Engineering

Abstract: Entropy measures characterize complex dynamics in electrocardiography (ECG) and heart rate variability (HRV), but their interpretation depends on estimator, signal representation, and analytical configuration. This methodology-oriented narrative review organized 126 references by evidentiary function and applied role-appropriate interpretation boundaries. Within this evidence set, 53 human heart-failure (HF) or HF-relevant reports underwent structured mapping by method family (template, multiscale, symbolic/distributional, spectral/wavelet, or conditional/coupling), research task, data provenance, and validation design. Findings varied with configuration, rhythm, preprocessing, and recording length, while reused datasets or shared cohorts limited the independence of report-level evidence. Of 12 reports presenting diagnostic/classification performance, only two reported participant-separated internal evaluation, without establishing a fully nested leakage-free pipeline. A non-exclusive set of 10 reports examined clinical outcomes or clinically relevant event associations, all within single cohorts. Neither subset included independent external patient-cohort validation or identified calibration. Entropy should therefore be treated as a configuration-, representation-, rhythm-, and task-specific research tool for dynamic phenotyping or a candidate component of multivariable models, rather than a standalone clinical measure. Progress requires complete estimator reporting and design-appropriate participant separation; candidates intended for prediction or decision support additionally require appropriate comparators, external validation, calibration, and decision-relevant evaluation.

Abstract:

Conventional silicon-based storage is approaching fundamental physical and economic limits under exponential data growth, motivating exploration of alternative archival substrates. This paper introduces Living Information Storage Systems (LISS), a framework that exploits self-replicating biological genomes as a simultaneous storage medium, replication engine, and error-repair substrate. Within LISS, the paper presents a formal, end-to-end, quantitative treatment of human somatic genomic storage comprising: (i) an information-theoretic capacity bound derived from the Shannon capacity of the genomic substitution channel; (ii) a closed-form effective capacity model \(C_{\text{eff}} = N_s \cdot L_p \cdot \eta \cdot (1-E)\); (iii) the Adaptive Safe-Harbor Encoding (ASHE) algorithm, formulated as a constrained nucleotide sequence optimisation with a multi-objective placement scoring function, targeting \(\eta\) = 1.75 bits/nt; (iv) a hierarchical Genomic Addressing Layer (GAL) with a CHR:LOCUS:BLOCK:OFFSET address space and 16-nt barcode scheme; (v) the Genomic Redundant Distributed Placement (GRDP) algorithm, framed as a Maximum Distance Separable (MDS) code over safe-harbor loci; (vi) a stochastic clonal cell-population model for long-term data integrity; and (vii) a five-tier governance framework with regulatory citations. Monte Carlo simulation (\(n=3{,}000\) trials) over an injection-deletion-substitution error channel yields post-ECC recovery rates of 91.53%–99.17% (95% CI) and raw BER of 41.8%–46.7% across 128 B–1 KB payloads. A mutation-drift model projects 77.9% data integrity at 50 years, and analytical MDS bounds predict that GRDP with \(k{=}3\) achieves end-to-end retrieval probability of 99.65% (dual-parity). A Gompertz-based clonal expansion model quantifies mosaicism degradation to 61.4% cell-fraction retention at 20 years, motivating ex vivo refreshal protocols. Four primary unsolved constraints for practical deployment are identified: prime-editing efficiency, innate immune response to CRISPR machinery, safe-harbor scarcity, and regulatory absence. To the author's knowledge, this constitutes among the first systematic, quantitative, end-to-end architectures for human somatic genomic storage in the literature.

Abstract: Background: Prostate cancer is a neoplastic disease that generally remains asymptomatic until its late-stage emergence. Among the techniques approved to date for prostate cancer diagnosis, prostate-specific antigen (PSA) screening is the most established method for its early detection. Currently, centralized clinical laboratories measure serum PSA using costly, time-consuming, multi-step immunoassays that rely heavily on lab technicians and highly sophisticated equipment. To address this issue, we developed a disposable, label-free electrochemical biosensor as a proof of concept for rapid, single-step, user-friendly, cost-effective detection of total PSA (tPSA) and free PSA (fPSA). Methods: The immunosensors were based on HNO₃-pretreated pencil graphite electrodes (PGEs) modified with graphene oxide (GO). Antigen concentrations were determined by measuring signals from the [Fe(CN)₆]^3-/4- redox probe using cyclic voltammetry (CV); the presence of antigen molecules increased electron-transfer resistance at the PGE surface. Results: The 30-minute tPSA biosensor exhibited a linear range and limit of detection (LOD) of 0.01-1000 ng/mL and 0.648 pg/mL, respectively. The 45-minute fPSA immunosensor had a 0.1-1000-ng/mL linear range and a 0.07-pg/mL LOD. Conclusions: The biosensors’ performance, along with selectivity, reproducibility, and stability analyses, and validation tests using commercial enzyme-linked immunosorbent assay (ELISA) kits with real human samples, provided early evidence supporting the potential clinical application of these immunosensors.

Abstract: Continuous gait monitoring is essential for clinical rehabilitation, yet traditional analysis systems are often confined to a laboratory and costly. This study develops a modular, low cost, multi-channel wearable capacitive sensing band for lower limb monitoring and as potential platform for rehabilitation research. The system utilizes repositionable conductive fabric electrodes on adjustable bands for the shank and thigh, employing an RC timing circuit to detect muscle deformation during walking and measure the capacitance via Wi-Fi for real time monitoring. Sensor characterization demonstrated high stability, with linearity exceeding 0.98, creep under 0.4%, and hysteresis below 3%, regardless of sensor conditions (Pristine, Aged, and Used). Moreover, comparison with OpenCap kinematics during normal walk across different gait phases showed strong to very strong waveform similarities with correlations exceeding 0.999 during terminal stance. Furthermore, the system demonstrated feasibility in identifying simulated hemiplegic, spastic, and Parkinsonian gait against normal walking, with abnormality thresholds reaching 40 % of the gait cycle points exceeding the abnormality threshold of |z|>2. These findings suggest the proposed modular system is an effective tool for identifying muscle deformation in lower limb, which is essential in gait assessment in clinical rehabilitation. Future studies should validate the device with actual patients with musculoskeletal disorders.

Abstract: Background: Developing Brain–Computer Interfaces (BCIs) for children with severe neuromotor disabilities remains challenging due to developmental neurophysiology and the high cognitive demands of motor imagery. Hybrid BCIs integrating electroencephalography (EEG) and electrooculography (EOG) can overcome these limitations by combining continuous motor-related control with reliable discrete commands. This multiple-case study investigated the feasibility of a hybrid EEG–EOG BCI integrated with a virtual reality (VR) powered-wheelchair simulator for mobility training. Methods: Six participants (one child, five adolescents) with severe neuromotor disabilities completed repeated sessions using a self-paced BCI. The system combined continuous kinesthetic motor imagery (KMI) for steering and EOG-detected voluntary blinks for discrete commands. We evaluated online classification performance (accuracy, sensitivity, specificity), user workload (NASA-TLX), and simulator-related discomfort (MSAQ) under fully and semi-immersive VR conditions. Results: Despite expected inter-subject variability, most participants achieved functional online VR wheelchair control, reaching a mean accuracy of 77% in successful sessions with balanced sensitivity and specificity. NASA-TLX scores indicated moderate cognitive workload but consistently low frustration. Zero dropouts and low MSAQ scores demonstrated excellent protocol tolerability, high engagement, and minimal cybersickness across both VR modalities. Conclusions: This study demonstrates the feasibility of combining hybrid EEG–EOG BCI control with VR for pediatric powered mobility training. The framework enabled functional control with high usability and minimal adverse effects, supporting the clinical potential of hybrid BCI–VR systems as accessible assistive technologies and providing a foundation for larger clinical trials.

Abstract: Reconstruction of large craniofacial bone defects typically relies on vascularized autografts, but involves donor-site morbidity and limited availability. Tissue-engineered vascularized grafts could address this gap; however, current strategies remain limited by in vitro cell manipulation, regulatory complexity, high costs, and difficult clinical scaling up t. We previously prefabricated a maxillary graft combining devitalized bone matrix (DBM), freshly-isolated human adipose-derived cells, known as the stromal vascular fraction (SVF), and Bone Morphogenetic protein-2 (BMP-2). We tested if cells were essential by developing a cell-free approach with only DBM (SmartBone) and BMP-2. SmartBone blocks were loaded with BMP-2 (60 µg/mL in a fibrin hydrogel) with or without SVF cells and implanted subcutaneously in nude mice for 12 weeks (ectopic model) or combined to an axial arteriovenous (AV) bundle in nude rats (785 mm³) for 2 or 12 weeks (vascularized model). Bone formation was assessed by X-ray tomography and histology. Across both mice and rats, BMP-2 alone was sufficient to generate robust and organized bone, and adding SVF cells provided no additional benefit. In mice, BMP-2 grafts formed lamellar bone with marrow, while grafts without BMP-2 produced only calcified fibrous tissue. Bone and marrow volumes did not differ between BMP-2 and BMP-2+SVF groups (p>0.05). In rats, both conditions showed significant increases in mineralized volume from 2 to 12 weeks (p<0.05). In conclusion, BMP-2 combined to DBM is enough to robustly generate vascularized bone grafts while using off-the-shelf, approved materials and microsurgical prefabrication, providing a simplified, translationally-feasible approach for reconstruction of large bone defects.

Abstract: On February 2, 2026, the U.S. Food and Drug Administration's (FDA) Quality Management System Regulation (QMSR) replaced the decades-old Quality System Regulation (QSR) at 21 CFR Part 820, incorporating ISO 13485:2016 by reference as the operative quality management system standard for finished medical device manufacturers. This review synthesizes the regulatory history, structural mechanics, and practical consequences of that transition, drawing on the Federal Register final rule, FDA's QMSR guidance and frequently asked questions, and published industry analyses. It then examines how the transition differentially affects three roles in the device supply chain: original equipment manufacturers (OEMs), contract manufacturers, and critical component suppliers. The review finds that while FDA has characterized QSR and QMSR as "substantially similar" in aggregate stringency, the removal of the former §820.180(c) exemption for internal audit, supplier audit, and management-review records is the single most consequential change for supply-chain participants, since it alters what was previously treated as protected internal deliberation into evidence subject to FDA review. The review concludes with practical recommendations for each entity type, including a crosswalk and gap assessment, a record-readiness review, and quality-agreement updates.

Abstract: Affordances in ecological psychology are relational properties of organism–environment systems, yet their quantitative characterization remains an open problem. Shannon information theory measures uncertainty and statistical dependence but does not by itself specify what environmental structure means for an acting organism. This article proposes a limited operational bridge between the two frameworks by treating Gibsonian specification as a relation to be quantified, rather than identifying it with Shannon entropy. Let A denote a locomotor affordance, such as stable continuation of walking, and let S denote a screw-structured haptic state derived from the coupled motion and loading of the body–environment system. The central quantities are the conditional distribution p(A | S), the residual conditional entropy H(A | S), and the mutual information I(A;S) = H(A)−H(A | S): stronger ecological specification corresponds to lower residual uncertainty and greater mutual information. Screw theory supplies the mechanical geometry through twists, helical axes, pitch, Plücker coordinates, and reciprocal screw conditions, including the projective reciprocity condition (K1 + K2) sinϕ + d cosϕ = 0. We hypothesize that screw-structured representations preserve affordance-relevant information more efficiently than Cartesian kinematics, yielding lower H(A | S_screw) and higher I(A;S_screw). A single-case gait demonstration and an explicit estimation protocol support the framework as a testable basis for quantifying ecological specificity in stable gait, perturbation, recovery, and fall-risk analysis.

Abstract: Pharmaceutical formulation and delivery-device development has historically proceeded by forward trial and error: a candidate material, formulation, or geometry is proposed, fabricated, and tested, and the cycle repeats until an acceptable product emerges. Machine learning enables an inverse alternative: given a target product profile — a desired release rate, gastric-residence time, transfection efficiency, or shape transformation — a model proposes the material composition or device geometry predicted to achieve it, collapsing what was previously dozens of design-build-test cycles into a handful. This review develops that inverse-design paradigm specifically for biopharmaceutical formulation and delivery, grounding each technical claim in verifiable, named research from the Massachusetts Institute of Technology and Harvard University. We examine four case studies: the directed message-passing graph neural network that Stokes, Yang, Collins, Barzilay, Jaakkola and colleagues at MIT used to discover the antibiotic halicin from a chemical library of over one hundred million candidates; the geometrically optimized, origami-inspired ingestible device that Traverso, Langer and colleagues at MIT and Brigham and Women’s Hospital/Harvard Medical School designed to self-deploy into a gastric-retentive torus; the algorithmically encoded, biomimetic 4D-printing platform that Lewis, Mahadevan and colleagues at the Harvard Wyss Institute developed to program anisotropic hydrogel swelling into predetermined shape changes; and the autonomous, machine-learning-guided closed-loop synthesis and formulation systems that Jensen and colleagues in the MIT Department of Chemical Engineering have advanced as “self-driving laboratories.” From these cases we distill a general method taxonomy — Bayesian optimization, generative models, graph neural networks, physics-informed machine learning coupled to finite-element simulation, and active-learning closed loops — and map each onto concrete biopharmaceutical inverse-design problems: ionizable-lipid and lipid-nanoparticle design for mRNA delivery, shape-memory device and 4D-printed geometry optimization, and controlled-release formulation screening. We close by examining the regulatory posture of the U.S. Food and Drug Administration toward artificial intelligence in drug development, including its January 2025 draft guidance and the January 2026 joint FDA-EMA guiding principles, and by outlining the data, validation, and interpretability barriers that stand between today’s proof-of-concept demonstrations and routine, regulator-accepted inverse design of biopharmaceutical products.

Abstract: Biomass conversion can support the circular economy when energy production is linked to the recovery of value from organic waste and residual biomass. This exploratory review examines how recent research has approached this relationship, with particular attention to the practical conditions that influence the use of biomass-to-bioenergy systems. A structured search of Scopus was used to identify relevant articles and reviews, which were examined through a critical thematic synthesis. The literature covers a broad range of feedstocks and conversion routes, including anaerobic digestion, pyrolysis, hydrothermal processing and integrated biorefinery systems. The findings show that the circular value of these technologies depends not only on energy production but also on how recovered products are used and whether the process can operate under realistic conditions. Feedstock availability, scale, logistics and access to suitable markets remain important constraints. Further progress will depend on evidence from longer-term operation and on assessments that better reflect the conditions in which these systems are implemented.

Abstract: Autonomic innervation governs cardiac development, functional maturation, and electrophysiological integration, yet in vitro models rarely reproduce the spatial organization through which sympathetic neurons reach the heart. We first established planar neuro-cardiac cocultures of intact neonatal rat superior cervical ganglia (SCG) and embryonic rat cardiac aggregates. Sympathetic axons were tyrosine hydroxylase (TH) positive, penetrated the cardiac aggregates, and increased the spontaneous contraction rate. RT-qPCR on these planar cocultures showed that innervated cardiac tissue upregulated contractile, gap-junction, and adrenergic transcripts (Actc1, Actn2, Myh6, Adrb1, Gja5/Cx40, Cx43, Cx45) relative to both freshly isolated tissue and cardiac-only cultures, without re-expression of the fetal isoform Myh7. We then developed 3D Tissue Engineered Innervated Cardiac Units (TE-ICUs), in which SCG and cardiac aggregates occupy opposite ends of a 3–10 mm methacrylated hyaluronic acid microcolumn. Sympathetic axons in isolation projected up to ~6 mm and released ~170 nM norepinephrine upon electrical stimulation. Within TE-ICUs, fasciculated TH+ tracts bridged the microcolumn and penetrated the cardiac unit, while cardiac aggregates self-organized into chamber-like compartments; the contraction rate was significantly higher than that of cardiac-only controls at 8 days in vitro. TE-ICUs provide an anatomically inspired platform for studying neuro-cardiac development and engineering innervated cardiac tissue.

Abstract: Lower back pain (LBP) affects an estimated 75–80% of individuals worldwide, with poor posture during strength training identified as a significant contributing factor. This study presents an intelligent, low-cost wearable system for real-time lumbar posture monitoring during Romanian Deadlifts. The system combines an MPU-9250 inertial measurement unit (IMU), an ESP32 microcontroller, a cloud-deployed Random Forest model (PostureProML), and a Flutter-based mobile application (PostureProne). The system achieved 89.2% classification accuracy on a held-out test set across three posture categories (proper, rounded, and arched), with cross-validated performance reaching 94.5%. Angular drift remained minimal (1.8°–7.1°), and battery life supported up to 4 hours of operation. Usability testing with fourteen participants (aged 18–25) indicated high acceptance, with 90% finding the app intuitive. By enabling immediate feedback and encouraging posture correction, this interdisciplinary system offers a tool to support posture awareness, with potential implications for injury risk reduction pending further validation.

Abstract: Conventional heart rate variability (HRV) analysis primarily characterizes R–R interval (RRI) variability, whereas the dynamic contribution structure among multiple electrocardiogram (ECG)-derived intervals and respiration remains less well understood. This study investigated the frequency-dependent organization of P–P interval (PPI), RRI, T–T interval (TTI), and respiration during postural change. Thirteen healthy young participants underwent simultaneous ECG and respiratory recording in the supine and standing positions. A four-variable multivariate autoregressive (MVAR) model was constructed, and relative power contribution (RPC) was quantified in the very-low-frequency, low-frequency (LF), and high-frequency (HF) bands and visualized as directed network topologies. Conventional HRV analysis confirmed marked autonomic modulation during standing, including shortened RRI and pronounced attenuation of HF power. RPC analysis identified 14 posture-related differences after false discovery rate correction across all 32 LF- and HF-band Source–Target comparisons. In the LF band, several TTI-related contributions increased during standing, including TTI→PPI and TTI→RRI. In the HF band, multiple PPI- and RRI-related cross-variable contributions and RRI self-contribution decreased, whereas TTI self-contribution increased markedly. Seven of the eight significant HF findings remained significant after excluding participants whose respiratory peak frequency entered the LF band. Furthermore, nine of the 14 significant RPC components remained significant across all tested P- and T-wave timing perturbations up to ±10 ms. These findings indicate that postural change reorganizes the frequency-dependent contribution structure among ECG-derived intervals and respiration. From a Network Physiology perspective, the proposed framework extends conventional RRI-centered HRV analysis by characterizing the multivariate structure and state-dependent reorganization of cardiac–respiratory dynamics.

Abstract: The objective of this study was to evaluate the effect of the carbon source (CS), the nitrogen source (NS), the percentage of pumpkin (PJ) in the juice, and heat treatment (TT) on the growth of Streptococcus thermophilus and Lactobacillus delbrueckii subsp. bulgaricus, in order to develop an inoculum adapted to a plant-based matrix. The experimental process included a screening design, maximum rise trajectory, and response surface optimization, supplemented by principal component analysis (PCA). Cell viability was determined by microbiological count (CFU/mL) and expressed as the logarithmic increase in cell viability (LIV). The screening identified CS, NS, and PJ as significant factors (p = 0.003). The optimization predicted a maximum LVI of 8.17 with 8.0% CS, 7.26% NS, and 7.28% PJ. The model had an R² of 89.22% and a standard error of less than 10%. PCA explained 79.81% of the variability through two components (PC1 = 61.02% and PC2 = 18.80%), associating PC1 with conductivity, salinity, and suspended solids, and PC2 with pH, total solids, and titratable acidity, demonstrating the relationship between microbial growth and physicochemical changes during fermentation.

Abstract: Continuous monitoring of surgeons' stress has the potential to improve surgical performance, training, and patient safety. However, developing reliable machine learning models for stress estimation is challenging because continuous psychological ground truth is impractical to obtain during surgery. Surrogate-label learning has emerged as a promising alternative, yet the physiological validity of the generated labels and the resulting model predictions remains largely unexplored. This study presents a multi-level physiological validation framework for evaluating personalized heart rate variability (HRV)-derived surrogate stress labels and corresponding deep learning models for continuous intraoperative stress estimation. The framework comprises two complementary validation levels. The first evaluates the physiological consistency of personalized clustering-derived surrogate labels against HRV-derived stress profiles using representative surgical cases, surgical workflow analysis, and correlation analysis. The second assesses the agreement between surrogate labels and convolutional neural network (CNN) and long short-term memory (LSTM) model predictions using prediction error metrics and Bland–Altman analysis across multiple temporal sequence lengths. Validation was performed using wearable electrocardiogram (ECG) recordings collected from 27 colorectal surgical procedures performed by seven surgeons. The surrogate labels demonstrated strong agreement with the HRV-derived reference (Spearman's ρ = 0.841, p = 0.036), while both CNN and LSTM models accurately preserved the physiological stress patterns, exhibiting moderate-to-strong correlations, low prediction errors, and acceptable agreement across all evaluated sequence lengths. The proposed framework extends conventional model evaluation by emphasizing physiological consistency in addition to predictive performance and provides a systematic methodology for validating surrogate-label-based physiological monitoring systems when continuous psychological ground truth is unavailable.

Abstract: In search for possible strategies to regenerate bony tissue of the non-load bearing regions, injectable hydrogels have received substantial attention considering their tunable mechanical properties, high biocompatibility, and customizable bioactivity along with the possibility for minimally invasive delivery. Currently, many injectable hydrogels (e.g., collagen, hyaluronic acid, chitosan, polyethylene glycol, alginate, fibrin, etc.) have found their utility in bone tissue repair and augmentation. However, they can be further tailored by incorporating various bioactive additives to achieve desirable functionality, known as composite hydrogel. Depending on the nature of bioactive additives, such injectable osteogenic composite hydro-gels (IOCHs) can serve as a reservoir for sustained release therapeutic drugs, antimicrobials, growth factors, and other biomolecules and act as temporary matrices to recruit and ac-commodate the cells of interest, thereby affording a favorable stimulatory microenvironment for bone regeneration. As noted above, the combination of osteogenic materials (such as bioglass, clays, calcium phosphate (CaP), metal-organic frameworks (MOFs), and carbon-based nanoparticles) with hydrogels can endow them with the capacity to promote osteogenic differentiation of those recruited cells in addition to improved mechanical strength and struc-tural stability of the hydrogel. This review summarizes a significant number of injectable osteogenic composite hydrogels and compares their bioactivity with particular emphasis on bone regeneration. In addition, this review also discusses the advantages and drawbacks as-sociated with the inclusion of various additives to formulate bioactive hydrogels. The review concludes with highlights of the foreseeable challenges using IOCHs as well as the prospects for both research and clinical applications in BTE.

Abstract: Heart rate variability (HRV) features provide compact representations of R-R interval (RRI) dynamics, but temporal information may be selectively lost during feature aggregation. This study investigated which aspects of temporal structure remain discriminable after RRI time series are compressed into HRV-based representations. Publicly available sinus rhythm and continuous atrial fibrillation (CAF) datasets were analyzed. For each RRI sequence, three temporal variants were generated: Forward, Reverse, and Shuffle. Three input representations were compared: raw RRI sequences, seven RRI-derived HRV features, and the same seven features supplemented with three temporal descriptors: Half Mean Difference, Linear Slope, and Difference Asymmetry (DAsym). Classification was evaluated across observation durations from 5 to 30 min using six machine-learning classifiers. Raw RRI classification was modest in sinus rhythm and approached chance level in CAF at durations of 10 min or longer. The 7-HRV representation substantially improved discrimination, particularly for Shuffle sequences, but Forward and Reverse remained more difficult to distinguish. Adding the three temporal features further improved classification, with accuracy reaching 0.803 in sinus rhythm and 0.923 in CAF. In CAF, DAsym showed the highest class-specific permutation importance for both Forward and Reverse at five of the six observation durations. Post-hoc analysis further showed that the magnitude of negative RRI changes exceeded that of positive changes in CAF (p<0.001). These findings indicate that feature aggregation does not uniformly eliminate temporal information: RRI-derived HRV features retain sensitivity to disruption of temporal organization but are comparatively insensitive to temporal direction. A small number of interpretable temporal descriptors may recover part of this lost directional information while preserving a compact representation. More broadly, compact physiological representations should be evaluated not only by how efficiently they can be transmitted, but also by what information they retain, what they omit, and whether the retained information is sufficient for the intended purpose.

Abstract: Dysphagia comprises aspiration risk as well as oral-phase impairment, post-swallow residue, reduced sensation, delayed pharyngeal onset, and incomplete hyolaryngeal elevation. Current daily-life approaches generally separate wearable sensing from electrical stimulation and rarely provide per-swallow, safety-bounded adaptation. This paper presents SafeSwallow™, a compact wearable neuroprosthesis that integrates around-ear or glasses electroencephalography (EEG) for anticipatory arming, submental surface electromyography (sEMG), laryngeal bioimpedance (BI), contact acoustics, and inertial measurement for swallow verification and elevation feedback. The controller combines reactive multimodal gating, anticipatory onset prediction, explicit electromechanical-delay (EMD) compensation, elevation-state estimation, and iterative learning control (ILC) of stimulation dose under a non-learned Tier-0 safety supervisor. The study evaluates the architecture on two complementary tracks: a physiology-inspired 1-kHz synthetic closed-loop experiment with healthy and Parkinsonian phenotypes and a public videofluoroscopy-aligned high-resolution cervical auscultation (HRCA) dataset. On held-out synthetic trials, closed-loop control achieved 100.0% Q1 force success with 0.0% misses and 0.0% false stimulation, while reducing normalized elevation error from 0.253 under matched open-loop dosing to 0.116. On the public HRCA test set, the improved causal front-end increased detection from 83.1% to 94.9% and modeled Q1 force success from 59.2% to 92.9%. These results support engineering feasibility of the proposed timing and control architecture, while demonstrating that clinical translation requires paired multimodal datasets, measured electrode-specific EMD, and prospective human validation.

Abstract: Secondary lymphedema is a persistent complication resulting from damage to the lymphatic system, including lymphatic vessel and lymph-node injury after cancer treatment. The authors previously proposed a lymphatic vessel concept, followed by fabrication of a first-generation (V1) prototype in 2024. The present study advances this work through simulation-driven redesign leading to a second-generation (V2) prototype and morphological characterization. Computational fluid dynamics (CFD) was used to reassess the initial flow boundary condition and evaluate valve openings. Assigning 4 L/day to a single micrometric vessel produced unrealistically high local velocities and pressure losses, prompting reassessment. Subsequent simulations adopted an inlet ve-locity of 0.0013 m/s and evaluated 60 and 120 µm valve openings, while 3 and 6 µm configurations were retained as narrow reference geometries. Larger openings produced substantially lower maximum velocities and pressure drops. The resulting V2 prototype has a nominal length of 3.32 mm, outer diameter of 150 µm, inner diameter of 140 µm, and approximately 5 µm wall thickness. Scanning electron microscopy (SEM) characterized the fabricated V2 structure at 23×–1500× magnification, confirming the tubular architecture and local measurements of lumen diameter and wall thickness. These findings support engineering development and identify parameters requiring experimental validation before biological or clinical assessment.

Abstract: In this paper, we propose a bioimpedance-based digital twin approach for organ-on-a-chip systems. Organ-on-a-chip platforms integrate microfluidic control, specific cell populations, extracellular matrices, mechanical stimuli and inter-tissue interactions to reproduce selected physiological functions under controlled conditions. Bioimpedance measurements allow non-destructive and repeated monitoring of cellular and tissue behavior. However, the electrical signal measured from an organ-on-a-chip system is not only dependent on the biological state. It is also influenced by the device geometry, electrode configuration, electrode-electrolyte interface, culture conditions and measurement stability. Therefore, an electrical signal cannot be directly considered as a biological state without considering the complete measurement system. We propose that the next development of organ-on-a-chip technology should move from continuous monitoring toward bioimpedance-based digital twins. The physical organ-on-a-chip system, integrated sensing system, measurement model, mechanistic model and data-driven model can be connected to form a continuously updated "experimental-computational" loop. In such a system, the objective is not only to monitor the sensor signal, but to estimate the current biological state, quantify the uncertainty associated with this estimation and predict the future response of the system to experimental perturbations. This approach could provide a transition from descriptive measurements of organ-on-a-chip systems toward predictive and adaptive microphysiological experiments.