Sensors

Research

18 pages, 1190 KB

Open AccessArticle

Fall-from-Bed Risk Prediction Using Physics-Based Bed Simulation

by Jaeyong Kim, Hyeonwoo Kim, Jihwan Won, Jiwoon Lee, Hyeonjung Kim, Sunwoo Yeon, Ryanghee Sohn, Youngho Cho and Cheolsoo Park

Sensors 2026, 26(10), 2979; https://doi.org/10.3390/s26102979 - 9 May 2026

Viewed by 296

Abstract

Fall-from-bed is a critical safety issue in hospitals and long-term care; however, large-scale real-fall data are rare, and collecting the data is ethically constrained. This study examined whether the fall-from-bed risk can be inferred from single static in-bed postures without temporal motion. We [...] Read more.

Fall-from-bed is a critical safety issue in hospitals and long-term care; however, large-scale real-fall data are rare, and collecting the data is ethically constrained. This study examined whether the fall-from-bed risk can be inferred from single static in-bed postures without temporal motion. We developed a physics-based bed–human simulator in MuJoCo and generated labeled episodes by sampling diverse initial configurations, rolling out uncontrolled dynamics for three seconds, and detecting falls by floor contact. Each initial state was represented as a 13-keypoint 2D skeleton in a bed-centric coordinate frame, normalized to fixed bed bounds, and supervised with a continuous risk label derived from time-to-fall using per-frame discounting on a 30 frame-per-second grid. On a pose-balanced simulated test set of 50,000 initial postures, the best-performing multilayer-perceptron-based predictor attained an area under the receiver operating characteristic curve of 0.9755, area under the precision–recall curve of 0.9771, F1-score of 0.9138, and mean squared error of 0.0374 (mean over five random seeds). Pose-stratified initialization improved performance relative to fully random sampling. Consistently high performance was observed across supine/prone/lateral subgroups, which improved with training set size. These results suggest that a static posture contains predictive information about fall risk under matched simulator dynamics, supporting the feasibility of posture-based risk scoring in the controlled settings. Full article

(This article belongs to the Special Issue AI-Enabled Sensing Technology for Smart Healthcare and Precision Diagnosis)

► Show Figures

Figure 1

26 pages, 678 KB

Open AccessArticle

Evaluating the Adversarial Robustness and Clinical Safety of Quantized Hierarchical Transformers for Edge-Based Malaria Microscopy

by Umar Hasan, Turki G. Alghamdi and Muhammad Ali Nayeem

Sensors 2026, 26(9), 2888; https://doi.org/10.3390/s26092888 - 5 May 2026

Viewed by 983

Abstract

Automated mobile microscopy in Internet of Things (IoT) networks is essential for scaling malaria screening in resource-constrained environments. Deploying standard convolutional architectures here introduces severe adversarial vulnerabilities. Post-Training Quantization (PTQ) mitigates hardware constraints by converting floating-point models to 8-bit integers (INT8); however, its [...] Read more.

Automated mobile microscopy in Internet of Things (IoT) networks is essential for scaling malaria screening in resource-constrained environments. Deploying standard convolutional architectures here introduces severe adversarial vulnerabilities. Post-Training Quantization (PTQ) mitigates hardware constraints by converting floating-point models to 8-bit integers (INT8); however, its impact on clinical safety and security remains unexplored. This study presents an adversarial audit of quantized Vision Transformers for medical edge deployment. We evaluated a Swin-Tiny transformer against ViT-Tiny and MobileNetV3 baselines using a 27,558-image malaria dataset and an out-of-distribution (OOD) White Blood Cell dataset. Our findings redefine the “Quantization Shield” hypothesis. PTQ compresses the Swin model by 3.9× (to 27.89 MB) with a negligible 0.11% accuracy drop, maintaining statistical reliability on OOD tests. However, the hypothesized architectural resilience shatters under white-box Projected Gradient Descent (PGD) attacks. Despite robustness against single-step attacks, both MobileNetV3 and the INT8 Swin-Tiny collapse to 0.00% accuracy under iterative PGD. Conversely, the quantized Swin-Tiny resists black-box transfer attacks from a surrogate, maintaining 81.00% accuracy. We conclude that while quantized Vision Transformers meet mobile sensor constraints, integer quantization provides zero innate defense against targeted iterative perturbations, exposing a critical vulnerability in diagnostic IoT networks. Full article

(This article belongs to the Special Issue AI-Enabled Sensing Technology for Smart Healthcare and Precision Diagnosis)

► Show Figures

Figure 1

24 pages, 5780 KB

Open AccessArticle

A Deep Learning-Guided Ensemble Empirical Mode Decomposition Method for Single-Channel Fetal Electrocardiogram Extraction

by Xiaojian Xu, Yifan Zhang, Yufei Rao, Yinru Xu, Yang Gao and Huating Tu

Sensors 2026, 26(7), 2037; https://doi.org/10.3390/s26072037 - 25 Mar 2026

Viewed by 498

Abstract

The fetal electrocardiogram (FECG) is critical for assessing fetal cardiac electrophysiology and detecting fetal distress and arrhythmias. Single-channel abdominal electrocardiogram (AECG) enables home-based monitoring but faces challenges posed by weak fetal signals, maternal interference, and the lack of spatial information. Ensemble Empirical Mode [...] Read more.

The fetal electrocardiogram (FECG) is critical for assessing fetal cardiac electrophysiology and detecting fetal distress and arrhythmias. Single-channel abdominal electrocardiogram (AECG) enables home-based monitoring but faces challenges posed by weak fetal signals, maternal interference, and the lack of spatial information. Ensemble Empirical Mode Decomposition (EEMD) is suitable for nonstationary AECG signals but relies on accurate selection of intrinsic mode functions (IMFs). In this study, a deep learning-guided method was proposed: a one-dimensional convolutional neural network (1D CNN) scored and selected EEMD-derived IMFs, followed by maternal QRS template subtraction and secondary EEMD purification to achieve automatic FECG extraction. Leave-one-subject-out (LOSO) cross-validation was performed on 15 simulated cases and 5 ADFECGDB records, yielding a mean AUC of 0.9282 ± 0.0189 for the IMF classifier. On the independent DaISy and NIFEA arrhythmia datasets, the proposed CNN-2×EEMD method achieved correlation coefficients of 0.94–0.96, F1-scores of 0.8372–0.9565 for fetal R-peak detection, and SNR improvements of 13.39–15.88 dB. This method outperformed conventional automatic selection methods and matched the performance of manual selection. Ablation studies validated the optimal network design and IMF selection strategy, while complexity analysis (0.08 GFLOPs, 2.24 ms latency) confirmed its suitability for real-time wearable deployment. Full article

(This article belongs to the Special Issue AI-Enabled Sensing Technology for Smart Healthcare and Precision Diagnosis)

► Show Figures

Figure 1

38 pages, 2058 KB

Open AccessArticle

AI-Enhanced Hybrid QAM–PPM Visible Light Communication for Body Area Networks

by Shreyash Shrestha, Attaphongse Taparugssanagorn, Stefano Caputo and Lorenzo Mucchi

Sensors 2026, 26(3), 971; https://doi.org/10.3390/s26030971 - 2 Feb 2026

Viewed by 687

Abstract

This paper investigates an artificial intelligence (AI)-enhanced visible light communication (VLC) system for body area networks (BANs) based on a hybrid modulation framework that jointly employs quadrature amplitude modulation (QAM) and pulse-position modulation (PPM). The dual-modulation strategy leverages the high spectral efficiency of [...] Read more.

This paper investigates an artificial intelligence (AI)-enhanced visible light communication (VLC) system for body area networks (BANs) based on a hybrid modulation framework that jointly employs quadrature amplitude modulation (QAM) and pulse-position modulation (PPM). The dual-modulation strategy leverages the high spectral efficiency of QAM together with the robustness of PPM to light-emitting diode (LED) nonlinearity and timing distortions, enabling simultaneous high-rate and reliable communication, two essential requirements in BAN applications. To address the nonlinear response of light-emitting diodes and the variability in indoor optical channels, the system integrates classical predistortion techniques with a deep learning equalizer combining convolutional neural network (CNN)–transformer layers. This hybrid model captures both local and long-range distortion patterns, improving symbol reconstruction for both modulation branches. The study further examines pilot-assisted equalization and adaptive bit loading, showing that these strategies strengthen link robustness under diverse channel conditions while enhancing spectral efficiency. The proposed architecture demonstrates that combining dual modulation with AI-driven equalization and adaptive transmission strategies leads to a more resilient and efficient VLC system, well-suited for the dynamic constraints of wearable and body-centric communication environments. Full article

(This article belongs to the Special Issue AI-Enabled Sensing Technology for Smart Healthcare and Precision Diagnosis)

► Show Figures

Figure 1

Review

Jump to: Research, Other

23 pages, 2268 KB

Open AccessReview

AI-Enabled Flexible Sensing Ecosystems for Parkinson’s Disease: Advancing Digital Biomarkers and Closed-Loop Interventions

by Jiadong Jin, Yongchang Jiang, Yukai Zhou, Wenkai Zhu, Jiangbo Hua, Wen Cheng, Yi Shi and Lijia Pan

Sensors 2026, 26(7), 2071; https://doi.org/10.3390/s26072071 - 26 Mar 2026

Viewed by 929

Abstract

Effective Parkinson’s disease (PD) management is hindered by the intermittent nature of clinical snapshots and the discomfort of rigid monitoring hardware. This review critically evaluates the synergy between flexible bioelectronics and artificial intelligence (AI) for continuous remote monitoring. Our analysis reveals that while [...] Read more.

Effective Parkinson’s disease (PD) management is hindered by the intermittent nature of clinical snapshots and the discomfort of rigid monitoring hardware. This review critically evaluates the synergy between flexible bioelectronics and artificial intelligence (AI) for continuous remote monitoring. Our analysis reveals that while material innovations have achieved milligram-level sensitivity, a significant ‘translational gap’ persists due to limited validation in real-world environments and small cohort sizes. We conclude that multimodal fusion architectures are essential for accurately mapping digital biomarkers to clinical gold standards such as MDS-UPDRS. By leveraging edge AI for privacy and closed-loop feedback for intervention, this integration facilitates the transition from reactive clinical visits to proactive, personalized digital home-care ecosystems. Full article

(This article belongs to the Special Issue AI-Enabled Sensing Technology for Smart Healthcare and Precision Diagnosis)

► Show Figures

Figure 1

Other

Jump to: Research, Review

11 pages, 481 KB

Open AccessProtocol

AI-Guided Remission: Protocol for a Conversational Agent (Chatbot) for Dosing Activity and Footwear Progression After Diabetic Limb Reconstruction

by Lucian M. Feraru, David C. Klonoff, Bijan Najafi, Magdalena Antoszewska and David G. Armstrong

Sensors 2026, 26(8), 2299; https://doi.org/10.3390/s26082299 - 8 Apr 2026

Viewed by 777

Abstract

Background: Diabetic foot ulcers recur frequently after healing. The first three months carry the highest risk. Remission is a vulnerable phase that demands precise self-care and timely feedback. Evidence supports thermometry and protective footwear with gradual return to activity, yet adherence at home [...] Read more.

Background: Diabetic foot ulcers recur frequently after healing. The first three months carry the highest risk. Remission is a vulnerable phase that demands precise self-care and timely feedback. Evidence supports thermometry and protective footwear with gradual return to activity, yet adherence at home is inconsistent. Objective: To describe the design and planned evaluation of a conversational agent (chatbot) that guides patients through the remission phase following diabetic limb reconstruction. Methods: This protocol describes a conversational agent (chatbot) that turns remission guidance into daily actions, grounded in clinical expertise and established care guidelines. Walking is dosed like a drug, with careful titration based on tissue response. The agent integrates automatic data capture (smartphone step counts, skin temperature, shoe step streams, smartwatch step streams, Bluetooth thermometry when available, and app session timestamps) with manual patient entries (shoe wear time, skin redness persistence, and symptom checks). It doses walking activity, guides footwear break-in, prompts photo-confirmed concerns, following clinician-informed rules and escalation pathways. We define data quality checks for missingness and physiologic plausibility, and the agent reinforces reducing weight-bearing activity when risk signals appear. We outline device drift. The study is designed as a single-arm feasibility pilot (n = 30) to assess engagement, safety, and implementation fidelity. Results: No clinical outcome results are reported because this is a protocol study and enrollment has not yet begun. This study presents the prespecified sensing-to-decision workflow, escalation logic, and pilot endpoints, along with internal technical verification procedures (e.g., message delivery reliability, data completeness checks, and rule-engine consistency testing). Conclusions: A remission chatbot is a plausible method to extend specialist support into the home, reflecting integration of clinical expertise with digital health tools. This protocol defines how feasibility, safety, and usability will be evaluated. Clinical efficacy should be confirmed in future studies. Full article

(This article belongs to the Special Issue AI-Enabled Sensing Technology for Smart Healthcare and Precision Diagnosis)

► Show Figures

Graphical abstract

54 pages, 2144 KB

Open AccessSystematic Review

Demystifying Artificial Intelligence: A Systematic Review of Explainable Artificial Intelligence in Medical Imaging

by Muhammad Fayaz, Kim Hagsong, Sufyan Danish, L. Minh Dang, Abolghasem Sadeghi-Niaraki and Hyeonjoon Moon

Sensors 2026, 26(7), 2131; https://doi.org/10.3390/s26072131 - 30 Mar 2026

Viewed by 975

Abstract

This comprehensive literature review explores the latest advancements in explainable artificial intelligence (XAI) techniques within the field of medical imaging (MI). Over the past decade, machine learning (ML) and deep learning (DL) technologies have made significant strides in healthcare, enabling advancements in tasks [...] Read more.

This comprehensive literature review explores the latest advancements in explainable artificial intelligence (XAI) techniques within the field of medical imaging (MI). Over the past decade, machine learning (ML) and deep learning (DL) technologies have made significant strides in healthcare, enabling advancements in tasks such as disease diagnosis, medical image segmentation, and the detection of various medical conditions. However, despite these successes, the widespread adoption of AI-driven tools in clinical practice remains slow, primarily due to the “black-box” nature of many AI models. These models make decisions without transparent reasoning, which poses significant barriers in critical medical and legal environments, where accountability and trust are paramount. This review investigates various XAI methods, focusing on both intrinsic and post-hoc techniques, to evaluate their potential in addressing these challenges. The paper examines how XAI can enhance the transparency of healthcare algorithms, thereby fostering greater trust and confidence among clinicians, patients, and regulators. Key challenges faced by XAI in healthcare, such as limited interpretability, computational complexity, and the absence of standardized evaluation frameworks, are discussed in detail. Furthermore, this work highlights existing gaps in the literature, including the lack of detailed comparative analyses of specific XAI techniques, especially in terms of their mathematical foundations and applicability across diverse medical imaging contexts. In response to these gaps, the paper introduces a new set of standardized evaluation metrics aimed at assessing XAI performance across various medical imaging tasks, such as image segmentation, classification, and diagnosis. The review proposes actionable recommendations for enhancing the effectiveness of XAI in healthcare, with a focus on real-world clinical applications. Unlike previous studies that focus on broader overviews or limited subsets of methods, this work provides a comprehensive comparative analysis of over 18 XAI techniques, emphasizing their strengths, weaknesses, and practical implications. By offering a detailed understanding of how XAI methods can be integrated into clinical workflows, this paper aims to bridge the gap between cutting-edge AI technologies and their practical use in medical settings. Ultimately, the insights provided are valuable for researchers, clinicians, and industry professionals, encouraging the adoption and standardization of XAI practices in clinical environments, thus ensuring the successful integration of transparent, interpretable, and reliable AI systems into healthcare. Full article

(This article belongs to the Special Issue AI-Enabled Sensing Technology for Smart Healthcare and Precision Diagnosis)

► Show Figures

Figure 1

21 pages, 2802 KB

Open AccessSystematic Review

Sensor-Based Technologies for the Detection of Unwanted Loneliness in Older Adults: A Systematic Review

by María Mercedes Párraga Vico, Juana María Morcillo Martínez, Juan F. Gaitán-Guerrero, Juan Luis Herreros Bódalo, Macarena Espinilla Estévez and Juan Carlos Cuevas Martínez

Sensors 2026, 26(7), 2028; https://doi.org/10.3390/s26072028 - 24 Mar 2026

Viewed by 850

Abstract

Background: Unwanted loneliness and social isolation in older adults are public health problems with negative effects on physical and mental health. The usual assessment tools, based on self-report questionnaires, have limitations in capturing these phenomena continuously and objectively. Objective: We aimed to [...] Read more.

Background: Unwanted loneliness and social isolation in older adults are public health problems with negative effects on physical and mental health. The usual assessment tools, based on self-report questionnaires, have limitations in capturing these phenomena continuously and objectively. Objective: We aimed to critically analyze recent scientific evidence on the use of passive sensor technologies combined with artificial intelligence for the detection of unwanted loneliness and social isolation in older adults. Methods: Studies were reviewed in databases (PubMed, Scopus, Web of Science, and IEEE Xplore) that used wearable devices, environmental sensors in the home, smartphones, and multimodal fusion approaches. This systematic review was conducted following the PRISMA 2020 guidelines. Results: Behavioral variables derived from passive monitoring, such as mobility, time away from home, sleep patterns, and digital interactions, are consistently associated with measures of loneliness and social isolation. Likewise, artificial intelligence models based on the combination of multiple data sources show better predictive performance than unimodal approaches. Conclusions: Sensor-based technologies can complement traditional assessment methods, although their practical application requires overcoming challenges related to methodological validation, user acceptance, and ethical considerations. Full article

(This article belongs to the Special Issue AI-Enabled Sensing Technology for Smart Healthcare and Precision Diagnosis)

► Show Figures

Figure 1

Journal Menu

Journal Browser

AI-Enabled Sensing Technology for Smart Healthcare and Precision Diagnosis

Share This Special Issue

Special Issue Editor

Special Issue Information

Keywords

Benefits of Publishing in a Special Issue

Published Papers (8 papers)

Research

Review

Other

Further Information

Guidelines

MDPI Initiatives

Follow MDPI