Skip to main content

Advancements in Smart Diagnostics for Understanding Neurological Behaviors and Biosensing Applications

Cover image for Research Topic "Advancements in Smart Diagnostics for Understanding Neurological Behaviors and Biosensing Applications"
70kviews
Editorial16 September 2025
Editorial: Advancements in smart diagnostics for understanding neurological behaviors and biosensing applications

illustrating how multimodal fusion and ensemble strategies can extract complementary neural and ocular markers of preference in ecologically valid scenarios. Their approach achieved around 84% accuracy with strong precision on positive preferences. Extending behavioral inference to social cognition, Bhutta et al. employ frontal fNIRS in an interactive setting to distinguish deception from truth-telling using deep neural networks, attaining approximately 88-90% accuracy and pointing to the feasibility of decoding complex, spontaneous behaviors beyond controlled paradigms. Sleep health emerges as a second thematic pillar that benefits from multimodal, temporally aware modeling across distinct populations. A contribution on sleep staging by Fan et al. transforms EEG and EOG into time-frequency sequences, couples long-range temporal modeling with multiscale feature extraction, and integrates modalities to mitigate heterogeneity introduced by obstructive sleep apnea (OSA). This design demonstrates broader applicability beyond healthy cohorts and enhances interpretability for clinical workflows, with performance around 80% in OSA cohorts and improvements over competitive baselines on public datasets. Complementing this system's view, a neonatal study by Siddiqa et al. identifies promising electrode configurations and informative signal features that sustain accurate sleep state classification. Notably, a single central channel maintained around 81% accuracy, and compact left hemisphere montages slightly outperformed right hemisphere channels. Study presented practical sleep monitoring strategy that prioritizes comfort, safety, and computational efficiency in newborn care. Methodological innovation in EEG decoding is highlighted by work that blends efficient sequence modeling with targeted attention and multiscale feature design. A study by Zhe Li advances a compact state space architecture paired with pyramidal convolutions and channel-spatial attention to improve EEG classification for brain-computer interfaces (BCI). It underscores the potential of latency-conscious designs for real-time use on a standard dataset while achieving about 97% performance with strong class-wise balance. In parallel, a clinical study by Umair et al. on major depressive disorder (MDD) detection demonstrates that ensembles leveraging transformer representations can deliver high-accuracy classification from EEG while operating within a decentralized, split-learning framework that keeps data local across nodes. The approach maintained over 95% accuracy across clients, and reached around 99% in centralized settings, aligning with institutional privacy requirements and offering a viable path to collaborative, ensemble learning without compromising data security. Beyond electrophysiology, imaging-centric contributions emphasize both diagnostic capability and pipeline security. A large-scale study on karyogram analysis by Tabassum et al. proposes a hybrid approach that pretrains the proposed classifier on unlabeled images and fine-tunes it to detect structural anomalies, complemented by techniques that localize abnormal regions. This design addresses the pervasive challenge of rarely labeled anomalies and demonstrates near state-of-the-art accuracy (about 99%), supporting early screening for chromosome-related neurodegenerative disorders with neurological impact. Complementing analytics with protection, Asiri et al. develop a lightweight bit-plane encryption scheme for medical images tailored to the internet of things (IoT) and edge devices. By leveraging chaotic map-based shuffling and diffusion, the method achieved high entropy (greater than 7.98), low or negative inter-pixel correlations, a vast key space, and robustness under occlusion. This study presented practical safeguards for data in transit and at rest in resource-constrained environments. Lastly, novel sensing modalities extend the scope of smart diagnostics to child-centered interaction and rehabilitation. Asahi et al. employ EIT-based tactile sensing, which presents a safe, integrated device that classifies children's power versus precision grips using features derived from voltage patterns and tomographic reconstructions. In a pediatric cohort, accuracies exceeded 85%, illustrating how contact-rich sensing can enable quantitative monitoring of developing motor skills and signify the design of pediatric HCI. Several cross-disciplinary and cutting-edge research focuses were explored in this research topic. First, multimodal fusion and multiscale representations consistently improve robustness to artifacts and population heterogeneity when EEG, EOG, and ET modalities are hybridized or link time-frequency transforms with attention and dilated convolutions. Second, temporal sequence modeling via transformers and state-space formulations captures long-range dependencies that static models miss, enabling more reliable decoding of stress, behavior, and sleep dynamics. Third, data-efficiency strategies, including unsupervised pretraining, synthetic data augmentation for class balancing with the synthetic minority over-sampling technique (SMOTE), and targeted feature engineering, addressed unlabeled data and imbalanced class distribution issues, which are common in clinical datasets and rare pathology scenarios. Fourth, interpretability is increasingly embedded through attention mechanisms, multiscale modules, and explicit localization, aligning model outputs with physiological prospects and aiding clinical acceptance. Finally, trustworthy deployment is advanced by privacy-preserving learning that limits data movement, lightweight encryption suited to edge devices, and practical design choices such as electrode optimization and a compact scheme that supports real-time, on-device feasibility. Conclusion: The presented research contributions validate a promising shift from individual performance gains to cutting-edge integrated pipelines that are multimodal, interpretable, cybersecure, and privacy-preserving by design. They demonstrate that modern sequence models and multiscale representations can decode complex neurobehavioral characteristics of active and passive brain activities (e.g., stress, consumer choice, deception, and sleep dynamics). It has been presented that data-efficient training enables intricate neuronal signatures to be captured efficiently (e.g., neonatal EEG and rare chromosomal anomalies). The contributions established that privacy-preserving analytics and lightweight cryptography are functional aspects for deployment in clinics and daily life (e.g., split learning sustaining over 95% accuracy, edge encryption with high entropy and resilience). To sustain this momentum, the research community should prioritize prospective and a wide range of demographic validation to establish generalization and to adopt shared data standards and benchmarks to strengthen reproducibility. Collaborative explainability should be adopted with clinicians, patients, and end users to support informed decisions. Decentralized learning and secure edge computing should continue to excel for equitable access. Convergence of ML and biosensing approaches, along with these research commitments, will continue to deliver reliable neurotechnology for diagnosis, monitoring, and HCI across the lifespan.

2,613 views

1 citations

Multimodal sleep staging network based on obstructive sleep apnea
ORIGINAL RESEARCH
18 December 2024
Multimodal sleep staging network based on obstructive sleep apnea
Jingxin Fan, 5 more and Xiaoling Peng
BackgroundAutomatic sleep staging is essential for assessing sleep quality and diagnosing sleep disorders. While previous research has achieved high classification performance, most current sleep staging networks have only been validated in healthy populations, ignoring the impact of Obstructive Sleep Apnea (OSA) on sleep stage classification. In addition, it remains challenging to effectively improve the fine-grained detection of polysomnography (PSG) and capture multi-scale transitions between sleep stages. Therefore, a more widely applicable network is needed for sleep staging.MethodsThis paper introduces MSDC-SSNet, a novel deep learning network for automatic sleep stage classification. MSDC-SSNet transforms two channels of electroencephalogram (EEG) and one channel of electrooculogram (EOG) signals into time-frequency representations to obtain feature sequences at different temporal and frequency scales. An improved Transformer encoder architecture ensures temporal consistency and effectively captures long-term dependencies in EEG and EOG signals. The Multi-Scale Feature Extraction Module (MFEM) employs convolutional layers with varying dilation rates to capture spatial patterns from fine to coarse granularity. It adaptively fuses the weights of features to enhance the robustness of the model. Finally, multiple channel data are integrated to address the heterogeneity between different modalities effectively and alleviate the impact of OSA on sleep stages.ResultsWe evaluated MSDC-SSNet on three public datasets and our collection of PSG records of 17 OSA patients. It achieved an accuracy of 80.4% on the OSA dataset. It also outperformed the state-of-the-art methods in terms of accuracy, F1 score, and Cohen's Kappa coefficient on the remaining three datasets.ConclusionThe MSDC-SSRNet multi-channel sleep staging architecture proposed in this study enhances widespread system applicability by supplementing inter-channel features. It employs multi-scale attention to extract transition rules between sleep stages and effectively integrates multimodal information. Our method address the limitations of single-channel approaches, enhancing interpretability for clinical applications.

4,819 views

12 citations

Artificial neural network models: implementation of functional near-infrared spectroscopy-based spontaneous lie detection in an interactive scenario
Deception is an inevitable occurrence in daily life. Various methods have been used to understand the mechanisms underlying brain deception. Moreover, numerous efforts have been undertaken to detect deception and truth-telling. Functional near-infrared spectroscopy (fNIRS) has great potential for neurological applications compared with other state-of-the-art methods. Therefore, an fNIRS-based spontaneous lie detection model was used in the present study. We interviewed ten healthy subjects to identify deception using the fNIRS system. A card game frequently referred to as a bluff or cheat was introduced. This game was selected because its rules are ideal for testing our hypotheses. The optical probe of the fNIRS was placed on the subject's forehead, and we acquired optical density signals, which were then converted into oxy-hemoglobin and deoxy-hemoglobin signals using the Modified Beer–Lambert law. The oxy-hemoglobin signal was preprocessed to eliminate noise. In this study, we proposed three artificial neural networks inspired by deep learning models, including AlexNet, ResNet, and GoogleNet, to classify deception and truth-telling. The proposed models achieved accuracies of 88.5%, 88.0%, and 90.0%, respectively. These proposed models were compared with other classification models, including k-nearest neighbor, linear support vector machines (SVM), quadratic SVM, cubic SVM, simple decision trees, and complex decision trees. These comparisons showed that the proposed models performed better than the other state-of-the-art methods.

6,505 views

12 citations