International Journal of Data Science, Bioinformatics and Cyber Security

A Survey on Nutri-Guide: An Intelligent System for Personalized Dietary Management and Nutritional Analysis

Mythri M — 2026-09-23

Maintaining a balanced diet is central to preventing lifestyle-related and chronic conditions such as obesity, diabetes, cardiovascular disease, and hypertension, yet conventional dietary assessment methods — food diaries, food-frequency questionnaires, and 24-hour dietary recall — are time-consuming, burdensome, and prone to memory-based bias. Recent advances in artificial intelligence, computer vision, and deep learning have introduced automated, image-based alternatives, in which a photograph of a meal, rather than a verbal or written description, becomes the primary data source for dietary assessment. This paper reviews and synthesizes recent research on image-based food recognition, food segmentation, portion and volume estimation, calorie and nutrient calculation, and personalized dietary recommendation, drawing on a methodological review of classification and volume-estimation algorithms, a deep-learning pipeline study covering dataset construction and nutrient-database matching, and an applied case system (IntelligentDine) that fine-tunes a 150-layer ResNet on the Food-101 dataset to jointly perform food identification and calorie regression, reporting 97.30% training accuracy and 98.02% validation accuracy. The reviewed studies show that convolutional neural networks, transfer-learning architectures such as ResNet and EfficientNet, and benchmark datasets such as Food-101 have substantially improved food-classification performance, while portion and volume estimation, dataset diversity (particularly for Indian and other regional cuisines), real-world robustness, privacy, and mobile deployability remain comparatively unresolved. Building on these findings, this paper proposes a unified conceptual framework that integrates image acquisition, preprocessing, food detection and classification, portion/volume estimation, nutrient-database lookup, and personalized health-based recommendation into a single intelligent dietary-assessment system, and outlines directions for future research toward accurate, real-time, and personalized dietary guidance.

Coordinated Neural Dynamics of Sleep-Dependent Memory Stabilization: Integrating Replay, Oscillatory Coupling, and Synaptic Regulation

Rimmy — 2026-09-24

Memory consolidation during sleep depends on complex interactions of neural oscillations, brain networks, and molecular machinery. Here, a structured narrative review of the neural mechanisms of sleep-dependent memory consolidation is presented and analyzed, concentrating on neural replay, hippocampal sharp-wave ripples (SWRs) and thalamocortical sleep spindles and cortical slow oscillations (SOs), and on hippocampo-neocortical communication, Synaptic Homeostasis, and neurochemical regulation. Research articles from January 2003 to September 2026 were retrieved through Pub Med/MEDLINE, Scopus, Web of Science, ScienceDirect, and IEEE Xplore and analyzed. Evidence from experimental studies involving humans and animals, including relevant methods, was thematically synthesized and comparatively evaluated with respect to mechanisms, study protocols, and the validity and limitations of the existing evidence and investigations. The literature consistently reveals that sleep-dependent replay and distribution of hippocampal-dependent memories during NREM sleep are critically coupled to interactions between hippocampal SWRs and sleep spindles and SOs, fitting the primary propositions of Active Systems Consolidation (ASC) theory. Specifically, neural replay and temporal coupling of oscillations may enhance hippocampo-neocortical communication, contributing to the persistence of encoded traces. Meanwhile, the hypothesis on Synaptic Homeostasis posits, with a functional approach for long-lasting potentiation, that synaptic upscaling, under sleep, might preserve only the salient memory engrams. Moreover, REM sleep could contribute significantly to consolidation of emotional memory as well as integration of the newly stored memory, in a manner not clearly determined for others. All these aspects are modulated through several neurochemical systems, mainly consisting of acetylcholine, glutamate, GABA, dopamine, and cortisol. Methodological advances such as intracranial EEG recording, neuroimaging techniques, targeted reactivation of memories, and non-invasive brain stimulation may confirm the hypotheses about the neural mechanisms. However, methodological questions remain, including differences between human and animal studies, a broad range of methodologies, many correlative studies in humans, and inconclusive evidence of causality, regarding neuronal replay and the coordination of oscillations with synaptic homeostasis and memory maintenance. Overall, the studies confirm that sleep is a very active state in which coordination across neural elements is maintained and that this process enhances not only memory maintenance but selectively modifies previously stored representations, integrating new information with existing information at the systemic level.

Benchmarking Machine-Learning Classifiers for Heavy-Rainfall Prediction from Monthly Station Data in Bangladesh: A First Step Toward Flood-Risk Assessment for Dhaka

Mahir Labib — 2026-10-14

Flooding and extreme rainfall pose recurrent risks to Bangladesh, and Dhaka in particular, where unplanned urbanization, drainage channel encroachment, and intensifying monsoon rainfall compound exposure. Reproducible, data-driven tools are needed to complement resource-intensive hydrodynamic models and qualitative surveys. This study benchmarks four machine-learning classifiers (logistic regression, Random Forest, XGBoost, and a sequence-based LSTM) for predicting heavy-rainfall months from monthly meteorological records of 31 stations across Bangladesh (1950–2022), as a first stage toward flood-risk assessment for Dhaka. A month was labelled heavy if the following month's rainfall exceeded the station's 75th percentile, and zero-filled rainfall records (21% of station-years) were removed beforehand. Models were trained on 1950–1999, tuned on 2000–2009, and tested on 2010–2022 (4,805 samples, 24% positive). All four classifiers reached ROC AUC values of 0.906–0.911 (LSTM 0.911, XGBoost 0.910, Random Forest 0.908, logistic regression 0.906), compared with 0.50 for a majority-class baseline. However, a seasonal climatology baseline that uses only station and calendar month reached 0.912, and no model exceeded it on AUC, PR AUC or F1. These results indicate that, at monthly resolution, three months of station-level meteorological history adds little beyond the monsoon cycle for predicting heavy-rainfall months, and that observed flood labels and terrain, drainage, and land-cover data would be needed for reliable flood-risk mapping.

Zero-Shot Anomaly Detection via Self-Supervised Vision Transformer and Hybrid Density Modeling

Kamran Khan — 2026-10-17

Pedestrian walkway surveillance requires robust anomaly detection for intelligent video analysis systems and public safety. Traditional supervised methods require labelled data for anomalous events, a costly, impractical expectation because anomalies are rare, unpredictable, and difficult to capture. This study proposes a zero-shot anomaly detection framework based on self-supervised Vision Transformers (ViT) that learns from normal data only. The insight is that an accurate representation of the normal is good enough to extend to anomalies, i.e., it is generalizable without training on the anomaly. The framework merges rotation prediction and contrastive learning objectives, yielding rich and generalizable representations of normal activity. For scoring anomalies, a hybrid detection strategy is used where Mahalanobis distance captures global distribution-based anomalies, and Local Outlier Factor (LOF) captures local density-based anomalies using a weighted fusion mechanism that incorporates these normalized scores into a weighted sum improves reliability. The attention maps from the ViT demonstrate interpretability in the localization of anomalies, revealing where unusual activity is happening in video frames. The effectiveness of this framework is demonstrated through extensive experiments on a synthetic pedestrian dataset and benchmark surveillance datasets, namely UCSDPed2 and ShanghaiTech, achieving AUROC scores of 0.9385, 0.968 and 0.742, respectively. The proposed framework outperforms those of supervised baselines, reconstruction-based methods, and zero-shot frameworks. Additionally, because only normal data is used for training the framework, it reduces annotation burden and makes anomaly detection feasible for real-world deployment in urban surveillance scenarios with better interpretability and efficiency.