2025

Alzheimer's disease classification using MRI scans is challenging due to class imbalance in medical imaging. This study examines the effects of data augmentation, weight balancing, and synthetic image generation on classification accuracy. Six deep learning models - EfficientNetB4, ResNet152, VGG19, Xception, InceptionV3, and DenseNet121 - were tested under three conditions: (1) no augmentation, (2) traditional augmentation, and (3) synthetic image augmentation. Models trained on imbalanced data performed poorly, while weight balancing and synthetic data improved accuracy and robustness. The best performance (97%) was achieved with EfficientNetB4 and Xception using synthetic images, highlighting synthetic data as a promising solution for class imbalance in MRI-based Alzheimer's classification

The rapid development of digital technologies requires the creation of secure and efficient methods for storing and managing confidential personal data. Modern personal documents such as birth certificates, identity cards, and passports are often vulnerable to fraud, forgery, and unauthorized access. This article proposes an innovative system that integrates color QR codes and blockchain technology for the secure storage and management of biometric and documentary information. The proposed approach combines steganography, cryptography, and modern data embedding techniques to enhance document security. The combination of encrypted QR codes with a decentralized blockchain-based database ensures confidentiality, reliable authentication, data transparency and traceability. This solution is especially relevant for storing official documents, such as birth certificates, with the inclusion of biometric data of citizens and their family members for improved identification, as well as enabling effective data extraction and prevention of misuse

Carbon dioxide (CO2) emissions remain a critical factor in global energy transformation and environmental sustainability. Fossil fuel combustion is the primary source of CO2 emissions, and as one of the world's major oil producers, Kazakhstan faces significant challenges in mitigating its environmental impact. This study applies deep learning (DL) predictive models, including recurrent neural networks (RNN), bidirectional LSTM (BiLSTM), and attention mechanisms, to forecast Kazakhstan's CO2 emission scenarios for the period 2019-2023 using historical national emission data. The models achieved high predictive accuracy (MSE = 0.031, MAE = 0.18, R2 = 0.97), and the results, supported by both survey data and simulations, indicate a potential downward trend in future CO2 emissions, reflecting growing public awareness and technological advancements. These findings highlight the potential of AI-based forecasting tools for supporting sustainable energy policy. Nonetheless, the relatively short forecasting horizon and small survey sample size represent limitations that should be addressed in future work