ArXiv最新文献

Previous studies have compared neural activities in the visual cortex to representations in deep neural networks trained on image classification. Interestingly, while some suggest that their representations are highly similar, others argued the opposite. Here, we propose a new approach to characterize the similarity of the decision strategies of two observers (models or brains) using decision variable correlation (DVC). DVC quantifies the image-by-image correlation between the decoded decisions based on the internal neural representations in a classification task. Thus, it can capture task-relevant information rather than general representational alignment. We evaluate DVC using monkey V4/IT recordings and network models trained on image classification tasks. We find that model-model similarity is comparable to monkey-monkey similarity, whereas model-monkey similarity is consistently lower. Strikingly, DVC decreases with increasing network performance on ImageNet-1k. Adversarial training does not improve model-monkey similarity in task-relevant dimensions assessed using DVC, although it markedly increases the model-model similarity. Similarly, pre-training on larger datasets does not improve model-monkey similarity. These results suggest a divergence between the task-relevant representations in monkey V4/IT and those learned by models trained on image classification tasks.

Polymer-based long-acting injectables (LAIs) have transformed the treatment of chronic diseases by enabling controlled drug delivery, thus reducing dosing frequency and extending therapeutic duration. Achieving controlled drug release from LAIs requires extensive optimization of the complex underlying physicochemical properties. Machine learning (ML) can accelerate LAI development by modeling the complex relationships between LAI properties and drug release. However, recent ML studies have provided limited information on key properties that modulate drug release, due to the lack of custom modeling and analysis tailored to LAI data. This paper presents a novel data transformation and explainable ML approach to synthesize actionable information from 321 LAI formulations by predicting early drug release at 24, 48, and 72 hours, classification of release profile types, and prediction of complete release profiles. These three experiments investigate the contribution and control of LAI material characteristics in early and complete drug release profiles. A strong correlation (>0.65) is observed between the true and predicted drug release in 72 hours, while a 0.87 F1-score is obtained in classifying release profile types. A time-independent ML framework predicts delayed biphasic and triphasic curves with better performance than current time-dependent approaches. Shapley additive explanations reveal the relative influence of material characteristics during early and for complete release which fill several gaps in previous in-vitro and ML-based studies. The novel approach and findings can provide a quantitative strategy and recommendations for scientists to optimize the drug-release dynamics of LAI. The source code for the model implementation is publicly available.

In the last century, most sensorimotor studies of cortical neurons relied on average firing rates. Rate coding is efficient for fast sensorimotor processing that occurs within a few seconds. Much less is known about long-term working memory with a time scale of hours (Ericsson and Kintsch, 1995). The discovery of millisecond-precision spike initiation in cortical neurons was unexpected (Mainen and Sejnowski, 1995). Even more striking was the precision of spiking in vivo, in response to rapidly fluctuating sensory inputs, suggesting that neural circuits could, in principle, preserve and manipulate sensory information through spike timing. High temporal resolution enables a broader range of neural codes. It could also support spike-timing-dependent plasticity (STDP), which is triggered by the relative timing of spikes between presynaptic and postsynaptic neurons in the millisecond range. What spike-timing mechanisms could regulate STDP in vivo? Cortical traveling waves have been observed across many frequency bands with high temporal precision. Traveling waves have wave fronts that could link spike timing to STDP. As a wave front passes through a cortical column, excitatory synapses on the dendrites of both pyramidal and basket cells are stimulated synchronously. Inhibitory basket cells form a calyx on pyramidal cell bodies, and inhibitory rebound following a strong transient hyperpolarization can trigger a backpropagating action potential, which arrives shortly after the excitatory inputs on pyramidal dendrites. STDP activated in this way could persist for hours, creating a second-tier network. This temporary network could support long-term working memory, a cognitive network riding above the long-term sensorimotor network. On their own, traveling waves and STDP have not yet yielded new insights into cortical function. Together, they could be responsible for how we think (Sejnowski, 2025).

Background: Whole-brain radiotherapy (WBRT) is a common treatment due to its simplicity and effectiveness. While automated Field-in-Field (Auto-FiF) functions assist WBRT planning in modern treatment planning systems, it still requires manual approaches for optimal plan generation including patient-specific hyperparameters definition and plan refinement based on quality feedback.

Purpose: This study introduces an automated WBRT planning pipeline that integrates a deep learning (DL) Hyperparameter Prediction model for patient-specific parameter generation and a large-language model (LLM)-based conversational interface for interactive plan refinement.

Methods: The Hyperparameter Prediction module was trained on 55 WBRT cases using geometric features of clinical target volume (CTV) and organs at risk (OARs) to determine optimal Auto-FiF settings in RayStation treatment planning system. Plans were generated under predicted hyperparameters. For cases in which the generated plan was suboptimal, quality feedback via voice input was captured by a Conversation module, transcribed using Whisper, and interpreted by GPT-4o to adjust planning settings. Plan quality was evaluated in 15 independent cases using clinical metrics and expert review, and model explainability was supported through analysis of feature importance.

Results: Fourteen of 15 DL-generated plans were clinically acceptable. Normalized to identical CTV D95% as the clinical plans, the DL-generated and clinical plans showed no statistically significant differences in doses to the eyes, lenses, or CTV dose metrics D1% and D99%. The DL-based planning required under 1 minute of computation and achieved total workflow execution in approximately 7 minutes with a single mouse click, compared to 15 minutes for manual planning. In cases requiring adjustment, the Conversational module successfully improved dose conformity and hotspot reduction.

Conclusions: The proposed system improves planning efficiency while maintaining clinically acceptable plan quality. It demonstrates the feasibility of combining DL-based hyperparameter prediction with LLM interaction for streamlined, high-quality WBRT planning.

In this work, we proposed virtual imaging simulators as an alternative approach to experimental validation of beam range uncertainty in complex patient geometry using a computational model of a human head and a photon-counting CT scanner. We validate the accuracy of stopping power ratio (SPR) calculations using a conventional stoichiometric calibration approach and a prototype software, TissueXplorer. A validated CT simulator (DukeSim) was used to generate photon-counting CT projections of a computational head model, which were reconstructed with an open-source toolbox (ASTRA). The dose of 2 Gy was delivered through protons in a single fraction to target two different cases of nasal and brain tumors with a single lateral beam angle. Ground truth treatment plan was made directly on the computational head model using clinical treatment planning software (RayStation). This plan was then recalculated on the corresponding CT images for which SPR values were estimated using both the conventional method and the prototype software TissueXplorer. The mean percentage difference in estimating the stopping power ratio with TissueXplorer in all head tissues inside the scanned volume was 0.28%. Stopping power ratios obtained with this method showed smaller dose distribution differences from the ground truth plan than the conventional stoichiometric calibration method on the computational head model. Virtual imaging offers an alternative approach to validation of the SPR prediction from CT imaging, as well as its effect on the dose distribution and thus downstream clinical outcomes. According to this simulation study, software solutions that utilize spectral information, such as TissueXplorer, hold promise for more accurate prediction of the stopping power ratio than the conventional stoichiometric approach.

Various theoretical and empirical studies have accounted for why humans cooperate in competitive environments. Although prior work has revealed that network structure and multiplex interactions can promote cooperation, most theory assumes that individuals play similar dilemma games in all social contexts. However, real-world agents may participate in a diversity of interactions, not all of which present dilemmas. We develop an evolutionary game model on multilayer networks in which one layer supports the prisoner's dilemma game, while the other follows constant-selection dynamics, representing biased but non-dilemmatic competition, akin to opinion or fad spreading. Our theoretical analysis reveals that coupling a social dilemma layer to a non-dilemmatic constant-selection layer robustly enhances cooperation in many cases, across different multilayer networks, updating rules, and payoff schemes. These findings suggest that embedding individuals within diverse networked settings-even those unrelated to direct social dilemmas-can be a principled approach to engineering cooperation in socio-ecological and organizational systems.

Alzheimer's disease (AD) is a multifactorial neurodegenerative disorder characterized by progressive cognitive decline and widespread epigenetic dysregulation in the brain. DNA methylation, as a stable yet dynamic epigenetic modification, holds promise as a noninvasive biomarker for early AD detection. However, methylation signatures vary substantially across tissues and studies, limiting reproducibility and translational utility. To address these challenges, we develop MethConvTransformer, a transformer-based deep learning framework that integrates DNA methylation profiles from both brain and peripheral tissues to enable biomarker discovery. The model couples a CpG-wise linear projection with convolutional and self-attention layers to capture local and long-range dependencies among CpG sites, while incorporating subject-level covariates and tissue embeddings to disentangle shared and region-specific methylation effects. In experiments across six GEO datasets and an independent ADNI validation cohort, our model consistently outperforms conventional machine-learning baselines, achieving superior discrimination and generalization. Moreover, interpretability analyses using linear projection, SHAP, and Grad-CAM++ reveal biologically meaningful methylation patterns aligned with AD-associated pathways, including immune receptor signaling, glycosylation, lipid metabolism, and endomembrane (ER/Golgi) organization. Together, these results indicate that MethConvTransformer delivers robust, cross-tissue epigenetic biomarkers for AD while providing multi-resolution interpretability, thereby advancing reproducible methylation-based diagnostics and offering testable hypotheses on disease mechanisms.

This study presents the design, simulation, and experimental validation of a dual-tuned concentric multimodal surface coil for 7T 1H/31P magnetic resonance spectroscopic imaging (MRSI), developed to significantly enhance 31P B1 efficiency while improving 1H performance. The coil architecture utilizes two interleaved sets of three concentric loop resonators. Intra-nucleus electromagnetic coupling within each three-loop set generates a spectrum of eigenmodes; the operational modes for 1H and 31P were specifically selected because their co-directed current distributions reinforce the magnetic field at the center, yielding B1 patterns that resemble those of conventional single-loop surface coils but with superior efficiency. Full-wave electromagnetic simulations and bench measurements on a fabricated prototype were conducted to characterize the multimodal resonance behavior, scattering parameters, B1 distribution, and 10-g local SAR, using size-matched conventional single-tuned loops as references. The results confirmed that the design reproducibly generated the predicted eigenmode ordering with sufficient spectral separation to prevent interference from parasitic or undesired modes. Notably, the multimodal design achieved an 83% boost in 31P B1 efficiency and a 21% boost in 1H B1 efficiency at the coil center compared to same-sized single-tuned references. Sufficient inter-nuclear decoupling was achieved to prevent signal leakage between channels, and simulations with a human head model confirmed that the peak 10-g local SAR remained comparable to conventional designs. These findings demonstrate that this multimodal concentric design offers a robust and highly efficient solution for multinuclear MRSI at ultrahigh fields, effectively mitigating the sensitivity limitations of X-nuclei without compromising proton-based imaging capabilities.

Pulmonary arterial hypertension (PAH) is a progressive cardiopulmonary disease that leads to increased pulmonary pressures, vascular remodeling, and eventual right ventricular (RV) failure. Pediatric PAH remains understudied due to limited data and the lack of targeted diagnostic and therapeutic strategies. In this study, we developed and calibrated multi-scale, patient-specific cardiovascular models for four pediatric PAH patients using longitudinal MRI and catheterization data collected approximately two years apart. Using the CRIMSON simulation framework, we coupled three-dimensional fluid-structure interaction (FSI) models of the pulmonary arteries with zero-dimensional (0D) lumped-parameter heart and Windkessel models to simulate patient hemodynamics. An automated Python-based optimizer was developed to calibrate boundary conditions by minimizing discrepancies between simulated and clinical metrics, reducing calibration time from weeks to days. Model-derived metrics such as arterial stiffness, pulse wave velocity, resistance, and compliance were found to align with clinical indicators of disease severity and progression. Our findings demonstrate that computational modeling can non-invasively capture patient-specific hemodynamic adaptation over time, offering a promising tool for monitoring pediatric PAH and informing future treatment strategies.

Wearable technologies have the potential to transform ambulatory and at-home hemodynamic monitoring by providing continuous assessments of cardiovascular health metrics and guiding clinical management. However, existing cuffless wearable devices for blood pressure (BP) monitoring often rely on methods lacking theoretical foundations, such as pulse wave analysis or pulse arrival time, making them vulnerable to physiological and experimental confounders that undermine their accuracy and clinical utility. Here, we developed a smartwatch device with real-time electrical bioimpedance (BioZ) sensing for cuffless hemodynamic monitoring. We elucidate the biophysical relationship between BioZ and BP via a multiscale analytical and computational modeling framework, and identify physiological, anatomical, and experimental parameters that influence the pulsatile BioZ signal at the wrist. A signal-tagged physics-informed neural network incorporating fluid dynamics principles enables calibration-free estimation of BP and radial and axial blood velocity. We successfully tested our approach with healthy individuals at rest and after physical activity including physical and autonomic challenges, and with patients with hypertension and cardiovascular disease in outpatient and intensive care settings. Our findings demonstrate the feasibility of BioZ technology for cuffless BP and blood velocity monitoring, addressing critical limitations of existing cuffless technologies.