In this paper, we have proposed a novel deep-learning model to process electrocardiogram (ECG) signals from single-lead ECG device. This is achieved by using a hybrid of CNN (convolutional neural network) and LSTM (long short-term memory) neural netwrok, where CNN extracts localized features from ECG signals, while the LSTM network extract temporal relationships. Moreover, a filtering-based method is proposed to generate synthetic data to overcome the data imbalance posed by the publicly available PhysioNet/CinC Challenge 2017 dataset. Specifically, three distinct ECG filters: Neurokit, BioSPPy, and EngZeeMod are used to perform data augmentation. As depicted by the simulation results, the proposed model exhibits superior performance with an F1 score of 99.56% providing a gain of 102.25 dB in terms of model size and performance gain of 1.06% outperforming existing state-of-the-art method enabling cost-effective real-time diagnosis with an inference time of 0.48×10-3 secs. Atrial fibrillation (A-Fib), a common cardiac arrhythmia, is linked to serious complications like heart failure and stroke, which early diagnosis could prevent. However, its paroxysmal and often asymptomatic nature complicates detection. In this paper, we propose a novel deep-learning model for single-lead ECG signal analysis, combining a convolutional neural network (CNN) to extract localized features and a long short-term memory (LSTM) network to capture temporal dependencies. To address data imbalance in the PhysioNet/CinC Challenge 2017 dataset, we employ a filtering-based approach with Neurokit, BioSPPy, and EngZeeMod for synthetic data generation. Simulation results show the model achieves a 99.56% F1 score, a 102.25 dB gain in efficiency, a 1.06% performance gain over the state-of-the-art method, and real-time diagnosis with a 0.48×10-3 second inference time. These simulation results confirm the superior performance of the proposed model.

Accurate skin lesion classification is hard because classes can look similar, datasets are imbalanced, and devices and domains vary. We introduce SynthraXCoreNet, a six CNN ensemble with ResNet50V2, ResNet101V2, ResNet152V2, DenseNet201, NASNetLarge, and Xception. Each backbone is trained per dataset, and at inference we average posterior probabilities from a shared 224×224×3 input. We assess probability calibration with reliability diagrams and Expected Calibration Error using temperature scaling. For interpretability we combine Integrated Gradients, Grad CAM variants, and saliency maps, and we score explanation quality by faithfulness, robustness, complexity, and axiomatic metrics. We include a small ablation that contrasts equal soft voting with weighted soft voting on Accuracy, Macro F1, and ECE. This work targets clinical decision support rather than autonomous diagnosis. The current scope is single lesion dermoscopic images. Multi center dermatologist reviewed prospective validation and extensions to multi lesion use are in progress. On four public datasets, test performance is reported as mean ± 95% Wilson CI. On HAM10000, Accuracy is 98.85% ± 0.68 and Macro F1 is 0.99. On ISIC2019, Accuracy is 95.30% ± 0.83 and Macro F1 is 0.95. On BCN20000, Accuracy is 95.00% ± 1.21 and Macro F1 is 0.95. On DERM12345, Accuracy is 96.63% ± 1.00 and Macro F1 is 0.97. Confusion analyses and calibration checks are consistent with strong accuracy and well calibrated probabilities. End to end inference latency is ≈ 461 ms per image with batch 1 on an RTX 4070 Ti in FP32, which supports interactive decision support at about 2.17 images per second or about 130 images per minute.