Published in 2026
Sandhya Yamala
Heart disorders are the leading cause of mortality worldwide. Around 80% of fatalities occurred in poor and middleincome nations. If present trends continue, it is anticipated that 23.6 million people would die from cardiovascular disease by 2030 (mostly as a result of heart attacks and strokes). With the increasing popularity of technologically advanced wearable devices, there is a greater chance to deliver an Internet of Things (IoT) solution. Regrettably, for persons who are suffering from an unexpected cardiac arrest, out-of-hospital survival chances are dismal. Heart rate, fluctuation in heart rate, changes in temperature of the body, Sugar levels, Blood pressure, and oxygen levels in the blood are important indicators that need to be checked on a regular basis in those who have heart disease. The objective of this research is to implement a smartphone application that can monitor these metrics for cardiac patients who need ongoing monitoring. The monitoring system employs wearable sensors to continually assess numerous properties. When the patients' predefined parameters reach the limit range then an alert message or email will be delivered to a doctor or any close family member. The heart rate module's data may be saved and retrieved in the future for medical use. Because of this wearable technology, patients may be mobile within a personal social setting, allowing them to lead their lives with assurance.
Internet of things, Heart Disorders, Sensors, Oxygen Levels, blood sugar, PPG signal, Arduino controller
R. Keerthigadevi, K. Santhanalakshmi
Communication may greatly be affected by speech and language disorders and therefore, early and proper diagnosis of the condition is the key to effective clinical treatment. The current paper introduces a single artificial intelligence system used in pathological speech classification and improvement through the combination of cross-modal learning and graph-based relational modeling. These two modalities are combined in the proposed system with the help of a cross-modal transformer which identifies the contextual dependencies between the modalities with the help of attention mechanisms. The model uses self-supervised pre training to enhance the quality of representation in a situation where limited labeled data are available and thus it is able to learn both generalized and discriminative speech features. Besides that, a graph neural network is used to model structural dependencies between speech segments, with nodes and edges respectively modeling feature embeddings and temporal continuity and phonetic similarity. This two-sided representation enables the framework to share the analysis of local variations and international speech patterns linked to such disorders like dysarthria and Parkinsonian speech. Moreover, a speech enhancement module is presented that is lightweight and helps to reduce noise interference and to enhance the quality of input signals, thus, boosting the performance of the downstream classification. Both cross-modal transformers and graph neural networks coupled to each other lead to a higher level of robustness, feature discrimination, and interpretability. According to the experimental evidence, the given approach is more effective than the
traditional ones, especially in noisy and low-resource conditions.
Speech disorder classification, cross-modal transformer, graph neural networks, self-supervised learning, speech enhancement, multimodal learning, dysarthria, acoustic features, linguistic features, pathological speech analysis
V. Priya, Preethi S
The issue of neonatal mortality prediction continues to be a thorn in the flesh of healthcare since the numerous interconnected factors of clinical, demographic, and imaging-related variables have an effect on infant outcomes. Risk assessment is a critical issue that requires timely intervention through early and accurate risk assessment especially in the neonatal intensive care unit where decisions have to be made in times of uncertainty. In the current study, it is suggested to use an EfficientNet-based multimodal fusion model combined with an artificial intelligence-based clinical model of predicting risks and mortality in neonatal disease prevention. The framework integrates profound visual representations of medical images of neonatal cases with the help of EfficientNet and represented clinical data (birth weight, gestational age, APGAR scores, and laboratory measurements). The use of a feature fusion approach can combine heterogeneous data sources to allow the model to model both spatial patterns as imaged and contextual relationships as available in tabular clinical attributes. Before the model training, the data are preprocessed through normalization, missing values, and features selection to enhance the quality of data and robustness of the model. The merged feature representation is then classified in a predictive model based on AI which is optimized to perform optimally. The experimental analysis proves that the suggested multimodal model with an experimental evaluation shows a better predictive ability than one-modality models based on the effective utilization of complementary information. The findings suggest that EfficientNet combined with clinical data modeling is a reliable and scalable decision-support model of neonatal outcome prediction in clinical practice.
Neonatal mortality prediction, EfficientNet, multimodal learning, feature fusion, artificial intelligence, medical imaging, clinical data, deep learning, healthcare analytics, decision support system.
M. Indhumathi, D. Banumathy
Predicting Bitcoin prices is inherently difficult due to the market’s highly volatile nature, nonlinear behavior, and sensitivity to global financial conditions. Traditional statistical models often fail to capture complex patterns, while standalone deep learning approaches may overlook underlying linear structures. To address these limitations, this paper presents a hybrid forecasting framework that combines ARIMA, Long Short-Term Memory (LSTM), and an attention mechanism in a unified, residual-driven architecture. Initially, the ARIMA model is applied to identify and model the linear components of the time-series data after ensuring stationarity. The residuals obtained from this stage, which represent nonlinear and unexplained variations, are then used as input to an LSTM network to learn long-term dependencies. Furthermore, an attention layer is incorporated to assign adaptive importance to different time steps, allowing the model to focus on the most influential historical patterns. This integrated approach enhances both prediction accuracy and interpretability, particularly under unstable market conditions. Experimental evaluation using standard performance metrics such as RMSE, MAE, and directional accuracy demonstrates that the proposed model consistently outperforms individual and baseline methods. The results highlight the effectiveness of combining statistical and deep learning techniques for robust cryptocurrency price forecasting.
Bitcoin price prediction, ARIMA, LSTM, attention mechanism, hybrid model, time-series forecasting, volatility analysis, and financial data modeling
Dr. D Banumathy, V Maheskumar, G Sanjay
The rapid expansion of digital infrastructure has significantly increased the complexity and frequency of cyber threats, necessitating advanced approaches for accurate and timely attack prediction. Traditional machine learning techniques have been extensively employed to analyze network traffic and detect anomalies based on historical data, offering reliable performance in identifying known attack patterns; however, their dependence on predefined features and labelled datasets often limits their ability to generalize emerging and sophisticated threats. In contrast, recent advancements in generative artificial intelligence introduce a more adaptive paradigm by enabling models to learn underlying data distributions and generate realistic representations of potential attack scenarios, thereby enhancing the capability to anticipate previously unseen vulnerabilities. This paper examines the evolution from conventional machine learning methods to generative AI-driven frameworks for cyber-attack prediction, emphasizing their comparative strengths and the potential for integrated approaches to improve the robustness, adaptability, and overall effectiveness of modern cybersecurity systems in dynamic environments.
Cyber Attack Prediction, Machine Learning, Generative Artificial Intelligence, Anomaly Detection, Network Security, Deep Learning, Threat Intelligence.
S. Janani, Dr. S.S. Suganya
This study presents a multi-stage hybrid deep learning and optimization approach to predict and classify the Water Quality Index (WQI). The goal was to identify a single process to create a model pipeline that will allow for the use of both feature selection and the use of deep learning combined with metaheuristics in the assessment of water quality. The objective of this research was to develop a 3-stage predictive model using Relief-GWO feature selection, optimized using ANN/DANN with ISLO and DNN using PSO or CCPSO or a Chaos-Driven Batalgorithm. Eighteen physical, chemical, and biological characteristics of water from 3,673 sampling sites were studied. The results obtained by the baseline models developed in Phase 1 showed that SVM was the best classifier at a rate of 91.29%.Phase 2 hybrid ANN-Maxout obtained an overall classification accuracy of 93.74%, whereas DANN-ISLO achieved a better regression R² value of 0.9806 than all other models. In phase 3, the DNN-CCPSO model had the best performance for regression as measured by the R² value of 0.9813, and DNN-ChaosBat was the best in terms of F1-Macro at 0.7269. Overall, this study demonstrated that a proposed multi-hybrid approach will provide superior results to many individual models used alone, providing significant improvements over time throughout each phase. The work presented here demonstrates how the use of artificial intelligence can be applied to support automatic water quality monitoring, detect contamination early on, and support governmentdecision-making relative to environmental policies.
Deep learning, Water Quality Index, Relief-GWO feature selection, Metaheuristic optimization, Hybrid neural networks