To address the challenges of caregiving for the bedridden elderly, traditional nursing beds often suffer from limited interaction and a lack of real‐time health monitoring. Most existing smart solutions rely heavily on cloud computing, which introduces significant system latencies (> 200 ms) and poses severe privacy risks for sensitive health data. This paper presents a fully localized intelligent nursing bed system designed for low‐latency and high‐privacy closed‐loop care. The hardware architecture is centered on the RK3588 embedded platform, integrated with a 6 TOPS NPU, and fuses a 24 GHz frequency‐modulated continuous‐wave (FMCW) millimeter‐wave radar with vision‐based sensing. At the circuit level, a specialized dual‐layer control board was developed to provide physical isolation between logic and execution units. This design utilizes 74HC238D decoding logic to expand control commands and incorporates a hardware‐level limit protection circuit based on 74HC08D AND gates, ensuring deterministic response and mechanical safety. To overcome feature extraction challenges and background interference, our framework enhances YOLOv8n with four key modules: C2f‐DysnakeConv for hand feature capture, SPPF‐LSKA for multiscale fusion, Hierarchical Attention Module (HAM) for background suppression, and Detect_DBB for efficient detection. For physiological monitoring, we employ EMD and Hilbert transform to extract vital signs from radar phase data. On the RK3588 platform, the optimized model achieves 94.85% precision and 93.76% mailto:mAP@0.5 at 53.1 FPS on the NBG‐7 gesture dataset. The system maintains end‐to‐end latency under 50 ms and vital sign errors within ±2 bpm (heart rate) and ±1.5 breaths/min (respiration), establishing a robust paradigm for privacy‐preserving, real‐time intelligent care through Edge AI and specialized hardware integration.
Accurate probabilistic prediction and uncertainty quantification are important in reliability engineering, survival analysis, predictive maintenance, healthcare, and risk assessment. This study proposes a deep neural distributional regression framework based on the Power Lindley distribution for modeling positive continuous and right‐censored outcomes. The framework models the distributional parameters as nonlinear functions of explanatory variables, thereby producing observation‐specific conditional distributions rather than only point predictions. Model estimation is performed through a censoring‐aware likelihood in which observed events contribute through the probability density function and right‐censored observations through the survival function. A systematic hyperparameter search and early stopping are used to improve optimization stability and generalization. The proposed method is evaluated through Monte Carlo simulations under nonlinear data‐generating mechanisms and an additional distributional misspecification scenario, and is further assessed using the Veterans′ Administration Lung Cancer and primary biliary cirrhosis benchmark datasets. Comparisons with linear regression, generalized linear models, random forest regression, and optimized deep neural regression demonstrate competitive predictive performance while preserving the principal advantage of direct probabilistic modeling. The framework provides individualized prediction intervals, conditional quantiles, survival probabilities, and other uncertainty measures. Additional interpretability analysis illustrates how explanatory variables influence the estimated distributional parameters. Overall, the proposed approach provides a flexible and interpretable framework for probabilistic prediction and uncertainty‐aware decision making in reliability and survival applications.
The improvement of soil performance using synthetic fiber reinforcement is a growing area of interest in geotechnical engineering. This study presents an experimental evaluation of the effects of synthetic PVA fiber additives on the mechanical, chemical, and microstructural properties of soil. A comprehensive dataset was utilized, including unconfined compressive strength (UCS) with a range of 160–430 kPa, split tensile strength (STS) with a range of 72–80 kPa, stress–strain behavior, disintegration rate, particle size distribution, pH with a range of 7.11–7.24, electrical conductivity (EC) of 4696, calcium carbonate (CaCO 3 ) content with a range of 11.45–13.67, and exchangeable sodium percentage (ESP) with a range of 3.81–5.2. Spectral data from energy dispersive x‐ray spectroscopy (EDS) and x‐ray diffraction (XRD) were analyzed to identify compositional and mineralogical changes resulting from fiber inclusion. The analysis revealed substantial improvements in mechanical strength and chemical stability, accompanied by observable microstructural enhancements. MATLAB‐based statistical and graphical analyses enabled a detailed correlation between fiber dosage, curing age, and soil behavior. The results indicate that synthetic fiber additives, when combined with MICP treatment, significantly improve soil properties relevant to engineering applications. Because the underlying measurements were adopted from a previously published dataset, the present contribution lies in the independent cleaning, restructuring, and unified MATLAB‐based reanalysis—curve fitting, regression, and cross‐parameter correlation—rather than in new laboratory testing. Disintegration index versus time, in minutes, increased with time depending on MICP. Although further field‐scale validation is encouraged, the observed trends suggest strong potential for applications in road subgrades, embankments, and stabilized earthworks.
Because of received signal strength (RSS) variability in indoor environments, multiple observations are often collected sequentially over time to reduce short-term fluctuations. However, these time-series observations may contain local outliers caused by measurement disturbances and environmental variability. The local outlier factor (LOF) algorithm can detect such outliers, but its performance depends on the distance metric, neighborhood size, and LOF threshold. Existing studies provide limited comparative evaluation of distance metrics for LOF-based outlier detection in time-series RSS data. This study addresses this gap by systematically evaluating four distance metrics—namely, absolute, squared, relative, and Canberra—within the LOF framework for time-series RSS outlier detection. Two RSS time-series datasets were evaluated using neighborhood sizes of k=10,20,and 30 and LOF thresholds of T=1.0,1.5,and 2.0. Performance was assessed using receiver operating characteristic area under the curve (ROC-AUC), precision-recall area under the curve (PR-AUC), precision, recall, and F1 score, with physically defined RSS bounds used to establish reference outliers. Dataset #1 contained 14 reference outliers, whereas Dataset #2 contained 175. The results show that distance metric selection affects threshold-based classification performance, although its effect varies across datasets and parameter settings. For Dataset #1, squared distance achieved perfect precision, recall, and F1 score at k=10 and T=1.5. For Dataset #2, squared distance achieved the highest F1 score of 0.2927 at k=30 and T=1.5, whereas Canberra distance achieved the highest ROC-AUC of 0.5812. The findings demonstrate that distance metric selection is an important consideration when configuring LOF for RSS outlier detection.
The effectiveness of externally bonded carbon fiber reinforced polymer (CFRP) systems for strengthening reinforced concrete (RC) beams is often limited by premature debonding. This study experimentally and numerically investigated the flexural behavior and debonding failure of RC beams strengthened with CFRP plates, with and without end U-wrap anchorage. Six simply supported RC beams were tested under four-point bending and divided into two concrete-strength groups, designated C20 and C25, with average 28-day cylinder compressive strengths of 20.2 and 25.1 MPa, respectively. Each group included an unstrengthened control RC beam, an unanchored CFRP-strengthened beam, and a CFRP-strengthened beam with end U-wrap anchorage. Comparisons within each group evaluated both the strengthening effect of the selected CFRP plate width and the additional contribution of the U-wrap to flexural capacity. CFRP plate widths of 80 and 50 mm were used for the C20 and C25 groups, respectively. These widths provided comparable ultimate-load increases of 15.72% and 15.94% for the corresponding unanchored beams relative to their control specimens. End U-wraps produced additional ultimate-load increases of 17.67% and 7.67% in the C20 and C25 groups, respectively. The greater additional capacity gain that was found in the C20 group likely reflects the combined effects of concrete strength, CFRP plate width, and CFRP–concrete bonding behavior. U-wrap anchorage changed the governing failure mode from plate-end debonding to intermediate crack-induced debonding, indicating greater utilization of the CFRP plate, although debonding was not eliminated. A three-dimensional nonlinear finite element model was validated against the experimental ultimate loads and load–deflection responses, achieving a mean absolute percentage error (MAPE) of 3.92% in the ultimate-load predictions. Representative numerical damage contours also provided qualitative support for the observed change in the critical debonding region.
