (OR = 1.836, 95% CI: 1.216–2.772, P = 0.004). This association was not observed in female patients. Subgroup analysis further revealed a stronger correlation between WWI and sarcopenia among male patients with higher HbA1c levels. In males, RCS regression demonstrated a non-linear positive correlation, with an inflection point at a WWI of 10.42 cm/√kg. Finally, the area under the ROC curve (AUC) for the WWI was 0.612.ConclusionsWWI emerges as a robust and independent risk factor for sarcopenia in male patients with T2DM. WWI may serve as an accessible and cost-effective tool for identify sarcopenia in patients with diabetes.

Performance of Radiomics-based machine learning and deep learning-based methods in the prediction of tumor grade in meningioma: a systematic review and meta-analysis.

Currently, the World Health Organization (WHO) grade of meningiomas is determined based on the biopsy results. Therefore, accurate non-invasive preoperative grading could significantly improve treatment planning and patient outcomes. Considering recent advances in machine learning (ML) and deep learning (DL), this meta-analysis aimed to evaluate the performance of these models in predicting the WHO meningioma grade using imaging data. A systematic search was performed in PubMed/MEDLINE, Embase, and the Cochrane Library for studies published up to April 1, 2024, and reporting the performance metrics of the ML models in predicting of WHO meningioma grade using imaging studies. Pooled area under the receiver operating characteristics curve (AUROC), specificity, and sensitivity were estimated. Subgroup and meta-regression analyses were performed based on a number of potential influencing variables. A total of 32 studies with 15,365 patients were included in the present study. The overall pooled sensitivity, specificity, and AUROC of ML methods for prediction of tumor grade in meningioma were 85% (95% CI, 79-89%), 87% (95% CI, 81-91%), and 93% (95% CI, 90-95%), respectively. Both the type of validation and study cohort (training or test) were significantly associated with model performance. However, no significant association was found between the sample size or the type of ML method and model performance. The ML predictive models show a high overall performance in predicting the WHO meningioma grade using imaging data. Further studies on the performance of DL algorithms in larger datasets using external validation are needed.

Rapid detection of carbapenem-resistant Escherichia coli and carbapenem-resistant Klebsiella pneumoniae in positive blood cultures via MALDI-TOF MS and tree-based machine learning models

BackgroundBloodstream infection (BSI) is a systemic infection that predisposes individuals to sepsis and multiple organ dysfunction syndrome. Early identification of infectious agents and determination of drug-resistant phenotypes can help patients with BSI receive timely, effective, and targeted treatment and improve their survival. This study was based on matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS), Decision Tree (DT), Random Forest (RF), Gradient Boosting Machine (GBM), eXtreme Gradient Boosting (XGBoost), and Extremely Randomized Trees (ERT) models were constructed to classify carbapenem-resistant Escherichia coli (CREC) and carbapenem-resistant Klebsiella pneumoniae (CRKP). Bacterial species were identified by MALDI-TOF MS in positive blood cultures isolated via the serum isolation gel method, and E. coli and K. pneumoniae in positive blood cultures were collected and placed into machine learning models to predict susceptibility to carbapenems. The aim of this study was to provide rapid detection of CREC and CRKP in blood cultures, to shorten the turnaround time for laboratory reporting, and to provide a basis for early clinical intervention and rational use of antibiotics.ResultsThe collected MALDI-TOF MS data of 640 E. coli and 444 K. pneumoniae were analysed by machine learning algorithms. The area under the receiver operating characteristic curve (AUROC) for the diagnosis of E. coli susceptibility to carbapenems by the DT, RF, GBM, XGBoost, and ERT models were 0.95, 1.00, 0.99, 0.99, and 1.00, respectively, and the accuracy in predicting 149 E. coli-positive blood cultures were 0.89, 0.92, 0.90, 0.92, and 0.86, respectively. The AUROC for the diagnosis of K. pneumoniae susceptibility to carbapenems by the DT, RF, GBM, XGBoost, and ERT models were 0.78, 0.95, 0.93, 0.90, and 0.95, respectively, and the accuracy in predicting 127 K. pneumoniae-positive blood cultures were 0.76, 0.86, 0.81, 0.80, and 0.76, respectively.ConclusionsMachine learning models constructed by MALDI-TOF MS were able to directly predict the susceptibility of E. coli and K. pneumoniae in positive blood cultures to carbapenems. This rapid identification of CREC and CRKP reduces detection time and contributes to early warning and response to potential antibiotic resistance problems in the clinic.Clinical trial numberNot applicable.

Lymphocyte-C-reactive protein ratio upon admission to predict disease progression and ICU admission in adult patients with diabetic ketoacidosis

This study aimed to investigate whether lymphocyte-C-reactive protein ratio (LCR) upon admission can predict disease progression and intensive care unit (ICU) admission in adult patients with diabetic ketoacidosis (DKA). A single-center retrospective study was conducted, including adult DKA patients admitted to the First Affiliated Hospital of Harbin Medical University between March 2018 and March 2023. Multiple demographic and clinical data were collected from the medical records upon admission and during hospitalization. Subsequently, sequential organ failure assessment (SOFA) score and LCR were calculated based on relevant clinical parameters within 24 h of admission. These indicators were compared among different disease severity groups, and factors related to severe DKA, concurrent acute kidney injury (AKI), and ICU admission were further analyzed. Receiver operating characteristic (ROC) curve analysis was performed to determine the sensitivity, specificity, area under the ROC curve (AUC), and cut-off value of LCR. A total of 271 adult DKA patients were enrolled and categorized into three groups: mild group (n = 42), moderate group (n = 64), and severe group (n = 165). Significant differences in demographic and clinical data were observed among these groups. Glasgow coma scale (GCS) score, LCR, pH, and bicarbonate (HCO3-) were identified as protective factors for severe DKA. Conversely, SOFA score, neutrophil count (NEUT), serum creatinine (SCr), and glucose (GLU) were risk factors for concurrent AKI. Concurrent AKI and SOFA score were risk factors for ICU admission, while pH was a protective factor. The areas under the ROC curve (AUC) of LCR to classify adult DKA patients into mild group, severe group, and ICU admission were 0.679, 0.718, and 0.621, respectively, with cut-off value of 212.80, 96.16, and 63.35, sensitivity of 54.8%, 76.4%, and 78.9%, and specificity of 76.0%, 62.4%, and 46.3%. LCR upon admission provides great potential to predict disease progression and ICU admission in adult patients with DKA.

Comparison of Neural Network Structures for Identifying Shockable Rhythm During Cardiopulmonary Resuscitation

Background/Objectives: Minimizing interruptions in chest compressions is very important when resuscitating patients with cardiac arrest. Recently, research has analyzed electrocardiograms (ECGs) during chest compressions using convolutional neural networks (CNNs). This study aimed to compare the accuracy of deeper neural networks and more advanced structures. Methods: ECGs with chest compression artifacts were obtained from patients who visited the emergency department of Korea University Ansan Hospital from September 2019 to February 2024. We compared the accuracy of a deeper CNN, long short-term memory (LSTM), and a CNN with an attention mechanism and residual block against a reference model. The input of the model was 5 s ECG segments with compression artifacts, and the output was the probability that the ECG with the artifacts was a shockable rhythm. Results: A total of 1889 ECGs with compression artifacts from 172 patients were included in this study. There were 969 ECGs annotated as shockable and 920 as non-shockable. The area under the receiver operating characteristic curve (AUROC) of the reference model was 0.8672. The AUROCs of the deeper CNN for five and seven layers were 0.7374 and 0.6950, respectively. The AUROCs of LSTM and bidirectional LSTM were 0.7719 and 0.8287, respectively. The AUROC of the CNN with the attention mechanism and residual block was 0.7759. Conclusions: CNNs with deeper layers or those incorporating attention mechanisms, residual blocks, and LSTM architectures did not exhibit better accuracy. To improve the model accuracy, a larger dataset or advanced augmentation techniques may be required, rather than complicating the structure of the model.

Performance of the FORD Versus Other Available Models for the Noninvasive Prediction of Pulmonary Hypertension in Patients with Interstitial Lung Disease

PurposePulmonary hypertension (PH) is associated with morbidity and mortality in patients with interstitial lung disease (ILD). Several prediction models have been proposed to predict PH in ILD patients. We sought to discern how previously described prediction models perform in predicting PH in patients with ILD.MethodsPatients with ILD who completed a baseline right heart catheterization, from Inova Fairfax Hospital, Northwestern Memorial Hospital, and Asan Medical Center in Korea were enrolled. The performance of various prediction models (FORD model, the FORD calculator, the PH‐ILD Detection tool, and the mean pulmonary artery pressure prediction model) were assessed using receiver operating characteristic (ROC) curves and area under the receiver operating characteristic curve (AUROC). There were four definitions of pulmonary hypertension against which the models were evaluated.ResultsThere were a total of 192 patients with ILD, of whom 32.8% (n = 63/192) met the modified 5th world symposium on PH definition of precapillary PH. Among the models assessed, the FORD calculator had an AUROC (0.733) that was marginally highest. Subgroup analysis revealed that the FORD index had the highest AUROC (0.817) in patients with idiopathic pulmonary fibrosis, while the FORD calculator had the highest AUROC (0.751) in patients with non-IPF ILD.ConclusionThe FORD model can be used to predict group 3 PH in both IPF patients and non-IPF ILD patients. It could serve as a tool for ILD patient selection for right heart catheterization as well as an enrichment tool for clinical trials targeting the pulmonary vasculature.

P0449 qFIT as a predictor of relapse in quiescent Crohn’s disease – initial results from the Stratifying Crohn’s Using Biomarker Assessment (SCUBA) study

Abstract Background The Stratifying Crohn’s Using Biomarker Assessment (SCUBA) study assessed a number of biomarkers (Faecal Calprotectin (FCP), quantitative faecal immunochemical testing (qFIT), C reactive protein (CRP), plasma and faecal zinc) in their ability to predict relapse in quiescent Crohn’s disease (CD). FCP is widely utilised as a surrogate marker in the monitoring of CD but is labour intensive and expensive. Population qFIT testing is established in colorectal cancer screening and is relatively cheap and simple to analyse. qFIT has been shown to predict relapse in ulcerative colitis1. Here we present a comparison of qFIT and FCP at predicting relapse in quiescent luminal CD. Methods A single centre prospective cohort study was performed. Patients with clinically quiescent CD (Harvey Bradshaw Index ≤3) were identified and provided paired, baseline FCP, qFIT and routine biochemical workup. Patients over 50 years were excluded due to overlay with bowel cancer screening programmes as were those with isolated upper GI or perianal disease. Patients were followed up over 1 year or until relapse defined by need for escalation of CD related treatment, steroids, surgery or hospitalisation for CD. Results 274 patients were recruited with 211 returning paired baseline faecal samples. Median age was 34 years (range 18-49) and 108/211 (51.1%) were female. Median HBI at submission was 1 and median FCP, qFIT and CRP was 61µg/g stool, 9 µg Hb/g and 2 mg/l respectively. 42/211 (19.9%) experienced a relapse during follow up. Both FCP and qFIT showed good ability to predict relapse at 1 year, performing comparably, with the area under the receiver operating characteristic curves (AUROC) 0.80 (95% CI 0.74- 0.85) vs 0.75 (95% CI 0.68-0.80)(p=0.28). Combination biomarker use did not significantly improve performance, AUROC 0.82 (95%CI 0.76- 0.87) p=0.34 and p=0.07 respectively. Optimal threshold for prediction of relapse was &amp;gt;180 µg/g stool for FCP (sensitivity 0.69 and specificity 0.85) and &amp;gt;9 µg Hb/g for qFIT (sensitivity 0.65 and specificity 0.82). Utilising these thresholds (figure 1) patients with both FCP and qFIT above threshold had a probability of relapse free survival of 16.7% (95% CI 0.20%-31.6%) at 1 year vs 95.4% (95% CI 91.7%-99.0%, p&amp;lt;.0001) in patients under threshold. In combination, FCP and qFIT had a positive predictive value of 50.1% and negative predictive value (NPV) of 97.6%. Conclusion qFIT performs well and comparably to FCP in predicting relapse in clinically quiescent luminal CD and may be used in preference to FCP being an easier and cheaper test to undertake. When combined at optimal thresholds, qFIT &amp; FCP show significant discriminatory ability in prediction of CD relapse-free survival with an excellent NPV.