hybrid-deep-reinforcement-learning-for-multimodal-biomedical-signal

Hybrid Deep Reinforcement Learning for Multimodal Biomedical Signal Fusion

Improving decision-making and diagnostic accuracy in health care requires combining multimodal biomedical signals such as electrocardiograms (ECG), ...

University of Belgrade

Choosing a comprehensive and cost-effective way of articulating and annotating the sentiment of a text is not a trivial task, particularly when dealing with short texts, in which sentiment can be expressed through a wide variety of linguistic and rhetorical phenomena. This problem is especially conspicuous in resource-limited settings and languages, where design options are restricted either in terms of manpower and financial means required to produce appropriate sentiment analysis resources, or in terms of available language tools, or both. In this paper, we present a versatile approach to addressing this issue, based on multiple interpretations of sentiment labels that encode information regarding the polarity, subjectivity, and ambiguity of a text, as well as the presence of sarcasm or a mixture of sentiments. We demonstrate its use on Serbian, a resource-limited language, via the creation of a main sentiment analysis dataset focused on movie comments, and two smaller datasets belonging to the movie and book domains. In addition to measuring the quality of the annotation process, we propose a novel metric to validate its cost-effectiveness. Finally, the practicality of our approach is further validated by training, evaluating, and determining the optimal configurations of several different kinds of machine-learning models on a range of sentiment classification tasks using the produced dataset.

A versatile framework for resource-limited sentiment articulation, annotation, and analysis of short texts.

In this paper, the influence of the variation in transformer oil temperature on the accuracy of the all-acoustic non-iterative method for partial discharge location in a power transformer is researched. The research can improve power transformers' testing and monitoring, particularly given the large transformer oil temperature variations during real-time monitoring. The research is based on quantifying the contribution of oil temperature to the standard combined measurement uncertainty of the non-iterative algorithm by using analytical, statistical, and Monte Carlo methods. The contribution can be quantified and controlled. The contribution varied significantly with different mutual placements of partial discharge and acoustic sensors. The correlation between the contribution and the mean distance between partial discharge and acoustic sensors was observed. Based on these findings, the procedure to quantify and control the contribution in practice was proposed. The procedure considers the specificity of the method's mathematical model (the assumption that the oil temperature is constant), the non-iterative algorithm's nonlinearity, and the large variations in transformer oil temperature. Existing studies did not consider the significant effect of the oil temperature on the combined measurement uncertainty of partial discharge location influenced by those phenomena. The research is limited to partial discharge located in the transformer oil.

Impact of Power Transformer Oil-Temperature on the Measurement Uncertainty of All-Acoustic Non-Iterative Partial Discharge Location.

Wall segmentation is a special case of semantic segmentation, and the task is to classify each pixel into one of two classes: wall and no-wall. The segmentation model returns a mask showing where objects like windows and furniture are located, as well as walls. This article proposes the module's structure for semantic segmentation of walls in 2D images, which can effectively address the problem of wall segmentation. The proposed model achieved higher accuracy and faster execution than other solutions. An encoder-decoder architecture of the segmentation module was used. Dilated ResNet50/101 network was used as an encoder, representing ResNet50/101 network in which dilated convolutional layers replaced the last convolutional layers. The ADE20K dataset subset containing only interior images, was used for model training, while only its subset was used for model evaluation. Three different approaches to model training were analyzed in the research. On the validation dataset, the best approach based on the proposed structure with the ResNet101 network resulted in an average accuracy at the pixel level of 92.13% and an intersection over union (IoU) of 72.58%. Moreover, all proposed approaches can be applied to recognize other objects in the image to solve specific tasks.

Wall segmentation in 2D images using convolutional neural networks.

Internet of Things (IoT) is one of the most important emerging technologies that supports Metaverse integrating process, by enabling smooth data transfer among physical and virtual domains. Integrating sensor devices, wearables, and smart gadgets into Metaverse environment enables IoT to deepen interactions and enhance immersion, both crucial for a completely integrated, data-driven Metaverse. Nevertheless, because IoT devices are often built with minimal hardware and are connected to the Internet, they are highly susceptible to different types of cyberattacks, presenting a significant security problem for maintaining a secure infrastructure. Conventional security techniques have difficulty countering these evolving threats, highlighting the need for adaptive solutions powered by artificial intelligence (AI). This work seeks to improve trust and security in IoT edge devices integrated in to the Metaverse. This study revolves around hybrid framework that combines convolutional neural networks (CNN) and machine learning (ML) classifying models, like categorical boosting (CatBoost) and light gradient-boosting machine (LightGBM), further optimized through metaheuristics optimizers for leveraged performance. A two-leveled architecture was designed to manage intricate data, enabling the detection and classification of attacks within IoT networks. A thorough analysis utilizing a real-world IoT network attacks dataset validates the proposed architecture's efficacy in identification of the specific variants of malevolent assaults, that is a classic multi-class classification challenge. Three experiments were executed utilizing data open to public, where the top models attained a supreme accuracy of 99.83% for multi-class classification. Additionally, explainable AI methods offered valuable supplementary insights into the model's decision-making process, supporting future data collection efforts and enhancing security of these systems.

Intrusion detection in metaverse environment internet of things systems by metaheuristics tuned two level framework.

Despite advancements in modern healthcare, diabetes mellitus remains a lifelong, incurable condition. Empowering patients through health education and self-management is essential in preventing disease progression. This study evaluates the effectiveness of My Diabetes Care, a mobile application featuring an animated conversational agent, Dia-vera, designed to support diabetes self-managementat home. Focusing on non-compliance behaviors, sedentary lifestyle, and uncontrolled HbA1c levels, data were collected from 200 purposively selected participants from rural health clinics in southern Pakistan. This study used artificial intelligence models with built-in explainability features applied to artificial neural networks, achieving 98% training accuracy and 95% testing accuracy. User-chatbot dialogues were analyzed for engagement, thematic queries, fallback responses, and silence periods. Dia-vera successfully answered 88.86% of the 2830 queries. Weekly dialogue averages dropped from 36 to 26.1 between study phases, providing insights for future refinement. High levels of participant acceptability and satisfaction were found using the System Usability Scale. The findings show that, especially in disadvantaged settings, integrating interpretable AI with conversational agents provides a user-friendly and scientifically supported method of diabetes self-managementassistance.In comparison to baseline, participants who used the intervention reported better adherence to medication and food regimens, showed increased involvement in physical activity, and showed small reductions in HbA1c levels. These results make the study's therapeutic relevance stronger and show a stronger connection between the intervention and the desired health behaviors. Using My Diabetes Care as a proof-of-concept implementation, this study offers a reproducible framework for creating intelligent, explainable digital health interventions.

My diabetes care: an AI-based mobile app with conversational agent for type 2 diabetes self-management.

Photovoltaic pumping systems have become a key solution for sustainable water supply, especially in remote and off-grid areas. Yet, their performance often drops under changing solar conditions. To address this, we introduce Spider-Tailed Horned Viper Optimization (STHVO), a novel nature-inspired MPPT technique specifically designed for such applications. A PV generator, a step-up converter, and a radial-flow pump powered by an induction motor are all part of the suggested configuration. The system was subsequently tested under standard irradiance (1000 W/m²) and real-world irradiance variations obtained from the Bni Hadifa region. By achieving 98.92% efficiency, delivering a peak hydraulic power of 72 W, and sustaining a steady 0.65 L/s flow rate, simulation results demonstrate that STHVO performs better than traditional tactics. It also ensures rapid tracking in less than 0.6 s and constant motor speed at 195 rad/s. These outcomes show how the method can enhance solar-powered systems' energy reliability and water delivery.

Bioinspired STHVO based MPPT control for grid connected photovoltaic water pumping systems.

Heart diseases (CVDs) are a major cause of morbidity and mortality in all global regions and thus there is the pressing need to develop early detection and effective management approaches. Traditional cardiovascular monitoring systems do not necessarily have real-time analyzing solutions and individual understanding, which leads to delayed interventions. Moreover, one of the greatest issues in digital healthcare applications remains to be data privacy and security.

Article	Year
A versatile framework for resource-limited sentiment articulation, annotation, and analysis of short texts. PloS one\| PubMed ID: 33180825	2020
Impact of Power Transformer Oil-Temperature on the Measurement Uncertainty of All-Acoustic Non-Iterative Partial Discharge Location. Materials (Basel, Switzerland)\| PubMed ID: 33809223	2021
Wall segmentation in 2D images using convolutional neural networks. PeerJ. Computer science\| PubMed ID: 37810356	2023
Intrusion detection in metaverse environment internet of things systems by metaheuristics tuned two level framework. Scientific reports\| PubMed ID: 39875592	2025
My diabetes care: an AI-based mobile app with conversational agent for type 2 diabetes self-management. Scientific reports\| PubMed ID: 41249394	2025
Bioinspired STHVO based MPPT control for grid connected photovoltaic water pumping systems. Scientific reports\| PubMed ID: 41507379	2026
Triboostcardio ensemble model for cardiovascular disease detection using advanced blockchain-enabled health monitoring. Frontiers in artificial intelligence\| PubMed ID: 41626575	2025

Bosko Nikolic

Biography

JoVE Journal Publications

Other Publications