Area Under The Precision-recall Curve Research Articles

4DCT image artifact detection using deep learning.

Four-dimensional computed tomography (4DCT) is an es sential tool in radiation therapy. However, the 4D acquisition process may cause motion artifacts which can obscure anatomy and distort functional measurements from CTscans. We describe a deep learning algorithm to identify the location of artifacts within 4DCT images. Our method is flexible enough to handle different types of artifacts, including duplication, misalignment, truncation, andinterpolation. We trained and validated a U-net convolutional neural network artifact detection model on more than 23000 coronal slices extracted from 98 4DCT scans. The receiver operating characteristic (ROC) curve and precision-recall curve were used to evaluate the model's performance at identifying artifacts compared to a manually identified ground truth. The model was adjusted so that the sensitivity in identifying artifacts was equivalent to that of a human observer, as measured by computing the average ratio of artifact volume to lung volume in a givenscan. The model achieved a sensitivity, specificity, and precision of 0.78, 0.99, and 0.58, respectively. The ROC area-under-the-curve (AUC) was 0.99 and the precision-recall AUC was 0.73. Our model sensitivity is 8% higher than previously reported state-of-the-art artifact detectionmethods. The model developed in this study is versatile, designed to handle duplication, misalignment, truncation, and interpolation artifacts within a single image, unlike earlier models that were designed for a single artifact type.

Inconsistency among evaluation metrics in link prediction

Abstract Link prediction is a paradigmatic and challenging problem in network science, which aims to predict missing links, future links and temporal links based on known topology. Along with the increasing number of link prediction algorithms, a critical yet previously ignored risk is that the evaluation metrics for algorithm performance are usually chosen at will. This paper implements extensive experiments on hundreds of real networks and 26 well-known algorithms, revealing significant inconsistency among evaluation metrics, namely different metrics probably produce remarkably different rankings of algorithms. Therefore, we conclude that any single metric cannot comprehensively or credibly evaluate algorithm performance. In terms of information content, we suggest the usage of at least two metrics: one is the area under the receiver operating characteristic curve (AUC), and the other is one of the following three candidates, say the area under the precision-recall curve (AUPR), the area under the precision curve (AUC-Precision), and the normalized discounted cumulative gain (NDCG). When the data is imbalanced, say the number of negative samples significantly outweighs the number of positive samples, the area under the generalized ROC curve (AUC-gROC) should also be used. In addition, as we have proved the essential equivalence of threshold-dependent metrics, if in a link prediction task, some specific thresholds are meaningful, we can consider any one threshold-dependent metric with those thresholds. This work completes a missing part in the landscape of link prediction, and provides a starting point toward a well-accepted criterion or standard to select proper evaluation metrics for link prediction.

A large scale multi institutional study for radiomics driven machine learning for meningioma grading

This study aims to develop and evaluate radiomics-based machine learning (ML) models for predicting meningioma grades using multiparametric magnetic resonance imaging (MRI). The study utilized the BraTS-MEN dataset’s training split, including 698 patients (524 with grade 1 and 174 with grade 2–3 meningiomas). We extracted 4872 radiomic features from T1, T1 with contrast, T2, and FLAIR MRI sequences using PyRadiomics. LASSO regression reduced features to 176. The data was split into training (60%), validation (20%), and test (20%) sets. Five ML algorithms (TabPFN, XGBoost, LightGBM, CatBoost, and Random Forest) were employed to build models differentiating low-grade (grade 1) from high-grade (grade 2–3) meningiomas. Hyperparameter tuning was performed using Optuna, optimizing model-specific parameters and feature selection. The CatBoost model demonstrated the best performance, achieving an area under the receiver operating characteristic curve (AUROC) of 0.838 [95% confidence interval (CI): 0.689–0.935], precision of 0.492 (95% CI: 0.371–0.623), recall of 0.838 (95% CI: 0.689–0.935), F1 score of 0.620 (95% CI: 0.495–0.722), accuracy of 0.729 (95% CI: 0.650–0.800), an area under the precision-recall curve (AUPRC) of 0.620 (95% CI: 0.433–0.753), and Brier score of 0.156 (95% CI: 0.122-0.200). Other models showed comparable performance, with mean AUROCs ranging from 0.752 to 0.784. The radiomics-based ML approach presented in this study showcases the potential for non-invasive and pre-operative grading of meningiomas using multiparametric MRI. Further validation on larger and independent datasets is necessary to establish the robustness and generalizability of these findings.

Integrated Knowledge Graph and Drug Molecular Graph Fusion via Adversarial Networks for Drug-Drug Interaction Prediction.

The Co-administration of multiple drugs can enhance the efficacy of disease treatment by reducing drug resistance and side effects. However, it also raises the risk of adverse drug interactions, presenting a challenging problem in healthcare. Various approaches have been developed to predict drug-drug interactions (DDIs) by leveraging both knowledge graphs and drug attribute information. While these methods have shown promise, they often fail to effectively capture correlations between biomedical information in the knowledge graph and drug properties. This work introduces a novel end-to-end DDI predictor framework based on generative adversarial networks. This framework utilizes a message-passing neural network to capture molecular structure information while employing the knowledge-aware graph attention network to capture the representation of drugs in the knowledge graph through considering the importance of different multihop neighbor nodes and relationships. The dual generative adversarial networks employ two generators and two discriminators in knowledge graph embedding and molecular topology embedding for adversarial training to capture the interrelations and complementary knowledge between molecular structure information and semantic information from the knowledge graph. comprehensive experiments have demonstrated that the proposed method outperforms state-of-the-art algorithms in binary classification, with improvements of 1.0% in accuracy, 0.45% in area under the receiver operating characteristic curve (AUC), 0.24% in area under the precision-recall curve (AUPR), and 0.98% in F1 score. Furthermore, for multiclass classification tasks, improvements were observed across various evaluation metrics, including 0.9% in accuracy, 0.25% in macro precision, 0.2% in macro recall, and 0.28% in macro F1. Additionally, ablation studies were conducted to showcase the effectiveness and robustness of our method in DDI prediction tasks.

Artificial intelligence in cardiology: a machine learning model for supraventricular tachycardia discrimination

Abstract Background Distinguishing atrioventricular nodal reentrant tachycardia (AVNRT) from orthodromic atrioventricular reentrant tachycardia (AVRT) presents a diagnostic challenge, with clinical and electrocardiogram (ECG) features playing a crucial role in differential diagnosis. Purpose The aim of this study was to devise and validate a predictive algorithm that utilises basic clinical and ECG parameters to accurately classify AVNRT versus AVRT. Furthermore, this algorithm was integrated into a user-friendly, web-based tool to support clinical decision-making. Methods We conducted a prospective analysis of 704 patients (median age: 53.6 years; 59.7% female) with electrophysiologically confirmed AVNRT (n=535) or AVRT (n=169). A comprehensive evaluation was performed using a structured questionnaire tailored to arrhythmia-specific symptoms. The study utilised advanced machine learning techniques for classification, employing feature selection methods to enhance predictive accuracy. The analysis included 19 demographic, clinical, and ECG variables, adopting an 80-20% split for training and internal validation. The algorithm's performance was evaluated against a model comprising all variables (Full Model), expert ECG interpretations (ExpertECG), objective ECG criteria (ECG model), and a baseline dummy classifier, with efficacy assessed by the Area Under the Receiver Operating Characteristic curve (AUC) and the Area Under the Precision-Recall Curve (AUPRC). Results The optimised model, utilising the XGBoost algorithm with six critical variables (initial age at first episode, biological sex, presence of neck palpitations, pseudoR wave in lead V1/pseudoS wave in inferior leads, visibility of retrograde P-waves, and tachycardia cycle length), exhibited superior performance, achieving an AUC of 0.916 and an AUPRC of 0.823. This model significantly outperformed other comparative models in terms of predictive accuracy (Figure 1). Implementation of the model into a web-based interface was accomplished using Shiny, enhancing its accessibility for clinical application (Figure 2). Conclusions This study illustrates the efficacy of employing machine learning algorithms to utilise bedside clinical and ECG variables for accurately differentiating between AVNRT and AVRT. The creation of a web-based calculator marks a significant advancement, empowering clinicians, especially non-specialists, with a potent tool for providing personalised diagnostic insights. Future research should focus on external validation to ascertain the model's generalisability and its applicability in broader clinical settings.Figure 1Figure 2

Machine learning approach for detection of valvular heart disease through digital cardiac auscultation: results from multi-center prospective cross-sectional study

Abstract Background Cardiac auscultation, an essential bedside diagnostic tool, often identifies significant valvular heart disease through murmurs. However, its accuracy is contingent upon the clinician's expertise, leading to a growing dependence on expensive imaging techniques. This study aimed to develop and validate a machine learning model for detecting valvular heart disease from cardiac auscultation. Methods The digital cardiac auscultation and associated echocardiograms were prospectively collected from three sites between May 1, 2022, and Dec 31, 2022. We trained, validated, and externally tested a machine learning based model to classify aortic stenosis, mitral stenosis, aortic regurgitation, mitral regurgitation, tricuspid regurgitation, oveall valvular heart disease from cardiac auscultation, age, sex, and comorbidity. We condcted training, and internal testing using datasets for eXtreme Gradient Boosting (XGBoost) machine learning model from hospital one and hospital two. The dataset from hospital three was employed for external testing. We evaluated the area under the receiver operating characteristic curve (AUROC), area under the precision recall curve (AUPRC), accuracy, sensitivity, specificity. Result We selected the cardiac auscultation and echocardiograms of a total of 4,321 patients, excluding those without valvular heart disease information, congenital heart disease, a history of valvular operation, and cardiac auscultation could not be interpreted. The diagnostic performance of heart murmurs for valvular heart disease across the three datasets generally demonstrated high sensitivity and PPV. Specifically, the sensitivity for aortic stenosis was 89.1%, mitral stenosis 88.8%, aortic regurgitation 88.5%, mitral regurgitation 89.8%, tricuspid regurgitation 90.1%, and for any valvular heart disease, it was 95.0%. However, the NPV was consistently lower, with values of 16.4%, 14.2%, 12.2%, 23.1%, 26.2%, and 65.3% for the respective conditions. In XGBoost machine learning model, the AUROC values for internal testing (N=1,403), along with their 95% confidence intervals, were as follows: aortic stenosis (0.93, 0.88–0.97), mitral stenosis (0.91, 0.81–0.97), aortic regurgitation (0.84, 0.79–0.90), mitral regurgitation (0.94, 0.90–0.98), tricuspid regurgitation (0.81, 0.75–0.87), and overall valvular heart disease (0.86, 0.82–0.89). For external testing (N=814), the AUROC values were: aortic stenosis (0.90, 0.83–0.96), mitral stenosis (0.88, 0.79–0.95), aortic regurgitation (0.94, 0.93–0.96), mitral regurgitation (0.93, 0.92–0.95), tricuspid regurgitation (0.89, 0.87–0.91), and overall valvular heart disease (0.92, 0.90–0.94). Conclusion The machine learning based model can accurately classify valvular heart diseases using information from digital cardiac auscultation.Figure 1.Study FlowFigure 2.AUC curve

A tri-light warning system for hospitalized COVID-19 patients: Credibility-based risk stratification for future pandemic preparedness

PurposeThe novel coronavirus pneumonia (COVID-19) has continually spread and mutated, requiring a patient risk stratification system to optimize medical resources and improve pandemic response. We aimed to develop a conformal prediction-based tri-light warning system for stratifying COVID-19 patients, applicable to both original and emerging variants. MethodsWe retrospectively collected data from 3646 patients across multiple centers in China. The dataset was divided into a training set (n = 1451), a validation set (n = 662), an external test set from Huoshenshan Field Hospital (n = 1263), and a specific test set for Delta and Omicron variants (n = 544). The tri-light warning system extracts radiomic features from CT (computed tomography) and integrates clinical records to classify patients into high-risk (red), uncertain-risk (yellow), and low-risk (green) categories. Models were built to predict ICU (intensive care unit) admissions (adverse cases in training/validation/Huoshenshan/variant test sets: n = 39/21/262/11) and were evaluated using AUROC ((area under the receiver operating characteristic curve)) and AUPRC ((area under the precision-recall curve)) metrics. ResultsThe dataset included 1830 men (50.2 %) and 1816 women (50.8 %), with a median age of 53.7 years (IQR [interquartile range]: 42–65 years). The system demonstrated strong performance under data distribution shifts, with AUROC of 0.89 and AUPRC of 0.42 for original strains, and AUROC of 0.77–0.85 and AUPRC of 0.51–0.60 for variants. ConclusionThe tri-light warning system can enhance pandemic responses by effectively stratifying COVID-19 patients under varying conditions and data shifts.

Machine learning for predicting device-associated infection and 30-day survival outcomes after invasive device procedure in intensive care unit patients

This study aimed to preliminarily develop machine learning (ML) models capable of predicting the risk of device-associated infection and 30-day outcomes following invasive device procedures in intensive care unit (ICU) patients. The study utilized data from 8574 ICU patients who underwent invasive procedures, sourced from the Medical Information Mart for Intensive Care (MIMIC)-IV version 2.2 database. Patients were allocated into training and validation datasets in a 7:3 ratio. Seven ML models were employed for predicting device-associated infections, while five models were used for predicting 30-day survival outcomes. Model performance was primarily evaluated using the receiver operating characteristic (ROC) curve for infection prediction and the survival model’s concordance index (C-index). Top-performing models progressively reduced the number of variables based on their importance, thereby optimizing practical utility. The inclusion of all variables demonstrated that extreme gradient boosting (XGBoost) and extra survival trees (EST) models yielded superior discriminatory performance. Notably, when restricted to the top 10 variables, both models maintained performance levels comparable to when all variables were included. In the validation cohort, the XGBoost model, with the top 10 variables, achieved an area under the curve (AUC) of 0.810 (95% CI 0.808–0.812), an area under the precision-recall curve (AUPRC) of 0.226 (95% CI 0.222–0.230), and a Brier score (BS) of 0.053 (95% CI 0.053–0.054). The EST model, with the top 10 variables, reported a C-index of 0.756 (95% CI 0.754–0.757), a time-dependent AUC of 0.759 (95% CI 0.763–0.775), and an integrated Brier score (IBS) of 0.087 (95% CI 0.087–0.087). Both models are accessible via a web application. The internally evaluated XGBoost and EST models demonstrated exceptional predictive accuracy for device-associated infection risks and 30-day survival outcomes post-invasive procedures in ICU patients. Further validation is required to confirm the clinical utility of these two models in future studies.

Risk-Specific Training Cohorts to Address Class Imbalance in Surgical Risk Prediction.

Machine learning tools are increasingly deployed for risk prediction and clinical decision support in surgery. Class imbalance adversely impacts predictive performance, especially for low-incidence complications. To evaluate risk-prediction model performance when trained on risk-specific cohorts. This cross-sectional study performed from February 2024 to July 2024 deployed a deep learning model, which generated risk scores for common postoperative complications. A total of 109 445 inpatient operations performed at 2 University of Florida Health hospitals from June 1, 2014, to May 5, 2021 were examined. The model was trained de novo on separate cohorts for high-risk, medium-risk, and low-risk Common Procedure Terminology codes defined empirically by incidence of 5 postoperative complications: (1) in-hospital mortality; (2) prolonged intensive care unit (ICU) stay (≥48 hours); (3) prolonged mechanical ventilation (≥48 hours); (4) sepsis; and (5) acute kidney injury (AKI). Low-risk and high-risk cutoffs for complications were defined by the lower-third and upper-third prevalence in the dataset, except for mortality, cutoffs for which were set at 1% or less and greater than 3%, respectively. Model performance metrics were assessed for each risk-specific cohort alongside the baseline model. Metrics included area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), F1 scores, and accuracy for each model. A total of 109 445 inpatient operations were examined among patients treated at 2 University of Florida Health hospitals in Gainesville (77 921 procedures [71.2%]) and Jacksonville (31 524 procedures [28.8%]). Median (IQR) patient age was 58 (43-68) years, and median (IQR) Charlson Comorbidity Index score was 2 (0-4). Among 109 445 operations, 55 646 patients were male (50.8%), and 66 495 patients (60.8%) underwent a nonemergent, inpatient operation. Training on the high-risk cohort had variable impact on AUROC, but significantly improved AUPRC (as assessed by nonoverlapping 95% confidence intervals) for predicting mortality (0.53; 95% CI, 0.43-0.64), AKI (0.61; 95% CI, 0.58-0.65), and prolonged ICU stay (0.91; 95% CI, 0.89-0.92). It also significantly improved F1 score for mortality (0.42; 95% CI, 0.36-0.49), prolonged mechanical ventilation (0.55; 95% CI, 0.52-0.58), sepsis (0.46; 95% CI, 0.43-0.49), and AKI (0.57; 95% CI, 0.54-0.59). After controlling for baseline model performance on high-risk cohorts, AUPRC increased significantly for in-hospital mortality only (0.53; 95% CI, 0.42-0.65 vs 0.29; 95% CI, 0.21-0.40). In this cross-sectional study, by training separate models using a priori knowledge for procedure-specific risk classes, improved performance in standard evaluation metrics was observed, especially for low-prevalence complications like in-hospital mortality. Used cautiously, this approach may represent an optimal training strategy for surgical risk-prediction models.

Machine learning-based diagnostic prediction of minimal change disease: model development study

Minimal change disease (MCD) is a common cause of nephrotic syndrome. Due to its rapid progression, early detection is essential; however, definitive diagnosis requires invasive kidney biopsy. This study aims to develop non-invasive predictive models for diagnosing MCD by machine learning. We retrospectively collected data on demographic characteristics, blood tests, and urine tests from patients with nephrotic syndrome who underwent kidney biopsy. We applied four machine learning algorithms—TabPFN, LightGBM, Random Forest, and Artificial Neural Network—and logistic regression. We compared their performance using stratified 5-repeated 5-fold cross-validation for the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC). Variable importance was evaluated using the SHapley Additive exPlanations (SHAP) method. A total of 248 patients were included, with 82 cases (33%) were diagnosed with MCD. TabPFN demonstrated the best performance with an AUROC of 0.915 (95% CI 0.896–0.932) and an AUPRC of 0.840 (95% CI 0.807–0.872). The SHAP methods identified C3, total cholesterol, and urine red blood cells as key predictors for TabPFN, consistent with previous reports. Machine learning models could be valuable non-invasive diagnostic tools for MCD.

MGACL: Prediction Drug-Protein Interaction Based on Meta-Graph Association-Aware Contrastive Learning.

The identification of drug-target interaction (DTI) is crucial for drug discovery. However, how to reduce the graph neural network's false positives due to its bias and negative transfer in the original bipartite graph remains to be clarified. Considering that the impact of heterogeneous auxiliary information on DTI varies depending on the drug and target, we established an adaptive enhanced personalized meta-knowledge transfer network named Meta Graph Association-Aware Contrastive Learning (MGACL), which can transfer personalized heterogeneous auxiliary information from different nodes and reduce data bias. Meanwhile, we propose a novel DTI association-aware contrastive learning strategy that aligns high-frequency drug representations with learned auxiliary graph representations to prevent negative transfer. Our study improves the DTI prediction performance by about 3%, evaluated by analyzing the area under the curve (AUC) and area under the precision-recall curve (AUPRC) compared with existing methods, which is more conducive to accurately identifying drug targets for the development of new drugs.

Risk for Suicide Attempts Assessed Using the Patient Health Questionnaire–9 Modified for Teens

Suicide is a leading cause of death in US youths. To assess whether screening with supplemental items 10 to 13 on the Patient Health Questionnaire-9 modified for teens (PHQ-9M) improves prediction of youth suicide attempts beyond the information provided by the first 9 items alone (the PHQ-9). This retrospective cohort study used a retrospective cohort of adolescents aged 12 to 17 years who were screened for depression in outpatient facilities within a pediatric health care system between January 1, 2016, and December 31, 2022, with up to 1 year of follow-up to assess the occurrence of suicidal behavior. Follow-up was completed on December 31, 2023. Screening with the PHQ-9M. This study developed and compared prediction using 3 Cox proportional hazards regression models (CR-9, CR-13, and CR-3) of subsequent suicide attempts, determined by the hospital's electronic health records up to 1 year following the last PHQ-9M screening. The CR-9 model used the PHQ-9 and the CR-13 model used all 13 items of PHQ-9M. The CR-3 model used the 3 most impactful variables selected from the 13 PHQ-9M items and PHQ-9 total score. All models were evaluated across 4 prediction horizons (30, 90, 180, and 365 days) following PHQ-9M screenings. Evaluation metrics were the area under the receiver operating characteristic curve (AUROC) and the area under the precision recall curve (AUPRC). Of 130 028 outpatients (65 520 [50.4%] male) with 272 402 PHQ-9M screenings, 549 (0.4%) had subsequent suicide attempts within 1 year following the PHQ-9M screening. The AUROC of the CR-9 model in the 365-day horizon was 0.77 (95% CI, 0.75-0.79); of the CR-13 model, 0.80 (95% CI, 0.78-0.82); and of the CR-3 model, 0.79 (95% CI, 0.76-0.81); the AUPRC of the CR-9 model was 0.02 (95% CI, 0.02-0.03); of the CR-13 model, 0.03 (95% CI, 0.02-0.03); and of the CR-3 model, 0.02 (95% CI, 0.02-0.03). The 3 most impactful items using adjusted hazard ratios were supplemental item 13 (lifetime suicide attempts; 3.06 [95% CI, 2.47-3.80]), supplemental item 10 (depressed mood severity in the past year; 2.99 [95% CI, 2.32-3.86]), and supplemental item 12 (serious suicidal ideation in the past month; 1.63 [95% CI, 1.25-2.12]). All of the models achieved higher AUROCs as prediction horizons shortened. In this cohort study of adolescent PHQ-9M screenings, the supplemental items on PHQ-9M screening improved prediction of youth suicide attempts compared with screening using the PHQ-9 across all prediction horizons, suggesting that PHQ-9M screening should be considered during outpatient visits to improve prediction of suicide attempts.

ECG data analysis to determine ST-segment elevation myocardial infarction and infarction territory type: an integrative approach of artificial intelligence and clinical guidelines.

Acute coronary syndrome (ACS) is one of the leading causes of death from cardiovascular diseases worldwide, with ST-segment elevation myocardial infarction (STEMI) representing a severe form of ACS that exhibits high prevalence and mortality rates. This study proposes a new method for accurately diagnosing STEMI and categorizing the infarction area in detail, based on 12-lead electrocardiogram (ECG) data using a deep learning-based artificial intelligence (AI) algorithm. Utilizing an ECG database consisting of 888 myocardial infarction (MI) patients, this study enhanced the generalization ability of the AI model through five-fold cross-validation. The developed ST-segment elevation (STE) detector accurately identified STE across all 12 leads, which is a crucial indicator for the clinical ECG diagnosis of STEMI. This detector was employed in the AI model to differentiate between STEMI and non-ST-segment elevation myocardial infarction (NSTEMI). In the process of distinguishing between STEMI and NSTEMI, the average area under the receiver operating characteristic curve (AUROC) was 0.939, and the area under the precision-recall curve (AUPRC) was 0.977, demonstrating significant results. Furthermore, this detector exhibited the ability to accurately differentiate between various infarction territories in the ECG, including anterior myocardial infarction (AMI), inferior myocardial infarction (IMI), lateral myocardial infarction (LMI), and suspected left main disease. These results suggest that integrating clinical domains into AI technology for ECG diagnosis can play a crucial role in the rapid treatment and improved prognosis of STEMI patients. This study provides an innovative approach for the diagnosis of cardiovascular diseases and contributes to enhancing the practical applicability of AI-based diagnostic tools in clinical settings.

Conditional probabilistic diffusion model driven synthetic radiogenomic applications in breast cancer.

This study addresses the heterogeneity of Breast Cancer (BC) by employing a Conditional Probabilistic Diffusion Model (CPDM) to synthesize Magnetic Resonance Images (MRIs) based on multi-omic data, including gene expression, copy number variation, and DNA methylation. The lack of paired medical images and genomics data in previous studies presented a challenge, which the CPDM aims to overcome. The well-trained CPDM successfully generated synthetic MRIs for 726 TCGA-BRCA patients, who lacked actual MRIs, using their multi-omic profiles. Evaluation metrics such as Frechet's Inception Distance (FID), Mean Square Error (MSE), and Structural Similarity Index Measure (SSIM) demonstrated the CPDM's effectiveness, with an FID of 2.02, an MSE of 0.02, and an SSIM of 0.59 based on the 15-fold cross-validation. The synthetic MRIs were used to predict clinical attributes, achieving an Area Under the Receiver-Operating-Characteristic curve (AUROC) of 0.82 and an Area Under the Precision-Recall Curve (AUPRC) of 0.84 for predicting ER+/HER2+ subtypes. Additionally, the MRIs served to accurately predicted BC patient survival with a Concordance-index (C-index) score of 0.88, outperforming other baseline models. This research demonstrates the potential of CPDMs in generating MRIs based on BC patients' genomic profiles, offering valuable insights for radiogenomic research and advancements in precision medicine. The study provides a novel approach to understanding BC heterogeneity for early detection and personalized treatment.

Prediction and validation of pathologic complete response for locally advanced rectal cancer under neoadjuvant chemoradiotherapy based on a novel predictor using interpretable machine learning

BackgroundPrecise evaluation of pathological complete response (pCR) is essential for determining the prognosis of patients with locally advanced rectal cancer (LARC) undergoing neoadjuvant chemoradiotherapy (NCRT) and can offer clues for the selection of subsequent treatment strategies. Most current predictive models for pCR focus primarily on pre-treatment factors, neglecting the dynamic systemic changes that occur during neoadjuvant chemoradiotherapy, and are constrained by low accuracy and lack of integrity. PurposeThis study devised a novel predictor of pCR using dynamic alterations in systemic inflammation-nutritional marker indexes (SINI) during neoadjuvant therapy and developed a machine-learning model to predict pCR. MethodsTwo cohorts of patients with LARC from center one from 2012 to 2017 and from center two from 2020 to 2023 were integrated for analysis. This study compared dynamic changes in blood indexes before and after neoadjuvant therapy and surgical operation. A least absolute shrinkage and selection operator (LASSO) regression analysis was conducted to mitigate collinearity and identify key indexes, constructing the SINI. Univariate and multiple logistic regression analyses were used to identify the independent risk factors associated with pCR. Additionally, 10 machine learning algorithms were employed to develop predictive models to assess risk. The hyperparameters of the machine learning models were optimized using a random search and 10-fold cross-validation. The models were assessed by examining various metrics, including the area under the receiver operating characteristic curves (AUC), the area under the precision-recall curve (AUPRC), decision curve analysis, calibration curves, and the precision and accuracy of the internal and external validation cohorts. Additionally, Shapley's additive explanations (SHAP) were employed to interpret the machine learning models. ResultsThe study cohort comprised 677 patients from the center one and 224 patients from the center two. Six key indexes were identified, and a predictive index, SINI, was constructed. Univariate and multiple logistic regression analyses revealed that SINI, clinical T-stage, clinical N-stage, tumor size, and the distance from the anal verge were independent risk factors for pCR in patients with LARC following NCRT. The mean AUC value of the extreme gradient boosting (XGB) model in the 10-fold cross-validation of the training set was 0.877. The XGB model demonstrated superior performance in the internal and external validation sets. Specifically, in the internal test set, the XGB model achieved an AUC of 0.86, AUPRC of 0.707, accuracy of 0.82, and precision of 0.80. In the external validation set, the XGB model exhibited an AUC of 0.83, AUPRC of 0.702, accuracy of 0.81, and precision of 0.81. Additionally, the predictions generated by the XGB model were analyzed using SHAP. ConclusionThis study involved developing and validating an XGB model using SINI to predict pCR in patients with LARC. Besides, a SINI-based machine learning model shows promise in accurately predicting pCR following NCRT in patients with resectable LARC, offering valuable insights for personalized treatment approaches.

External validation of an artificial intelligence multi-label deep learning model capable of ankle fracture classification

BackgroundAdvances in medical imaging have made it possible to classify ankle fractures using Artificial Intelligence (AI). Recent studies have demonstrated good internal validity for machine learning algorithms using the AO/OTA 2018 classification. This study aimed to externally validate one such model for ankle fracture classification and ways to improve external validity.MethodsIn this retrospective observation study, we trained a deep-learning neural network (7,500 ankle studies) to classify traumatic malleolar fractures according to the AO/OTA classification. Our internal validation dataset (IVD) contained 409 studies collected from Danderyd Hospital in Stockholm, Sweden, between 2002 and 2016. The external validation dataset (EVD) contained 399 studies collected from Flinders Medical Centre, Adelaide, Australia, between 2016 and 2020. Our primary outcome measures were the area under the receiver operating characteristic (AUC) and the area under the precision-recall curve (AUPR) for fracture classification of AO/OTA malleolar (44) fractures. Secondary outcomes were performance on other fractures visible on ankle radiographs and inter-observer reliability of reviewers.ResultsCompared to the weighted mean AUC (wAUC) 0.86 (95%CI 0.82–0.89) for fracture detection in the EVD, the network attained wAUC 0.95 (95%CI 0.94–0.97) for the IVD. The area under the precision-recall curve (AUPR) was 0.93 vs. 0.96. The wAUC for individual outcomes (type 44A-C, group 44A1-C3, and subgroup 44A1.1-C3.3) was 0.82 for the EVD and 0.93 for the IVD. The weighted mean AUPR (wAUPR) was 0.59 vs 0.63. Throughout, the performance was superior to that of a random classifier for the EVD.ConclusionAlthough the two datasets had considerable differences, the model transferred well to the EVD and the alternative clinical scenario it represents. The direct clinical implications of this study are that algorithms developed elsewhere need local validation and that discrepancies can be rectified using targeted training. In a wider sense, we believe this opens up possibilities for building advanced treatment recommendations based on exact fracture types that are more objective than current clinical decisions, often influenced by who is present during rounds.

Deep Learning Model for Predicting Neurodevelopmental Outcome in Very Preterm Infants Using Cerebral Ultrasound

ObjectiveTo develop deep learning (DL) models applied to neonatal cranial ultrasound (CUS) and clinical variables to predict neurodevelopmental impairment (NDI) in very preterm babies (VPI) at 3 years of corrected age. Patients and MethodsA retrospective study of a cohort of VPI (220-306 weeks' gestation) born between 2004 and 2016 in Nova Scotia, Canada. Clinical data at hospital discharge, and CUS images at three time points were used to develop DL models using elastic net (EN) and convolutional neural network (CNN). The models' performance was compared using Precision Recall Area Under the Curve (PR-AUC) and area under Receiver Operation Characteristic curve (ROC-AUC) with their 95% Confidence intervals. ResultsOf 665 eligible VPI, 619 (93%) infants with 4184 CUS images were included. The CNN model combining CUS, and clinical variables showed better performance (PR-AUC 0.75, C.I. 072 – 0.79 and ROC-AUC 0.71, CI 0.67 - 0.74) in the prediction of positive NDI outcome compared to the traditional models based solely on clinical predictors (PR AUC 0.60 (0.52-0.68), and ROC-AUC of 0.72 (0.68-0.75). When analyzed by the CUS plane and acquisition timepoint, the model utilizing the anterior coronal plane at 6 weeks of age provided the highest predictive accuracy: PR-AUC 0.81 (0.77-0.91) and ROC-AUC 0.78 (0.66-87). ConclusionWe developed and internally validated a DL prognostic model to predict NDI in VPI at 3 years utilizing CUS and clinical predictors. Early and accurate identification of infants at risk for NDI enables referral to targeted interventions which improves functional outcomes.

Predicting Mortality in Trauma Research: Evaluating the Performance of Trauma Scoring Tools in a South African Population.

Background Trauma is a leading cause of death and disability in low-resource settings, yet trauma severity scores are seldom validated in these contexts. There is a pressing need to better characterize and compare trauma scoring tools, especially within research frameworks. This study aimed to evaluate the performance of various trauma scoring tools in predicting in-hospital mortality among trauma patients in South Africa. Methods This study conducted a secondary analysis of existing data from the multicenter Epidemiology and Outcomes of Prolonged Trauma Care (EpiC) study, which included 13,548 adult trauma patients aged 18 years and older, collected between August 2021 and March 2024. The predictive ability of the scoring tools was assessed by calculating the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC). Results The mortality rate was 2.5% (n = 298). The Kampala Trauma Score (KTS) demonstrated the highest predictive ability for seven-day in-hospital mortality, with an AUROC of 0.95 and an AUPRC of 0.53. Similarly, the Trauma and Injury Severity Score (TRISS) and the New Injury Severity Score (NISS) also exhibited strong predictive capabilities, with AUROC values of 0.96 and AUPRC values of 0.62 for TRISS and an AUROC of 0.96 and AUPRC of 0.53 for NISS. In contrast, the Revised Trauma Score and Mechanism, Glasgow Coma Scale, Age, and Arterial Pressure (MGAP) showed lower predictive performance, with AUROC values of 0.87 (AUPRC = 0.51) and 0.86 (AUPRC = 0.47), respectively. Conclusions The KTS exhibited optimal performance characteristics for retrospectively predicting mortality in our cohort, outperforming other scoring tools. Notably, it is also the simplest scoring tool, featuring the fewest variables compared to other trauma severity assessments. These findings highlight the necessity for external validation of trauma scoring tools in resource-limited populations to ensure their applicability and effectiveness in trauma research across diverse healthcare settings.

Interictal stereo-electroencephalography features below 45 Hz are sufficient for correct localization of the epileptogenic zone and postsurgical outcome prediction.

Evidence suggests that the most promising results in interictal localization of the epileptogenic zone (EZ) are achieved by a combination of multiple stereo-electroencephalography (SEEG) biomarkers in machine learning models. These biomarkers usually include SEEG features calculated in standard frequency bands, but also high-frequency (HF) bands. Unfortunately, HF features require extra effort to record, store, and process. Here we investigate the added value of these HF features for EZ localization and postsurgical outcome prediction. In 50 patients we analyzed 30 min of SEEG recorded during non-rapid eye movement sleep and tested a logistic regression model with three different sets of features. The first model used broadband features (1-500 Hz); the second model used low-frequency features up to 45 Hz; and the third model used HF features above 65 Hz. The EZ localization by each model was evaluated by various metrics including the area under the precision-recall curve (AUPRC) and the positive predictive value (PPV). The differences between the models were tested by the Wilcoxon signed-rank tests and Cliff's Delta effect size. The differences in outcome predictions based on PPV values were further tested by the McNemar test. The AUPRC score of the random chance classifier was .098. The models (broad-band, low-frequency, high-frequency) achieved median AUPRCs of .608, .582, and .522, respectively, and correctly predicted outcomes in 38, 38, and 33 patients. There were no statistically significant differences in AUPRC or any other metric between the three models. Adding HF features to the model did not have any additional contribution. Low-frequency features are sufficient for correct localization of the EZ and outcome prediction with no additional value when considering HF features. This finding allows significant simplification of the feature calculation process and opens the possibility of using these models in SEEG recordings with lower sampling rates, as commonly performed in clinical routines.

Hemolysis Index, Carboxyhemoglobin, and Methemoglobin for the Early Identification of Patients at Risk for Cardiac Surgery–Associated Acute Kidney Injury

Hemolysis is a contributor to CS-AKI. Biochemistry analyzers provide a hemolysis index to quantify in vitro hemolysis, a condition that can, for example, affect the accuracy of potassium concentration measurements. We aimed to assess whether the postoperative plasma level of the hemolysis index (HIpostoperative) could aid the early recognition of patients at risk for cardiac surgery-associated acute kidney injury (CS-AKI) and also to evaluate other hemolysis indicators: plasma carboxyhemoglobin (COHbpostoperative) and methemoglobin (MetHbpostoperative). One-year retrospective study. University hospital. Patients undergoing elective cardiac surgery. None. In 1090 patients, the median HIpostoperative was higher in patients who developed CS-AKI compared to patients who did not (11 mg/dL [interquartile range (IQR), 5-38 mg/dL] v 7 mg/dL [IQR, 3-16 mg/dL]; p < 0.001). HIpostoperative refined the early recognition of CS-AKI: the area under the precision-recall curve (AUPRC) for HIpostoperative was 37% (95% confidence interval [CI], 31%-42%), whereas the AUPRC associated with no discriminative power, equal to the prevalence of CS-AKI in the whole population, was 21%. Among the 611 patients with measurements for all 3 biomarkers, the AUPRC of HIpostoperative was higher than that of COHbpostoperative or MetHbpostoperative (+6.6% and +7.4% respectively; p < 0.0001 for both). Unlike COHbpostoperative or MetHbpostoperative, the incorporation of HIpostoperative into a model (trained on a sample then validated in another sample) of CS-AKI early recognition significantly enhanced its performance, with a +1.9% (95% CI, 1.6%-2.1%) increase in AUPRC (p < 0.0001). Elevated HIpostoperative represents an early alert signal for the development of CS-AKI. Our findings support the incorporation of HIpostoperative, a readily available biomarker, into predictive scores of CS-AKI.