Although tracheal gas insufflation (TGI) and aspiration of expiratory gas in the dead space (ASPIDS) have been shown to enhance CO₂ removal by washing out the anatomical dead space, clinical implementation has been limited due to complications including unintended airway pressure increases and mucosal damage, as well as their complexity. We developed a novel respiratory management system combining continuous intratracheal gas suctioning with pressure control ventilation, utilizing the leak compensation function of modern ventilators. The aim of this study was to document the preliminary physiological observations of this system in a proof-of-concept porcine model. Two female pigs weighing approximately 30 kg were intubated with 7.5 mm cuffed endotracheal tubes and paralyzed under general anesthesia. Each pig was studied under four different ventilatory conditions: two with healthy lungs and two with experimentally induced lung injury (created by repeated whole-lung lavage with normal saline and injurious ventilation until PaO₂/FIO₂ < 300 mmHg). A 10 Fr closed suction catheter was inserted into the trachea. Using the leak compensation function of a Puritan Bennett™ 840 ventilator, heated and humidified gas was automatically supplied at the same flow rate as the continuous suctioning flow (13-14 L/min), maintaining constant circuit pressure. Pressure-controlled ventilation was applied, and arterial blood gas was analyzed before, during, and after continuous suctioning. PEEP was increased by 1 cmH₂O during suctioning to compensate for pressure loss through the endotracheal tube. A total of eight experimental runs were performed (two pigs × four conditions each: two healthy, two injured lung conditions). Baseline PaCO₂ ranged from 42.2 to 71.8 mmHg. During suctioning, PaCO₂ decreased in seven conditions (reductions: 0.9 to 10.1 mmHg; 1.8% to 14.1%) and increased in one condition (3.0 mmHg; 5.9%). The largest reductions occurred in conditions with higher baseline PaCO₂. After cessation of suctioning, PaCO₂ returned toward baseline levels. No ventilator alarms or operational problems occurred during the experiments. In a proof-of-concept study using a porcine model with both healthy and injured lungs, continuous intratracheal gas suctioning combined with ventilator leak-compensation may enhance CO₂ removal efficiency without increasing driving pressure. This technique has potential as a simple adjunct to conventional mechanical ventilation for enhancing CO₂ removal in patients with respiratory failure, while further investigation is required for safety and clinical applicability.
Despite the well-documented efficacy of exposure therapy for phobic disorders, its real-world implementation remains limited due to barriers in accessibility, acceptability, and generalization. This proof-of-concept study examined the feasibility and preliminary effects of a novel gamified augmented reality (AR) intervention designed to support in vivo exposure by embedding playful, non-threatening virtual stimuli into real-world anxiety-provoking contexts. Twenty individuals with various phobic disorders participated in a multiple-baseline design. At a randomly assigned time point, participants used a smartphone-based AR application, or in some cases, a VR headset version, to interact with virtual game elements within individualized fear-relevant environments. The task involved physically navigating individualized fear-relevant environments while collecting color-coded virtual objects presented via augmented reality within the real-world setting. Linear mixed-effects analyses indicated a significant reduction in self-reported situational fear following the intervention, with a large within-subject effect size estimate (Cohen's d = 2.21). Secondary outcomes related to anxiety sensitivity, agoraphobic cognitions, and depressive symptoms also showed pre-post improvements with small to moderate effect sizes. No significant differences were observed across device types or treatment settings. The findings provide preliminary support for a low-threshold, smartphone-based AR approach that may complement exposure-based interventions in naturalistic settings. However, given the small and heterogeneous sample and the absence of a control group, these results should be interpreted cautiously. Future randomized controlled studies with larger and more homogeneous samples, validated behavioral outcome measures, and follow-up assessments are needed to clarify efficacy, mechanisms of change, and long-term effects.
Radiomics-based modeling has shown promise for characterizing tumor heterogeneity, but its integration with causal machine learning for treatment-effect estimation remains underexplored in osteosarcoma. This study aimed to develop a proof-of-concept radiomics-based causal machine learning framework for exploratory estimation of average and individual treatment effects associated with neoadjuvant chemotherapy cycle intensity in osteosarcoma. This retrospective single-center study included 34 patients with osteosarcoma who underwent neoadjuvant chemotherapy followed by surgical resection. Radiomic features were extracted from pre-treatment T1-weighted magnetic resonance imaging and combined with baseline clinical variables. Three causal meta-learners-S-Learner, T-Learner, and X-Learner-were implemented to estimate counterfactual survival probabilities under high-cycle and low-cycle neoadjuvant chemotherapy strategies. Average treatment effects and individual treatment effects were derived from the predicted potential outcomes. The proposed framework enabled estimation of population-level and individualized treatment-effect measures using integrated radiomic and clinical covariates. The estimated average treatment effects differed in magnitude and direction across meta-learners, indicating instability of treatment-effect estimation in this small cohort. Confidence intervals crossed zero for two of the three learners, and model performance metrics were interpreted only as technical indicators of feasibility rather than evidence of generalizable predictive validity. This study demonstrates the methodological feasibility of combining radiomics with causal machine learning for exploratory treatment-effect estimation in osteosarcoma. Given the limited sample size, retrospective single-center design, treatment-group imbalance, missing outcome information, and uncertainty of causal assumptions, the findings should be regarded as hypothesis-generating rather than clinically actionable. Larger multicenter studies with standardized imaging protocols, adequate event counts, longer follow-up, and prospective validation are required before the translational relevance of radiomics-based causal treatment-effect estimation can be assessed.
Transforaminal lumbar interbody fusion (TLIF) is a modality for treatment of chronic lower back pain (CLBP) and/or lumbosacral radiculopathy (LR). Up to 28% of patients with indications for surgical treatment of CLBP/LR continue to have CLBP/LR after surgery despite a technically successful spinal surgery, likely due to development or persistence of neuropathic pain. Dorsal root ganglion (DRG) stimulation is often used for the treatment of chronic neuropathic pain. Thus, we present a novel technique of integrating DRG stimulation using Direct Visual Placement with open lumbar or lumbosacral decompression and instrumented fusion in a single surgical procedure. In an uncontrolled case series for a proof of concept study, we combined DRG stimulation with TLIF in a single surgical procedure to evaluate feasibility and safety. Fifteen patients with CLBP/LR and indication for TLIF received DRG stimulator placement concomitantly during TLIF. Safety of the combined approach, namely adverse event (AE), was evaluated. Also, change in back pain and leg pain visual analog scale (VAS) scores, as well as opioid usage, from baseline to 12 weeks, 6 months, and 12 months postoperatively, was tracked. No AE was serious, and was related to the 2 procedures, TLIF and DRG stimulation, being combined. One instance of electrode migration without loss of therapy was recorded. Stimulation parameters were 20 Hz frequency, 250 microsecond pulse width, and less than 1 mA amplitude. Back pain VAS scores improved 67%, 72%, and 71% from baseline at 12 weeks, 6 months, and 12 months, respectively. Leg pain VAS scores improved 67%, 69%, and 69% from baseline at 12 weeks, 6 months, and 12 months, respectively. Responder (VAS score reduction of at least 50%) analysis of back pain and leg pain VAS scores showed that 53.3% and 60.0% of subjects responded to stimulation at 10-20 days for back pain and leg pain, respectively. Responder rates increased to 66.7% and 73.3% at 12 months for back pain and leg pain, respectively. Responder rates for "either" or "both" were 80.0% and 66.0% at 12 months, respectively. Opioid usage was down to 20% from 12 weeks to 12 months. This series demonstrates the feasibility of integrating DRG stimulator placement with open lumbar or lumbosacral decompression and instrumented fusion. The Direct Visual Placement of DRG stimulator demonstrated no TLIF and DRG stimulation combination related AEs, showed feasibility, reduced pain and improved quality of life of the subjects, thus presenting a new treatment approach for CLBP/LR in a single procedure.
This proof-of-concept study explores the application of solvatochromism to resolve differences in the polarity microenvironment induced by nonpolar hydrocarbon compounds, specifically cycloalkanes. A solvatochromic probe based on Betaine 30 or Reichardt's dye was designed, and UV-vis absorption spectra were recorded for a set of structurally diverse cycloalkanes. Multivariate analysis, incorporating Principal Component Analysis (PCA) and K-means clustering, was applied to the spectral data to introduce a novel polarity classification capable of capturing subtle differences in the polarity microenvironment of cycloalkanes, compounds traditionally regarded as uniformly nonpolar. This solvatochromic approach demonstrated high sensitivity in detecting and differentiating small polarity variations among closely related isomers, revealing that molecular conformation, stereochemistry, and the number, type, and position of alkyl substituents exert measurable effects on the polarity microenvironment of hydrocarbons. These findings are particularly relevant to the development of alternative synthetic aviation fuels, where small variations in cycloalkane composition can significantly influence fuel performance, including energy density, combustion efficiency, and material compatibility (e.g., O-ring swelling). Overall, this work underscores the utility of solvatochromic probes for differentiating among structurally similar organic compounds, revealing quantitative polarity microenvironment differences across cycloalkanes and provides a preliminary polarity microenvironment classification framework for evaluating subtle intermolecular interaction differences to complement traditional methods based solely on solvatochromic parameters such as ET(30) or normalized ENT values.
The future of cerebellar stimulation should be moving out of the well-resourced research environment, so implementation needs to expect regulation, home-based models, and popular trust. An example of how a change in policy can affect the feasibility, cost and innovation schedule is the European reclassification of some non-medical NIBS devices as Class III medical devices [1]. There is also evidence in the literature of stakeholder research suggesting that parties concerned with the perceived benefit, with safety and with fear of misuse, affect the acceptability-variables that can dictate trial participation and ultimate adoption.
Distraction osteogenesis (DO) for limb lengthening often requires prolonged fixation while the regenerate consolidates. Adjuncts that improve the regenerate quality could reduce morbidity and treatment time. Hyperelastic Bone™ (HEB) is a 3D-printed hydroxyapatite (HA)/polylactic-co-glycolic acid (PLGA) scaffold, but its efficacy as an osteotomy-site adjunct during DO is unclear. After IACUC approval, rabbits underwent left tibial lengthening with a mini-rail external fixator distracted at 0.75 mm/day to 20% tibial length and received no scaffold (control), traditional Hyperelastic Bone™ (10% PLGA/90% HA), or a biphasic formulation (10% PLGA/70% HA/20% β-tricalcium phosphate [β-TCP]) placed subperiosteally at the osteotomy site. The contralateral tibiae served as references. After 8 weeks of consolidation, radiographs, micro-CT (prespecified endpoint: regenerate BMD), and torsional testing (prespecified endpoint: stiffness) were performed. Fourteen rabbits underwent attempted lengthening. Intraoperative fractures (n = 2) and fixation loss (n = 3) reduced complete terminal datasets to nine, all of which completed micro-CT and torsion testing. Radiographs confirmed regenerate formation. Regenerate BMD remained below contralateral values with controls being intermediate to the traditional and biphasic HEB formulations (60.5%, 63.8%, and 54.7% of contralateral, respectively). Despite no mineral density improvement, scaffold-treated regenerate demonstrated higher torsional stiffness and strength metrics, greatest with the biphasic formulation, with lower failure displacement and higher polar moment of inertia. Statement of Clinical Significance: A scaffold that improves functional regenerate performance could enable earlier safer weight bearing and device removal and reduce nonunion/refracture risk in DO.
Persistent emotion lability and impulsive behavior in bipolar disorder (BD) are thought to be due to impaired affective cognitive control. Neural synchrony via theta-gamma phase-amplitude coupling (PAC) in the prefrontal cortex facilitates cognitive control and is impaired in BD. This pilot study aimed to investigate the feasibility and efficacy of transcranial alternating current stimulation (tACS) in targeting the neural coordination (i.e., theta-gamma PAC) of cognitive control in a double-blind, randomized, controlled crossover trial. Participants (N = 18) with BD completed an emotional Go/No-Go task during six 5-min stimulation blocks interleaved with 2-min resting blocks. Participants received active- and sham-stimulation 1 week apart; sham comprised a ramp up/down at the beginning and end of the block to mimic stimulation sensations. Tolerability was measured using a side effect questionnaire after active and sham stimulation. Behavioral measures of discriminability (d') and reaction time during the task, and EEG measures of theta-power and theta-gamma PAC during the interleaved resting states were compared for active versus sham. Active did not differ from sham stimulation on any side effects (all p's > 0.11). Effect sizes when comparing discriminability, theta-power, and PAC across active versus sham were small but showed a consistent pattern of improvement in the active relative to the sham condition. This pilot study showed that 30-min tACS is tolerable to BD patients and demonstrated preliminary promise in enhancing affective cognitive control. Future studies should determine the optimal dosage to produce more substantial and enduring effects. ClinicalTrials: NCT05480124.
Highly effective CFTR modulators, particularly elexacaftor/tezacaftor/ivacaftor (ETI), produce rapid clinical improvements in people with cystic fibrosis. Yet early effects may be difficult to capture with spirometry or sweat test in patients with a mild disease or atypical mutations. Exhaled breath is rich in volatile organic compounds (VOCs) reflecting metabolic and inflammatory processes. We aimed to determine whether ETI induces early, measurable changes in breath composition and whether these relate to clinical outcomes. Ten adults initiating ETI were enrolled in a prospective, open-label study with breath sampling at baseline, week one and month one. VOCs were measured using real-time proton-transfer-reaction - mass spectrometry (PTR-MS). Longitudinal changes were assessed using multilevel statistics, including univariate linear mixed-effects models, and multivariate repeated measures ANOVA-simultaneous component analysis plus (RM-ASCA+); repeated-measures correlations examined associations with lung function and sweat chloride concentration. Results were compared with a healthy cohort. Amongst the eight clinical responders, 11 features changed significantly after ETI initiation. Eight differed from healthy controls at baseline and shifted towards healthy levels over one month. RM-ASCA+ identified monotonous and non-monotonous patterns capturing various dynamics such as acute, progressive or delayed metabolic responses. A 11-feature PLS-DA model classified visits with high accuracy (AUC=0.84-0.96). Ten VOCs correlated with clinical readouts. Tentatively identified features pointed towards a shift in the microbiome and/or energy metabolism. ETI induces rapid alterations in exhaled VOCs, many trending towards healthy values and correlating with clinical improvement. Real-time breath analysis offers a promising non-invasive surrogate for early monitoring of therapeutic response.
Transaction verification is essential to blockchain security. As blockchain data continue to grow, resource limited nodes may be forced to operate as non-full nodes, which weakens independent verification and may increase centralization risk. To address this issue, the stateless blockchain technology has been proposed, which uses the accumulators to combine the UTXO set into one fixed-size commitment. However, they suffer from two critical limitations: (i) the inability to support script validation duo to the lack of scriptSig, and (ii) the absence of an outsourcing mechanism to ensure that task executors reliably provide the appropriate witness for the nodes just recovered from failures. We propose LSTVS, a lightweight stateless transaction verification architecture for UTXO based blockchains, which extends RSA accumulator based stateless verification with script-based authorization verification, cache assisted stale proof tolerance, and outsourced witness updates. First, we incorporate UTXO fields associated with transactions into the membership witness to enable digital signature verification. Second, we reconstruct the transaction data format to prevent the exponential growth of transaction reference fields. Finally, we introduce an outsourcing mechanism to improve the transaction verification rate while minimizing computational resource consumption. Experimental results show that the proposed architecture supports the core validation dimensions of UTXO based stateless verification, including existence verification, unspent status verification, and script-based authorization verification, while avoiding UTXO scale dependent proof growth and introducing input dependent transient witness overhead. Compared with existing state of the art RSA accumulator-based schemes, LSTVS improves the transaction verification rate and reduces the local witness update overhead for intermittently online nodes.
In the field of colonoscopy, robotic systems have been developed to support or replace human operators due to a shortage of trained endoscopists. We developed the Autonomous Colonoscope Robot System (ACRS), based on the Endoscopic Operation Robot version 4, to evaluate whether expert-derived operational data can enable autonomous colonoscope insertion. ACRS was trained using insertion data obtained from an expert endoscopist operating a standardized colonoscopy training model. Automated insertions were evaluated using Pattern 1 of the model, a highly controlled configuration without substantial loop formation. Completely automated insertions were designated Level 4, whereas insertions requiring some manual assistance were designated Level 3. Level 4 insertion times were compared with manual insertions performed by an expert and trainees. Of the 72 automated insertions at Level 3 or higher, 62 are classified as Level 4, giving a success rate of 86.1% (95% CI, 75.9-93.1%). The average insertion time for Level 4 procedures is 2.92 ± 1.20 minutes, significantly longer than that of the expert (1.43 ± 0.32 minutes), but comparable to the time taken by trainees (2.97 ± 1.32 minutes; errors are standard deviations). ACRS demonstrates proof-of-concept feasibility for autonomous colonoscope insertion under simplified, controlled model conditions. Further validation in more complex models, animal studies, and clinical settings is required before translational application. Colonoscopy is an important test for finding colorectal cancer early. An endoscope is a thin, flexible tube with a camera that is inserted into the colon, but safe and painless insertion requires considerable skill. We developed the Autonomous Colonoscope Robot System (ACRS), based on a master–slave robotic system, in which a doctor’s hand movements are transmitted to a robotic device that holds and moves the endoscope. Using insertion data from an expert doctor in a colonoscopy training model, we trained an artificial intelligence model to control the robot automatically. In a simplified model setting, ACRS achieved fully automated insertion in many trials. This proof-of-concept study may support future development of robotic systems that help make colonoscopy safer and more widely available.
The mitochondrial theory of aging has been proposed, which suggests that the accumulations of multiple in mtDNA with age induce decreased mitochondrial respiratory function, ultimately leading to aging. This hypothesis is widely accepted because mtDNA mutator mice (Polgmut/mut mice) with a homozygous proofreading-deficient mutation in Polg gene accumulate random point mutations in mtDNA, resulting in mitochondrial respiratory dysfunction and premature aging phenotypes. However, the accurate mtDNA mutation frequency in Polgmut/mut mice has remained unclear, and the causal relationship between accumulation of random point mutations in mtDNA and mitochondrial respiratory dysfunction in Polgmut/mut mice has been debated. Then, we verified the exact mtDNA mutation frequency in Polg mice and experimentally varied mtDNA mutation frequency to test their correlation with mitochondrial respiratory activity. Our results showed that, regardless of mtDNA mutation frequency, mitochondrial respiratory activity was mildly reduced in Polg+/mut mice with a heterozygous proof-reading deficient mutation in Polg and severely reduced Polgmut/mut mice. Moreover, by varying mtDNA mutation frequency, some Polg+/+ mice showed mtDNA mutation frequency equivalent to those of Polgmut/mut mice, but mitochondrial respiratory function in Polg+/+ mice was normal. These results suggest that a reduced mitochondrial function in Polg mice is correlated with Polg mutation rather than accumulation of random mutations in mtDNA. Thus, this study proposed the need to reconsider the hypothesis that accumulation of random mutations in mtDNA induces a decreased respiratory function, which form the core of the mitochondrial theory of aging.
Temporal interference transcranial alternating-current stimulation (TI-tACS) enables non-invasive deep brain neuromodulation but requires bulky, expensive equipment. This work presents an integrated wearable TI-tACS device addressing these barriers. A battery-powered system (55$\times 40\times$25 mm) featuring a quadratic-boost converter generates bipolar $\pm$15 V from 3.7 V battery at moderate duty cycles ($D$=0.55-0.68). Validation included electrical equivalent-load testing, electromagnetic modeling, saline phantom measurements, and proof-of-concept in vivo measurements in rats ($N$=5). The proposed device achieves >50× volume reduction compared to existing TI-tACS implementations, with ${\$}$300 USD cost, 1.25% mean current error, and 0.28% mean THD. Phantom measurements matched simulations. Proof-of-concept in-vivo measurements demonstrated depth-dependent electric-field distributions containing both carrier-frequency and beat-frequency spectral components in rat brain tissue. The system successfully delivers TI-tACS waveforms and demonstrates interference field generation in biological tissue. This compact, affordable platform addresses critical hardware accessibility gaps, providing a foundation for preclinical TI-tACS research and enabling portable experimental paradigms.
Renal cell (RCC) and breast carcinoma (BC) frequently develop lytic bone metastases (BM) that are usually treated with systemic and locoregional therapies. Surgery plays a crucial role when pathological fractures or lesions involve a load-bearing bone and consists in the removal of tumor area followed by replacement with implant or prosthesis. In the present study, a proof of concept of novel locoregional therapies for BM patients was evaluated, consisting in the functionalization of biomorphic apatitic scaffolds with proved intrinsic regenerative properties with Everolimus and an anti-RANKL antibody, selected as relevant anticancer and anti-resorptive biomolecules, respectively. Upon confirmation of the biocompatibility of unloaded scaffolds with healthy bone cells, the effective antitumoral activity of released EVE on BC cell lines (MCF-7 and MDA-MB-231) and RCC cell line (Caki-2) was demonstrated. The inhibitory effects of drugs on osteoclast differentiation were validated using peripheral blood mononuclear cells (PBMCs). Preclinical validation was performed on monocultures and cocultures of cancer and bone cells sharing the same culture medium. In silico analyses were exploited to predict cancer cell behavior on medicated scaffolds providing complementary insights while reducing the need for extensive wet-lab experimentation. Finally, as a preliminary proof of translational relevance, the medicated scaffolds were tested with RCC patient-derived explants.
Background: Many diseases evolve through complex, nonlinear trajectories shaped by interacting genetic, cellular, and environmental factors over time. Such dynamics are difficult to represent using static risk models, particularly in biologically heterogeneous conditions such as pediatric acute lymphoblastic leukemia (ALL). Here, we present a large concept model (LCM) as a mechanistic, hypothesis-generating framework for simulating longitudinal disease trajectories using pediatric ALL relapse dynamics as a proof-of-concept exemplar. Methods: We developed a causal longitudinal modeling framework implemented within the aiHumanoid v11.0 platform to characterize post-remission relapse dynamics. Seven clinically relevant ETV6::RUNX1-based genotypic profiles were simulated from remission baseline (T0) through 2 post-remission intervals (T1 = 3 months; T2 = 6 months). Longitudinal remission-to-relapse changes were evaluated across genotype- and age-defined virtual cohorts using descriptive nonparametric effect-size-oriented measures. Relapse dynamics were summarized using 2 composite system-level metrics: the relapse risk score and relapse pressure index. Results: The model generated distinct genotype- and age-associated trajectory patterns across relapse-relevant biological domains and produced composite measures reflecting modeled relapse pressure within the simulation environment. Greater relapse-associated biological divergence was observed in selected genotype-age strata, particularly in domains related to clonal evolution, treatment resistance, and minimal residual disease. Conclusions: This ALL-focused proof of concept demonstrates the architectural and analytic potential of mechanistic trajectory simulation for hypothesis generation, longitudinal systems modeling, and future integration with real-world longitudinal datasets.
Endoscopic sinus surgery (ESS) carries risks such as vision loss and intracranial injury due to the proximity of critical structures and unrecognised anatomical variants. We developed convolutional neural networks (CNNs) to detect free-floating anterior ethmoidal arteries (FFAEA), Onodi cells, and Haller cells on coronal sinus CT, and evaluated BioMedCLIP, a biomedical vision-language model (VLM), in a few-shot setting. CT scans from 122 ESS patients were anonymised, standardised coronal CT images were captured, and variant presence was annotated. Five ImageNet-pretrained CNN backbones were assessed using repeated patient-wise five-fold cross-validation across 40 configurations. The best CNNs achieved balanced accuracies of 77.8 ± 2.3% (FFAEA), 74.6 ± 2.5% (Onodi), and 63.7 ± 6.2% (Haller). BioMedCLIP achieved 65.5 ± 3.2%, 63.8 ± 1.8%, and 73.5 ± 3.6%, respectively, outperforming CNNs for Haller cell detection while providing competitive performance for the other variants. These models demonstrate proof-of-concept performance for automated identification of selected sinonasal variants on standardised coronal CT images under internal patient-wise cross-validation.
This study introduces a simulation-based wearable biomechanical sensor network framework intended to support real-time fatigue monitoring and performance prediction in gymnastic training environments. The work was motivated by the recognised need for objective, quantitative fatigue indicators that complement, rather than replace, the subjective ratings and laboratory-based assays currently used in gymnastics. Although sensor-based fatigue monitoring has been examined in endurance sports, the multi-planar, high-dynamic skills of gymnastics remain largely under-served; the framework therefore couples a sport-specific multi-modal sensor topology with personalised machine learning, an integration that to our knowledge has not been reported in the gymnastics literature. The proposed multi-node architecture integrates accelerometers, gyroscopes and force sensors, together with multi-feature fusion and machine learning-based fatigue classification, signal processing routines, and personalised prediction modules that capture the temporal dependencies of fatigue progression. In a computational validation built around virtual athlete models, the framework achieved a detection accuracy of approximately 93% across the six fatigue severity levels examined, while personalised algorithms outperformed generic counterparts by 12-18% within the same simulation pipeline. We emphasise that these figures are simulation-derived proof-of-concept results; they should not be interpreted as empirical validation, and confirmation through studies with real athletes wearing physical sensors in actual training environments is still required. The performance prediction module retained acceptable forecasting accuracy up to a 60-minute horizon, which in principle could support proactive training load management once the framework has been empirically validated. The real-time processing pipeline is, in principle, fast enough to feed immediate coaching feedback, and the modular architecture is intended to accommodate different gymnastic disciplines in future field studies. Taken together, the present work lays methodological groundwork for evidence-based training optimisation that, pending real-world validation, may eventually contribute to performance development and athlete-safety practice through objective biomechanical monitoring and predictive analytics.
The circadian clock is an endogenous oscillator with a period of approximately 24 hours, with the central pacemaker localized in the suprachiasmatic nucleus (SCN). Jet lag occurs because the clock requires several days to entrain to a new environment. To address this, it is common practice to use bright light exposure-a potent zeitgeber-at specific times. Here, we propose an alternative approach based on a mathematical limit cycle model: reducing the amplitude of the circadian oscillator to enhance its responsiveness to timing cues. As a proof of concept, we generated a transgenic rat line expressing a dominant-negative (DN) form of BMAL1 in the nervous system to create a model with weakened circadian function. We developed an expression vector using the mouse Prion protein promoter to express a BMAL1 DN lacking the C-terminus, a region critical for CRY1 interaction. We confirmed that the behavioral rhythms of BMAL1 DN(+) Tg rats exhibited lower amplitudes compared to their DN(-) littermates under both light-dark (LD) and constant darkness (DD) conditions. Furthermore, ex vivo SCN organ cultures from BMAL1 DN(+) rats showed lower amplitude rhythms than those from controls. When exposed to bright light at ZT14 (two hours after lights-off) and released to DD, BMAL1 DN(+) rats showed a significantly greater phase response. Additionally, following an abrupt shift in the LD cycle, BMAL1 DN(+) rats required fewer days to re-entrain to the new environment. These suggest that reducing the amplitude of the circadian oscillator enhances phase responsiveness and accelerates entrainment to new light-dark cycles.
Chimeric antigen receptor (CAR) T cell therapy for solid tumors is constrained by the scarcity of safe, uniformly expressed cell-surface targets. Here we identify glycoprotein NMB (GPNMB)-an MiT/TFE-family fusion-driven protein-as being highly, homogeneously and stably expressed in primary and relapsed alveolar soft-part sarcoma (ASPS) and translocation renal cell carcinoma. We develop a GPNMB-directed CAR T cell product, GCAR1, which demonstrates potent activity against patient-matched cells, organoids and xenograft models. Post hoc interim analysis of a first-in-human open-label, individual-participant trial ( NCT07104682 ) for a participant with relapsed/refractory, metastatic ASPS showed that GCAR1 induces stable disease for up to 3 months, accompanied by resolution of many nontarget lesions (primary endpoint), and is well tolerated. GCAR1 T cells expand in peripheral blood as a polyclonal population and remain detectable for 1 month. Spatial transcriptomics identified immunosuppressive niches in a treatment-resistant lesion and immune checkpoint blockade synergized with GCAR1 in a xenograft model. Altogether, our data provide a proof of concept for treating GPNMB-expressing solid tumors with GCAR1 and more broadly targeting surface antigens driven by oncogenic gene fusions with CAR T cell therapies.
Long-duration spaceflight exposes crews to isolated, confined, and extreme environments where chronic stress and limited real-time support heighten interpersonal tensions. Large language models (LLMs) could serve as onboard artificial intelligence teammates for de-escalation, but could also exacerbate tensions by endorsing hostility or exclusion. This study evaluates whether LLMs demonstrate prosocial conflict mediation behaviors vs. failure modes that could elevate interpersonal risk. We tested four LLMs (GPT-5, Claude Sonnet 4, Grok 4, Gemini 2.5 Pro) using a Dartmouth PATH Program scenario where one crewmember criticizes an absent teammate and tries to recruit another against him. Models were evaluated across seven test batteries: adversarial framing, backstory manipulation, team role variations, aggressor identity variations, target identity variations, mental health sensitivity, and direct ethics questions. We scored 196 responses using a three-point rubric (defended excluded teammate, neutral, or endorsed exclusion). Claude Sonnet 4 achieved perfect scores (3.0). GPT-5 averaged 2.97. Gemini 2.5 Pro (2.81) showed vulnerability to narrative capture, shifting from defending the excluded crewmember to endorsing criticism when given unfavorable backstory. Grok 4 (2.85) exhibited significant failure under social proof pressure, explicitly endorsing exclusion. All models showed consistent prosocial responses regardless of demographic factors and correctly identified ethical issues when questioned directly. LLMs show promise as behavioral health teammates but have critical vulnerabilities. They can be manipulated into endorsing teammate exclusion through backstory reframing, peer pressure, or authority hierarchies, requiring safety protocols before deployment. Dvorak C, Buckey JC Jr. Large language models as behavioral health teammates in long-duration spaceflight. Aerosp Med Hum Perform. 2026; 97(7):516-523.