Parkinson’s disease is a progressive neurodegenerative disorder that primarily affects the motor system and is often difficult to diagnose in its early stages due to subtle and age-overlapping symptoms. To address this challenge, we propose a hybrid deep learning-based diagnostic framework using T2-weighted MRI scans from the PPMI dataset consisting of 77 healthy control (HC) and 223 Parkinson’s disease (PD) subjects. The images undergo standardized preprocessing, including intensity normalization and contrast-limited adaptive histogram equalization (CLAHE), to enhance structural visibility in the midbrain region. A hybrid architecture combining EfficientNetB0 for deep feature extraction with an XGBoost classifier is introduced to improve class discrimination across four categories: HCF, HCNF, PDF, and PDNF. The proposed model achieves a testing accuracy of 99.02%, demonstrating strong performance with high precision, recall, and F1-scores across all classes. These results highlight the significance of the hybrid approach in effectively handling limited and imbalanced MRI datasets while improving early PD detection. The framework shows strong potential for deployment as a fast, reliable, and clinically adaptable diagnostic tool.

Background: For melanoma, an aggressive form of skin cancer, early and accurate detection is critical to improving clinical outcomes. Deep learning–based image analysis has emerged as a promising approach for computer-assisted evaluation of dermoscopic lesions. However, model performance is often affected by limited dataset size, class imbalance, overfitting, and differences in network architecture and optimization strategies. Objectives: This study was aimed at developing and comparatively evaluating convolutional neural network (CNN) and DenseNet-based deep learning models for automated classification of dermoscopic skin lesion images. The effects of data augmentation, network architecture, optimizer selection, and training duration on classification performance were examined. Methods: Publicly available dermoscopic skin lesion images obtained from Kaggle were evaluated using a conventional CNN, DenseNet169, and DenseNet201. The original dataset contained 2357 images representing multiple benign and malignant skin lesion categories. Images were resized to 250×250 pixels, labeled, and randomly partitioned into training and testing sets at an 80:20 ratio. To address class imbalance, image augmentation using transformations, including ±10° shear, was performed, increasing the dataset size to 6857 images, with approximately 500 additional samples generated per class. Models were trained using adaptive moment estimation (Adam) and root mean square propagation (RMSprop) optimizers, and evaluated after 20, 60, and 100 epochs. Performance was assessed using accuracy, precision, recall (sensitivity), specificity, F ₁ -score, confusion matrices, and training and validation loss. Results: Data augmentation substantially improved model performance and reduced evidence of overfitting compared with the original dataset. On the augmented dataset, DenseNet169 and DenseNet201 achieved accuracy and F ₁ -scores exceeding 94% after 100 epochs, consistently outperforming the conventional CNN. DenseNet169 achieved up to 95% validation accuracy, with only minor performance differences between the Adam and RMSprop optimizers. The CNN showed greater susceptibility to overfitting on the smaller dataset; its F ₁ -score, calculated over the complete dataset, increased from 31% after 20 epochs to 59% after 60 epochs and 87% after 100 epochs before augmentation. Across the comparative experiments, the DenseNet architectures demonstrated more stable validation performance and better generalization following augmentation. Conclusions: DenseNet169 and DenseNet201 demonstrated stronger and more stable classification performance than the conventional CNN for dermoscopic skin lesion classification, particularly after augmentation of the training dataset. Increasing the dataset size and improving class balance led to reduced overfitting and improved model performance. These findings support further investigation of DenseNet-based approaches as potential clinical decision support tools for melanoma and skin lesion assessment. However, prospective evaluation using independent, multicenter datasets and diverse patient populations is required before clinical implementation.

Background Atrial fibrillation (AF) is a leading cause of cardioembolic stroke, yet the independent contribution of prediabetes to recurrent ischemic stroke risk in AF patients with prior cerebrovascular events remains poorly defined. Prior cross-sectional studies have yielded conflicting results, and no large propensity-matched investigation has examined longitudinal time-to-event outcomes in this population. Methods Using the TriNetX US Collaborative Network (67 healthcare organizations), adults with AF and prior ischemic stroke or transient ischemic attack were classified as having prediabetes (hemoglobin A1c 5.7%–6.4% or ICD-10 R73.03) or normoglycemia (hemoglobin A1c ≤5.6%). Patients with any form of diabetes mellitus were excluded. One-to-one propensity score matching balanced demographics and comorbidities. The primary outcome was recurrent ischemic stroke. Secondary outcomes included all-cause mortality, gastrointestinal (GI) bleeding, and intracranial hemorrhage (ICH). Cumulative risk and Kaplan-Meier survival analyses were performed. Results After matching, 80,335 patients per cohort were analyzed (mean age 70 years; 54.7% male; all standardized mean differences <0.02). Cumulative recurrent stroke risk did not differ between groups (60.0% vs 60.4%; risk ratio 0.993, 95% CI 0.982–1.005; P = 0.264). Time-to-recurrence was significantly longer in the prediabetes cohort (median 1,794 vs 1,695 days; hazard ratio [HR] 0.960, 95% CI 0.943–0.978; P < 0.001). Prediabetes was associated with lower all-cause mortality (HR 0.891, 95% CI 0.873–0.910; P < 0.001) and lower GI bleeding (HR 0.918, 95% CI 0.889–0.947; P = 0.003). ICH did not differ significantly (HR 0.961, 95% CI 0.917–1.007; P = 0.065). Conclusions Prediabetes was not associated with increased recurrent stroke risk in AF patients with prior cerebrovascular events. The observed longer time-to-recurrence and lower mortality may reflect a metabolic surveillance effect warranting prospective confirmation.