Objectives: Melanoma is the deadliest form of skin cancer, but it can be fully cured through early detection and treatment in 99% of cases. Our aim was to develop a non-invasive machine learning system that can predict the thickness of a melanoma lesion, which is a proxy for tumor progression, through dermoscopic images. This method can serve as a valuable tool in identifying urgent cases for treatment.Methods: A modern convolutional neural network architecture (EfficientNet) was used to construct a model capable of classifying dermoscopic images of melanoma lesions into three distinct categories based on thickness. We incorporated techniques to reduce the impact of an imbalanced training dataset, enhanced the generalization capacity of the model through image augmentation, and utilized five-fold cross-validation to produce more reliable metrics.Results: Our method achieved 71% balanced accuracy for three-way classification when trained on a small public dataset of 247 melanoma images. We also presented performance projections for larger training datasets.Conclusions: Our model represents a new state-of-the-art method for classifying melanoma thicknesses. Performance can be further optimized by expanding training datasets and utilizing model ensembles. We have shown that earlier claims of higher performance were mistaken due to data leakage during the evaluation process.

Background/Objectives: Donor-derived cell-free DNA (dd-cfDNA) is a marker of graft injury that increases in acute rejection and has excellent negative predictive value. Left ventricular global longitudinal strain (LVGLS) and strain-derived myocardial work indices are novel echocardiographic parameters with growing applications. Still, they have been poorly investigated in heart transplant (HTx) recipients so far. We sought to examine the diagnostic impact of left ventricular longitudinal strain-derived indices in diagnosing myocardial injury as assessed by dd-cfDNA after HTx. Methods: Since October 2022, HTx recipients have been shifted from our endomyocardial biopsy (EMB)-based rejection surveillance protocol to a monthly dd-cfDNA-led rejection assessment. We analysed the percentage of donor-derived to total cell-free DNA. For echocardiographic analysis, patient selection was restricted to those transplanted ≥ 6 months. We used 2D speckle-tracking echocardiography to assess LVGLS and strain-derived myocardial work parameters. Results: We analysed four hundred and forty-nine dd-cfDNA samples from seventy-one patients until November 2024. The mean dd-cfDNA fraction remained very low (0.13 ± 0.06%). Eighty-eight percent of surveillance EMBs that would have otherwise been performed were avoided. The mean LVGLS was lower than the literature reference values. We found no correlation between dd-cfDNA and LVGLS. Transplanted hearts had different myocardial work indices than the reference values reported in the literature. Conclusions: dd-cfDNA effectively rules out clinically significant acute rejection and decreases the need for invasive surveillance EMBs. LVGLS seems less sensitive than dd-cfDNA for the identification of myocardial injury in the early stages of HTx rejection in patients at low risk for rejection.

Over the past decades, we have witnessed an immense expansion in the arsenal and performance of machine learning (ML) algorithms. One of the most important fields that could benefit from these advancements is biomedical science. To streamline the training and evaluation of binary classifiers, we constructed a universal and flexible ML framework that uses tabular biomedical data as input. Our framework requires the input data to be provided as a comma-separated values file, in which rows correspond to subjects and columns represent different features. After reading the content of this file, the framework enables the users to perform outlier detection, handle missing values, rescale features, and tackle class imbalance. Then, hyperparameter tuning, feature selection, and internal validation are performed using nested cross-validation. If an additional dataset is available, the framework also provides the option for external validation. Users may also compute SHapley Additive exPlanations values to interpret the individual predictions of the model and identify the most important features. Our ML framework was implemented in Python (version 3.9), and its source code is freely available via GitHub. In the second part of this paper, we also demonstrate the usage of the framework through a case study from the field of cardiovascular imaging. The proposed ML framework enables the efficient training and evaluation of binary classifiers on tabular biomedical data. We hope our framework will serve as a useful resource for both learning and research purposes and will promote further innovation. Over the past decades, we have witnessed an immense expansion in the arsenal and performance of machine learning (ML) algorithms. One of the most important fields that could benefit from these advancements is biomedical science. To streamline the training and evaluation of binary classifiers, we constructed a universal and flexible ML framework that uses tabular biomedical data as input. Our framework requires the input data to be provided as a comma-separated values file, in which rows correspond to subjects and columns represent different features. After reading the content of this file, the framework enables the users to perform outlier detection, handle missing values, rescale features, and tackle class imbalance. Then, hyperparameter tuning, feature selection, and internal validation are performed using nested cross-validation. If an additional dataset is available, the framework also provides the option for external validation. Users may also compute SHapley Additive exPlanations values to interpret the individual predictions of the model and identify the most important features. Our ML framework was implemented in Python (version 3.9), and its source code is freely available via GitHub. In the second part of this paper, we also demonstrate the usage of the framework through a case study from the field of cardiovascular imaging. The proposed ML framework enables the efficient training and evaluation of binary classifiers on tabular biomedical data. We hope our framework will serve as a useful resource for both learning and research purposes and will promote further innovation.

Background: Right ventricular (RV) function is an important predictor of morbidity and mortality in various cardiovascular conditions. Nevertheless, its echocardiographic assessment is challenging due to its complex anatomy and location in the chest, resulting in limited inter-observer reproducibility. Objectives: We aimed to develop a novel deep learning model – EchoNet-RV – to segment the RV in apical 4-chamber view (A4C) echocardiographic videos and estimate RV fractional area change (RVFAC). Methods: For training EchoNet-RV, 7,169 expert-annotated A4C echocardiographic videos were used. The model’s performance was evaluated on a held-out internal test set of 1,320 A4C videos and two international external test sets of 3,107 and 1,077 A4C videos from two separate centers. Additionally, the associations between the predicted RVFAC values and the composite endpoint of heart failure hospitalization or all-cause death were also analyzed in the first external test set. Results: EchoNet-RV segmented the RV with Dice coefficients of 0.893 (0.891–0.895), 0.797 (0.796–0.798), and 0.788 (0.785–0.790) and predicted RVFAC with mean absolute errors of 5.795 (5.560–6.031), 5.830 (5.692–5.970), and 6.362 (6.064–6.660) percentage points in the held-out test set and the two external test sets, respectively. In 500 randomly selected videos from the external test sets, EchoNet-RV’s prediction error was significantly lower than the inter-observer variability (p<0.001). Moreover, it identified RVFAC <35% with areas under the receiver operating characteristic curve of 0.859 (0.843–0.876), 0.725 (0.710–0.740), and 0.684 (0.653–0.713) in the three test sets. EchoNet-RV also outperformed two multi-task models, EchoPrime and PanEcho, in estimating RVFAC and identifying RV dysfunction in the external test sets. In the first external test set, predicted RVFAC values were inversely associated with the composite endpoint (adjusted HR: 0.948 [0.917–0.979], p<0.001), independent of age, sex, cardiovascular risk factors, and left ventricular systolic function. Conclusions: EchoNet-RV enables the rapid and automated assessment of RVFAC, with strong potential to become a valuable tool for the echocardiographic evaluation of RV function and disease surveillance. CONDENSED In this study, we developed EchoNet-RV, an echocardiography-based DL model for automated RV segmentation and RVFAC estimation, and evaluated its performance on two international external datasets. EchoNet-RV demonstrated robust performance in RV segmentation, RVFAC estimation, and RV dysfunction detection, with prediction errors significantly lower than inter-observer variability. It also outperformed two multi-task models, EchoPrime and PanEcho, in estimating RVFAC and identifying RV dysfunction. Moreover, the model’s predictions were also associated with adverse clinical outcomes. EchoNet-RV enables rapid and automated RVFAC assessment, with strong potential to become a valuable tool for the echocardiographic evaluation of RV function and disease surveillance.

"Addendum to: A machine learning framework for performing binary classification on tabular biomedical data" published on 30 Apr 2025 by Akadémiai Kiadó.

Background: Orthotopic heart transplantation (OHT) remains the gold standard for end-stage heart failure, yet individualized risk assessment for postoperative mortality remains challenging. We aimed to develop and interpret random forest-based models for predicting 30-day and 1-year mortality and to examine whether the key predictors differ between the 30-day and 1-year models. Methods: We analyzed 581 patients who underwent OHT between 2012 and 2024. The 30-day and 1-year mortality rates were 9.9% and 17.6%, respectively. Eighty-seven preoperative and forty-eight postoperative variables were considered as input features for model development. Random forest models were trained and validated using five-fold cross-validation, and explainability was assessed using SHapley Additive exPlanations (SHAP). Results: Using preoperative features only, the random forest models achieved AUCs of 0.62 (95% CI, 0.48–0.75) for 30-day and 0.67 (95% CI, 0.56–0.78) for 1-year mortality. SHAP analysis revealed that early mortality predictions were primarily driven by features reflecting acute physiological stress—hepatic dysfunction, inflammation, and hemodynamic instability—whereas long-term predictions were increasingly influenced by renal function, metabolic reserve, and frailty. Incorporating postoperative features improved performance (AUC 0.98 [95% CI, 0.97–0.99] and 0.86 [95% CI, 0.80–0.92], respectively), with model predictions dominated by the severity and persistence of organ dysfunction: short-term risk driven by hepatic injury, hemodynamic compromise, and critical illness, and long-term risk by sustained hepatic and renal impairment, metabolic resilience, and duration of circulatory support. Conclusions: Random forest models integrating preoperative and immediate postoperative data could predict short- and mid-term mortality after OHT. SHAP analysis demonstrated temporal shifts in the most important predictors, supporting the role of dynamic, data-driven risk assessment in transplant care.