Biology subjects

Chen, B.

Publications and source records attributed to Chen, B..

At least 19 recordsLinked to original sources

bioRxiv2026

Preclinical Evaluation of Itareparib, a brain penetrant PARP1-selective radioligand, in Glioma Rat Model and Nonhuman Primate

Rationale: Poly(ADP-ribose) polymerase-1 (PARP1) is a central mediator of DNA damage response and is upregulated in glioma, supporting it as a molecular imaging biomarker. However, quantitative PET assessment of PARP1 in infiltrative tumors with a preserved blood-brain barrier remains an unmet need. In this study, we evaluated if [18F]Itareparib, a brain-penetrant, non-trapping, PARP1-selective radioligand with reversible kinetics, could improve brain tumor diagnosis and enable target occupancy studies in neurooncology. Methods: [18F]Itareparib was synthesized and evaluated in an orthotopic rat model of BT142 IDH1-mutant gliomas. Dynamic PET scans were performed with or without the administration of the PARP1-selective inhibitor AZD9574. Standardized uptake value ratios (SUVRs) for tumors were calculated using the contralateral region as reference and compared with distribution volume ratios (DVRs) calculated using kinetic modeling. PARP1-specific binding and tracer stability were assessed using autoradiography, immunofluorescence, biodistribution, and radiometabolite analysis. Brain kinetics of [18F]Itareparib were evaluated in a healthy nonhuman primate (NHP) at baseline and blocking conditions. Results: [18F]Itareparib rapidly entered the rat brains and showed 4-fold higher uptake in BT142 tumors than in the contralateral brain. AZD9574 reduced tumor uptake by 77%, supporting the PARP1-specific binding of [18F]Itareparib in the tumors. SUVR(30-60 min) showed the closest agreement with DVR. Autoradiographic tumor-to-reference mean ratio was 7.3, and AZD9574 reduced the tumor-to-reference ratio by 56%. PARP1 immunofluorescence was higher in tumor than the rest of the brain and correlated with autoradiographic results. In rats, the parent fraction of [18F]Itareparib exceeded 99% in brain through 60 min post tracer injection (p.i.) and was approximately 88% in plasma at 60 min p.i. In the NHP, [18F]Itareparib showed rapid brain uptake, and VT decreased by 21+/-6 % after blocking with AZD9574. Conclusion: [18F]Itareparib showed brain penetration, high in vivo metabolic stability, reversible binding kinetics, and PARP1-specific tumor uptake. These findings support further development of [18F]Itareparib for the quantitative PET imaging of PARP1 in brain tumors.

cancer biology↗

bioRxiv2026

In Vivo Transcriptomics-Based Framework for Systematic Indication Expansion

Capturing the systemic effects of drugs in vivo remains a fundamental challenge for drug repurposing strategies. Here, we present an in vivo transcriptomics-based drug repurposing framework that profiles multi-organ RNA-seq from drug-treated rodents to model systemic drug responses and match tissue-specific drug signatures against a compendium of disease signatures for indication expansion. We applied this framework to semaglutide, a GLP-1 receptor agonist originally developed for diabetes and subsequently expanded for several other indications. By harmonizing more than 900 human disease transcriptomic signatures via agentic AI and using Z-transformed reversal of gene expression scoring to quantify drug-disease relationships, our approach successfully recovered established semaglutide indications while deprioritizing Alzheimers disease, an indication for which semaglutide was not effective. Top predictions for common diseases were supported by real-world electronic health record data from the Mount Sinai Health System and several predictions for rare indications are proposed for further investigation. This framework provides a scalable strategy for indication expansion across diverse therapeutic agents, especially investigational compounds for which real-world evidence remains sparse.

systems biology↗

bioRxiv2026

Tonotopic mapping in children shows an overrepresentation of high frequencies in autism

Altered sensitivity to sounds despite normal peripheral hearing is common in neurodevelopmental disorders but the underlying mechanisms remain poorly understood. Here we investigate tonotopic organization in 86 children (6-11 years) with autism spectrum and related neurodevelopmental disorders as well as typically developing (TD) peers, using a child-friendly, passive listening, functional magnetic resonance imaging design. In TD children, tonotopic maps in primary auditory cortex resembled the typical high-low-high frequency gradients previously reported in adults, while children with autism showed an overrepresentation of high frequencies consistent with what has been observed in animal models of autism. Higher frequency tuning further correlated with greater caregiver-reported sound but not touch sensitivities. Importantly, tonotopy is a core organizational feature of auditory cortex that develops early in life and remains relatively stable after the critical period of heightened plasticity ends. Thus, these results suggest that atypical auditory processing in autism may result from early disruption to tonotopic map formation. Longitudinal neuroimaging studies are needed to understand the genetic and environmental causes of altered auditory cortex topology and its long-term consequences for higher-level auditory, language, and social communication development.

neuroscience↗

bioRxiv2026

Benchmarking longitudinal local centiles for radiographic knee osteoarthritis prognosis

Cartilage thickness on MRI is a structural biomarker of osteoarthritis (OA), yet anatomical variation and aging can mask early disease, including changes occurring 12-36 months before incident OA. We propose longitudinal local centile maps that use normative growth charts as prior information and integrate a knee's earlier scans to identify cartilage that is thinner or thicker than expected. We built the reference charts from 5,045 scans of 957 persistently radiographically normal knees in the Osteoarthritis Initiative. Under participant-separated nested cross-validation, longitudinal centile profiles achieved higher 12- and 24-month AUCs for predicting incident radiographic OA than regional PCA, radiomics, cross-sectional centile features, and raw-map deep features across most classifiers. These conventional descriptors added little when combined with centile profiles, and cross-sectional centiles added little beyond longitudinal centiles. Deep centile features achieved the best AUCs (0.954 [95% CI 0.931-0.974] at 12 months and 0.969 [0.949-0.985] at 24 months; balanced accuracy 0.917 and 0.931) and outperformed raw-map deep features in all 12 classifiers after correction for multiple comparisons. No representation consistently improved 36-month prediction. Longitudinal centiles transform cartilage thickness maps into reference-based deviation maps: deep longitudinal centile features provide the highest predictive accuracy, while regional longitudinal centile profiles provide interpretable prognosis with the second-highest accuracy.

bioengineering↗

bioRxiv2026

Auditing Protein-Protein Interaction Signals with Sparse Autoencoder Fingerprints

Protein language models have become a dominant foundation for sequence-based protein-protein interaction (PPI) prediction, but their generalization remains limited under stringent evaluation, and benchmark accuracy alone cannot reveal whether a PPI predictor learns partner-specific biological signals or exploits contextual shortcuts. Here we introduce AuditPPI, an interpretable framework that transforms sparse-autoencoder (SAE) features from a frozen protein language model into order-invariant pair fingerprints, recasting PPI prediction as an auditable tabular-learning problem. AuditPPI achieves competitive predictive performance across diverse PPI benchmarks while enabling multiscale auditing of the information supporting its predictions. Protein-level audits show that partner-independent participation signals remain substantial in conventional benchmarks, including protein-disjoint benchmark, but are largely insufficient for pair-level discrimination in topology- and degree-controlled benchmark. Pair-level analyses further show that removing protein identity overlap does not eliminate contextual structure: protein-disjoint benchmarking retains strong subcellular co-localization signals and benchmark-dependent feature matching captured by SAE co-activation and absolute-difference. Structural analyses, however, provide little evidence that these predictive signals are specifically grounded in PPI interfaces: feature enrichment at interfaces for protein-disjoint benchmark was no longer detectable after controlling for surface exposure, and contact-specific enrichment was observed for only a small fraction of testable SAE feature pairs. AuditPPI therefore separates predictive success from mechanistic fidelity and provides a practical framework for evaluating whether sequence-based PPI predictors capture partner-specific biological signals instead of contextual shortcuts.

bioinformatics↗

bioRxiv2026

Estimand-aware and donor-aware triangulation of predefined gene-set signals across human tendon transcriptomic datasets

We analysed five predefined programmes across three human tendon transcriptomic contexts: acute exercise-associated fibroblast pseudobulks (E-MTAB-15400; 4 control and 4 exercise samples), mechanically stretched donor-derived cells (GSE150482; 3 normal and 3 tendinopathy donors), and paired lesional versus grossly normal-appearing tendon (GSE26051; 23 donor pairs). Inference was conducted at the sample, donor or donor-pair level. Non-stimulated GSE150482 single-cell RNA sequencing comprised only N1 and D1, whereas stretched single-cell RNA sequencing comprised N1-N3 and D1-D3; the conditions also used Drop-seq and 10X Chromium, respectively. A donor-replicated disease-by-stimulation interaction is therefore not estimable. Within-context Benjamini-Hochberg adjustment supported acute mechanical-response, extracellular-matrix (ECM) and protein-folding differences and paired-lesion collagen, integrin and ECM differences, but no stretched donor-background difference. Exact label permutations, a conservative global 15-test adjustment, leave-one-unit analyses, a Fib1-Fib4 pooled sensitivity and a paired exact Wilcoxon sensitivity qualified these findings. Integrin was fully nested in ECM, while collagen overlapped 52/79 genes with ECM. Collagen fibril organization showed the most directionally consistent matrix-associated pattern across the three non-equivalent contrasts, without establishing a disease-specific loading response or causal mechanism. The analysis demonstrates why public tendon datasets must be integrated by estimand and biological inference unit rather than pooled as interchangeable replications.

bioinformatics↗

bioRxiv2026

PCDH17 regulates lysosomal degradative capacity to promote autophagy attenuation during prolonged starvation

Autophagy is induced by nutrient starvation to recycle intracellular constituents; however, its activity must subsequently be attenuated during prolonged nutrient deprivation. The mechanisms underlying this attenuation in mammalian cells remain incompletely understood. Here, using complementary HaloTag-based assays, we show that autophagic activity declines during prolonged starvation in HeLa cells. A genome-wide CRISPR/Cas9 knockout screen designed to identify cells that sustain autophagic activity under these conditions identified protocadherin 17 (PCDH17) as a regulator of autophagy attenuation. PCDH17 depletion maintained autophagic activity during prolonged starvation without detectably altering mTORC1 signaling, ULK1 abundance, or the proximal machinery of autophagosome formation. Instead, PCDH17 depletion increased lysosomal abundance, acidification, and proteolytic activity, whereas PCDH17 overexpression produced reciprocal effects. We further identified a lysosome-associated PCDH17 subpopulation that is supplied predominantly through the biosynthetic ER-Golgi pathway. This pool undergoes proteolytic processing and lysosomal turnover, with starvation preferentially accelerating degradation of the C-terminal fragment while preserving a comparatively stable N-terminal fragment. Together, these findings identify PCDH17 as an unexpected negative regulator of lysosomal function and demonstrate that modulation of lysosomal degradative capacity contributes to autophagy attenuation during prolonged starvation.

cell biology↗

bioRxiv2026

Bumetanide Strengthens Residual Brain Bladder Communication After Spinal Cord Injury

Neurogenic bladder is one of the most disabling consequences of spinal cord injury (SCI), yet it remains unclear whether residual brain bladder communication persists after injury and can be therapeutically strengthened. Here, we integrated analysis of a clinical SCI cohort with studies in a mouse model that recapitulates key features of human neurogenic bladder. SCI markedly impaired both descending and ascending limbs of the spinobulbospinal micturition reflex, reducing bladder responses evoked by stimulation of the pontine micturition center (PMC) and bladder filling-induced activation of the periaqueductal gray. Despite this marked functional impairment, pseudorabies virus tracing, optogenetics, and machine learning assisted three dimensional imaging revealed persistent bladder connected neurons within the spared thoracic interlesion region. Neurons within this region remained responsive to descending PMC input, identifying the interlesion network as a candidate substrate for residual brain bladder communication. Early continuous intrathecal bumetanide improved urinary storage and emptying, enhanced descending PMC to bladder and ascending bladder to brain signaling, and increased the functional engagement of bladder-connected thoracic neurons without increasing urine production. Bumetanide also increased transsynaptic labeling between the bladder and PMC, consistent with enhanced polysynaptic brain bladder connectivity. Together, these findings indicate that brain bladder communication is not completely lost after SCI and may be supported by a functionally compromised residual spinal network that remains amenable to therapeutic reinforcement. Targeting spared autonomic circuitry may therefore provide a strategy for improving bladder function after SCI.

neuroscience↗

bioRxiv2026

Cardiomyocyte prohibitin ablation reprograms cardiac metabolism revealing a pathogenic role for mTORC1 in dilated cardiomyopathy

Maintaining cardiac structure and function throughout the lifespan requires a delicate balance in carbon allocation between energetic and biosynthetic processes. At the nexus of this balance are prohibitins-1 and -2 (PHB1, 2) which form a ring-like complex in mitochondrial and plasma membranes responsible for coordinating cellular growth, metabolism and autophagy. Here we describe how ablation of the PHB complex in cardiomyocytes of adult mice (cPHB1KO) causes unrestrained mechanistic target of rapamycin complex 1 (mTORC1) activity and a Warburg-like reprogramming of glucose metabolism in heart toward enhanced de novo amino acid biosynthesis. These changes are accompanied by disruptions in mitochondrial Ca2+ handling and impaired autophagy, leading to severe dilated cardiomyopathy and mortality within 12 weeks. Using pharmacological and nutritional approaches, we further show that mTORC1 inhibition attenuates pathologic cardiac remodeling only in female cPHB1KO mice. Our findings illustrate novel mechanisms linking the PHB complex with altered carbon flux and pathogenesis of cardiomyopathy.

cell biology↗

bioRxiv2026

A pretrained unified model enables cellular functional profile prediction and multi-objective virtual drug screening

Cells are characterized by molecular states, coordinated molecular interactions, regulatory programs, and responses to perturbations. Systematic mapping of these cellular functional profiles across biological contexts remains experimentally costly and fragmented. Here we present InsilicoCell, a pretrained multi-modal, multi-task model that unifies prediction of cellular functional profiles spanning molecular states, molecular interactions, and perturbation-induced responses. Built on a supervised transformer architecture and pretrained on more than 88 million measurements across seven tasks, including drug sensitivity, drug-induced gene expression, and drug-protein binding, InsilicoCell learns a shared representation that links molecular profiles to cellular phenotypes, improves performance over task-specific models, and generalizes to unseen entities, contexts, and conditions. InsilicoCell extends beyond cell line systems to patient, spatial and single-cell settings, and enables multi-objective virtual drug screening. It identifies novel candidate compounds with experimental validation, including c-Myc activity inhibitors, antifibrotic agents and stemness-inducing compounds. Together, InsilicoCell provides a scalable framework for predictive cellular biology and therapeutic discovery.

bioinformatics↗

bioRxiv2026

Spatial multi-omics analysis reveals vimentin-high macrophages-endothelial cells niche shapes CAFs heterogeneity in colorectal cancer metastasis

The spatial architecture of the tumor microenvironment (TME) is pivotal in the progression of colorectal cancer (CRC) liver metastasis. By applying high-plex spatial multi-omic mapping and neighborhood analysis to a discovery cohort of colorectal cancer primary tumor (PT) and paired liver metastases (LM), we identified a specialized vimentin-high macrophages-endothelial cells niche that orchestrates cancer-associated fibroblast (CAF) phenotypes. Mechanistically, in primary tumors, vimentin-high macrophages secrete INHBA to activate the ACVR2/TGF-{beta} axis in endothelial cells, driving CAFs toward a myCAF phenotype. Conversely, in liver metastases, these macrophages secrete CXCL9 to trigger the PI3K-Akt/NF-[kcy]B/CXCL12 cascade in endothelial cells, directing CAFs toward an iCAF state. Clinically, high niche activity predicts poor survival. Divergent endothelial signaling in primary versus metastatic lesions exposes site-specific stromal vulnerabilities for therapeutic targeting, with architectural features discernible from routine histopathology. These findings reveal a site-specific regulatory mechanism of the macrophage-endothelial niche, offering a novel and clinically significant biomarker for CRC prognosis.

cancer biology↗

bioRxiv2026

Automating scientific annotations for open transcriptomic profiles via multi-stage agents

Public transcriptomic repositories contain millions of samples, yet their large-scale reuse is hindered by heterogeneous and inconsistently reported metadata. In the Gene Expression Omnibus (GEO), key biological information is often distributed across study- and sample-level records, requiring context-dependent interpretation. Here we present GEOMeta, a large language model (LLM)-based multi-stage workflow with task-specialized agents for automated GEO metadata curation. The pipeline separates metadata retrieval, task-specific information extraction, field standardization, ontology mapping and quality control. Using GEOMeta, we generated standardized annotations for approximately 600,000 human bulk RNA-seq samples. To demonstrate its utility, we benchmarked transcriptome representation models for predicting sex, age, tissue and disease from transcriptome embeddings. We further prospectively annotated newly submitted GEO studies and evaluated 22 frontier LLMs. Recent open-source Flash models achieved annotation quality comparable to leading reasoning models while reducing costs by an order of magnitude. GEOMeta provides a scalable resource and reproducible framework for metadata curation.

bioinformatics↗

bioRxiv2026

DuplexFM: Transferable small-RNA target representations link miRNA interactions to siRNA efficacy prediction

MicroRNAs (miRNAs) and small interfering RNAs (siRNAs) share Argonaute-mediated guide-target recognition, yet quantitative siRNA efficacy measurements are substantially scarcer and more costly to generate than miRNA-target interaction data. We therefore asked whether miRNA interaction data could provide transferable supervision for siRNA efficacy prediction. Here we present DuplexFM, a biologically grounded framework that uses sample-specific gates to integrate five evidence sources: pairing and sequence-context priors, duplex energetics, experimentally supervised mRNA accessibility, target-to-guide cross-attention, and contextual token-pair compatibility. The accessibility expert, trained on nucleotide-resolution icSHAPE measurements, achieved a held-out nucleotide-level Pearson correlation of 0.627 and evaluated accessibility at seed match and energy-supported candidate sites. On miRBench v7, three independently trained DuplexFM models achieved a macro APS of 0.873 (SD = 0.002), soft-voting increased this to 0.876 and yielded the highest APS on all four test sets. We then froze the miRNA-trained representation and trained only a lightweight residual head with 24 siRNA-specific descriptors. Transfer improved Pearson and Spearman correlations, AUPRC, and F1 over the descriptor-only baseline in all six evaluation settings. The ensemble achieved the highest Pearson and Spearman correlations in four settings, whereas OligoFormer remained stronger on Huesken and Takayuki. These findings show that experimentally grounded accessibility and miRNA-derived interaction representations provide complementary, transferable information, supporting a parameter-efficient route towards unified modeling of Argonaute-guided RNA regulation. Code and data are available at https://github.com/cbaiming/DuplexFM.

bioinformatics↗

bioRxiv2026

Immunometabolic Reprogramming by Thyroid-Stimulating Antibodies Drives Orbital Adipogenesis via Histone Lactylation in Thyroid Eye Disease

Thyroid eye disease (TED) is conventionally viewed as an autoimmune inflammatory disorder. However, the metabolic determinants driving orbital tissue remodeling remain largely undefined. Here, we uncover a paradigm-shifting mechanism whereby thyroid-stimulating antibodies (TSAbs) instigate profound metabolic reprogramming in orbital fibroblasts (OFs), driving a switch toward glycolysis and markedly elevating lactate production via the CREB/LDHA axis. Critically, we demonstrate that this TSAbs-induced metabolic perturbation is not a mere byproduct but the principal driver of pathology. Mechanistically, lactate functions as a signaling metabolite that directly potentiates adipogenesis via histone lactylation-dependent transcriptional activation of the master adipogenic regulators PPAR{gamma} and C/EBP. Collectively, our findings underscore metabolic reprogramming as a central pathogenic mechanism in TED, whereby autoimmune cues drive disease progression through metabolic-epigenetic crosstalk. This work substantiates the paradigm of TED as an immune- metabolic disorder.

immunology↗

bioRxiv2026

Spatial Glyco-Codes Define Human Liver Pathology and Progression

Glycosylation is a fundamental process regulating cellular function, tissue organization, and disease progression. However, comprehensive glycan profiling at single-cell spatial resolution remains largely inaccessible, particularly in clinical archival tissues. Here we develop spatial-GPT, a multimodal platform for simultaneous profiling of glycans, proteins, and/or transcripts in archival formalin-fixed paraffin-embedded (FFPE) tissues. Using a panel of 30 DNA-encoded lectins recognizing major mammalian glycan motifs and structural classes, sequencing-based spatial-GPT (DBiT-GPT) mapped the spatial glycome, proteome, and transcriptome across 16 human liver specimens encompassing steatosis, fibrosis, cirrhosis, and hepatocellular carcinoma (HCC), leading to identification of spatial glyco-codes - combinatorial glycan states associated with distinct cellular identities, tissue features, and pathological processes. Unexpectedly, glyco-codes alone were sufficient to resolve major cell types, disease states, and HCC subtypes, revealing a previously unappreciated level of biological information encoded within the tissue glycome. Spatial glycomics uncovered tumor-like glyco-codes in premalignant regions, suggesting that glycan reprogramming may precede overt malignant transformation. Using imaging-based single-cell spatial glycan-protein profiling (CODEX-GP), we track glyco-codes across the whole-tissue architecture of 3 representative HCC samples. We further examined the glyco-codes across more than 300 patient specimens and quantified cell-type- and disease-specific glyco-codes as well as glycan-defined immune-evasion, T-cell-exhaustion, and steato-fibrotic niches. Together, these findings establish spatial glyco-codes as a previously unrecognized layer of tissue organization that encodes cellular identity, tissue function, and disease progression. The ability of glyco-codes to distinguish major liver pathologies across independent patient cohorts further highlights their potential as a new class of molecular histopathology biomarkers.

pathology↗

bioRxiv2026

Relapse-founding cancer persister cells in follicular lymphoma

Follicular lymphoma (FL) is an incurable, prototypical relapse-remitting cancer, implying the existence of therapy-persistent cells that survive frontline treatment and seed disease recurrence1-3. However, these persister cells remain difficult to study directly in patients because immediate post-treatment sampling is ethically and practically challenging. Using a genetically defined mouse model that allows sampling of persistent cells immediately after frontline R-CHOP therapy, we prospectively isolate and functionally define relapse-founding cancer persister cells (CPC). The CPC is an IgM memory-like B-cell with high germinal center re-entry capacity. This state represents a discrete component of a heterogeneous residual pool indicating that residual disease is polytypic and that relapse potential may depend on which cells persist rather than on residual tumour burden alone. By integrating mouse CPC with human FL datasets, we show that an analogous transcriptional programme is detectable at diagnosis and is enriched in patients with inferior clinical outcome across independent cohorts4,5. These findings support the concept that relapse risk is linked to a conserved, genotype-agnostic CPC programme present before therapy. To explore therapeutic vulnerabilities, we developed a scalable in-vitro platform that models the CPC-like state and used it to identify sensitivity to histone deacetylase inhibition. Romidepsin and panobinostat killed CPC-like cells in-vitro, and decreased therapy-persistent cells after R-CHOP treatment in-vivo and in patient-derived organoids. Together, these data define a tractable CPC state in FL, with a validated clinical readout and an immediately testable therapeutic entry point, opening CPC-directed strategies for durable FL control.

cancer biology↗

bioRxiv2026

Pan-Cancer Driver Mutation Signatures Define a Molecular Taxonomy of Tumors

Cancers with similar histology often exhibit divergent clinical behavior, reflecting molecular heterogeneity not captured by current classification systems. Although driver mutations are central to tumorigenesis, their broader systems-level consequences have not been systematically leveraged. We integrated genomic and transcriptomic data across cancers to define driver mutation signatures (DMS), coordinated transcriptional programs associated with cancer driver mutations. From 121 candidate drivers, we derived 90 robust signatures and quantified their activity in individual tumors using mutation signature scores (MSS). DMS analysis revealed a hierarchical organization of tumors into molecular subgroups that transcended tissue boundaries while preserving driver-associated features. Continuous MSS profiles further defined high-resolution molecular fingerprints for individual tumors. DMS provides a quantitative framework for tumor classification and patient stratification and links driver-associated programs to potential therapeutic vulnerabilities. Together, these findings establish a pan-cancer molecular taxonomy that bridges genotype and phenotype and may inform precision oncology.

cancer biology↗

bioRxiv2026

Towards Passive Dietary Monitoring: Dilated CNN-Based Meal Detection Using Ambulatory High-Resolution Electrogastrography

BackgroundObjective tools for longitudinal dietary monitoring, despite its importance in the health triad of diet, sleep, and exercise, remain limited. Widely available ambulatory options include continuous glucose monitoring, which is a delayed response, and manual logging, which is rife with human error. Clinical tools such as gastric emptying scintigraphy are impractical for everyday use. High-resolution electrogastrography (HR-EGG) offers an alternative by treating gastric myoelectric activity as a biomarker of digestive state. However, its utility for ambulatory meal detection remains unclear. We hypothesize that HR-EGG and accelerometry together encode distinct postprandial gastric signatures to enable automated meal detection, and that postural context represents a relevant source of variation in signal detectability. MethodsHR-EGG and accelerometry data were collected from seven healthy adults across sixteen 150-minute meal sessions under IRB-approved protocol. Each session included a 30-minute fasted baseline, consumption of a standardized meal, and 90-minute postprandial period of sitting, walking, and lying in a randomized order. Features of the gastric slow wave, including raw and normalized bandpower, phase gradient directionality (PGD), wave direction, and wave speed, were extracted alongside triaxial accelerometer magnitude. A dilated one-dimensional convolutional network (1D CNN) was trained to classify meal consumption at five-minute resolution using leave-one-subject-out cross-validation. Postural effects on gastric myoelectric metrics were assessed using the Friedman test. ResultsThe model achieved a mean AUROC of 0.925 (95% CI: [0.857, 0.993]) and mean AUPRC of 0.824 (95% CI: [0.668, 0.980]; null model: 0.20). Feature ablation showed PGD as the most informative input ({Delta}AUPRC = -0.188), with wave propagation speed the least informative ({Delta}AUPRC = -0.105). Walking produced the highest signal-to-noise ratio (9.94 dB), lying had the most stable gastric rhythm (89.1% normogastric), and sitting demonstrated the greatest frequency instability (dominant frequency standard deviation = 0.825 cpm). ConclusionA dilated 1D CNN applied to spatiotemporal HR-EGG features enables temporally aware passive meal detection across ambulatory contexts. This study framework addresses a gap between clinical need for objective dietary monitoring and the limitations of current detection methods.

bioengineering↗