Nature Biomedical Engineering

Deep learning models utilizing longitudinal healthcare data have significantly advanced epidemiological research. However, contemporary transformer-based models increasingly rely on computationally intensive pre-training steps that entail processing massive real-world datasets with cost-prohibitive hardware. We introduce the Temporal Encoder with Late Fusion (TELF), a lightweight end-to-end predictive model featuring an encoder-only architecture for processing medical codes, followed by post-encoder concatenation with demographic variables. TELF learns code embeddings on-the-fly, thereby bypassing the resource-intensive pre-training bottleneck. Furthermore, its late-fusion design preserves the integrity of the temporal attention mechanism before integrating static demographic predictors. We evaluated TELF using an administrative claims database across three distinct cohorts: pancreatic cancer (n=53,661), type 2 diabetes (n=78,756), and heart failure (n=72,540). TELF consistently outperformed traditional machine learning baselines, including XGBoost, LightGBM, and logistic regression. Specifically, TELF achieved AUCs of 0.9150, 0.8199, and 0.8721 for pancreatic cancer, type 2 diabetes, and heart failure, respectively, compared with 0.9044, 0.7908, and 0.8535 for XGBoost and 0.9014, 0.7800, and 0.8466 for logistic regression. Beyond predictive superiority, TELF's isolated temporal attention mechanism enables population-level motif mining. By extracting high-attention temporal sequences, we mapped aggregated patient journey pathways, revealing interpretable clinical trajectories preceding disease onset. Collectively, these results demonstrate that TELF provides a resource-efficient and accessible framework for advanced temporal modeling in clinical and epidemiological research.

Selecting an effective treatment relies on accurately anticipating patient's response to alternative interventions. However, forecasting longitudinal clinical trajectories remains difficult because electronic health records contain heterogeneous, irregularly sampled data over extended time periods. These issues are especially relevant for laboratory measurements, which are central for diagnostics, assessment of therapeutic responses, and tracking disease progression in routine clinical practice. However, existing deep learning methods for counterfactual prediction usually assume regularly sampled data, an assumption incompatible with the irregular, heterogeneous data-generation processes of real-world clinical practice. Here we present the Time-Aware G-Transformer, which integrates causal G-computation with time-aware attention to predict counterfactual outcomes on irregular data. By explicitly conditioning on the timing of future observations and encoding measurement patterns, the model captures temporal dynamics that previous methods overlook. Evaluated on synthetic tumor growth data and on 90,753 cancer patient trajectories from an academic medical center, our approach demonstrates superior long-horizon (> 1 day) prediction accuracy and uncertainty calibration compared to state-of-the-art baselines. These results demonstrate that embedding temporal relations directly into the attention mechanism enables robust integration of patient history data for evaluating potential treatment strategies in personalized medicine.

Sleep disturbances are pervasive, debilitating non-motor symptoms of Parkinson's disease (PD), where sleep spindle deficits directly drive cognitive decline and disease progression. Current adaptive deep brain stimulation (aDBS) for PD is largely limited to motor symptom management, with no established technical foundation for sleep spindle-targeted closed-loop modulation. The functional role of the basal ganglia in human sleep spindle regulation remains incompletely characterized, and no robust cross-subject pipeline exists to decode these transient events from clinically implanted DBS electrodes. Here, we developed a connectomics-guided meta-learning framework for cross-subject sleep spindle decoding and anticipatory prediction, using whole-night synchronized basal ganglia local field potential and polysomnography data from 17 PD patients with bilateral DBS implants. Our framework achieved 92.63% accuracy for concurrent spindle decoding and 83.44% accuracy for 2-second-ahead prediction, with optimal signals localized to the limbic subthalamic nucleus and <50 ms total latency meeting real-time closed-loop requirements. This work defines the neuroanatomical substrate of basal ganglia spindle signaling in PD, establishes the cross-subject spindle decoding pipeline for clinical DBS systems, and provides a critical translational foundation for sleep-targeted closed-loop aDBS to mitigate PD non-motor burden.

Precision neuroimaging aims to deliver individualized assessments of brain health, yet a single structural MRI does not yield a multidimensional, quantitative summary of an individual's current health or future risk. Existing approaches optimize task-specific objectives, yielding representations entangled with cohort- or disease-specific signals rather than capturing biologically grounded patterns of anatomical variation. Here, we introduce NeuroFM, a foundation model trained exclusively on 100,000 healthy synthetic volumes to predict morphometric and demographic targets. Without exposure to diagnostic labels, NeuroFM organizes brain MRIs into population-level patterns that encode meaningful brain health differences. These representations transfer across five neuroscience domains without adaptation and support simple linear readouts for clinical, cognitive, developmental, socio-behavioural, and image quality control. Evaluated on 136,361 real volumes spanning multiple cohorts, NeuroFM generalizes across domains and enables individual-level brain health profiling, estimating future dementia risk years before diagnosis. Together, these findings establish a disease-naive foundation model paradigm for precision neuroimaging.

Allogeneic cell therapies require the coordinated expression of multiple immunomodulatory genes, yet multigene circuits that function in permissive cell lines often fail in differentiated human tissues for unclear reasons. Here, we systematically dissect how transcriptional architecture governs functional immunoregulation in engineered human keratinocyte and fibroblast lines. Using site-specific large-cargo integration (eePASSIGE) as an enabling tool, we determined that genomic insertion efficiency was not the limiting factor for phenotype; rather, promoter arrangement and gene order dictated expression hierarchy. A single-promoter EF1-IDO1-T2A-GFP design that expressed robustly in HEK293T cells was nearly silent in skin-derived cells, preventing reporter-based enrichment. In dual- and tri-modular cassettes, we observed severe transcriptional interference: a downstream CMV promoter driving GFP or PD-L1/iCasp9 (via EMCV-IRES) markedly suppressed the upstream EF1-IDO1 unit, despite intact integration (resulting in [~]175-625-fold attenuation), demonstrating strong promoter interference within the circuit. Functionally, co-culture assays revealed a hierarchical immunomodulatory logic: high IDO1 expression proved to be a requisite threshold for T-cell suppression, whereas PD-L1 provided measurable benefit only against highly activated, PD-1+ T cells in vitro. Collectively, these data establish a site-specific framework for generating immune-tuned skin cells and define essential design rules for avoiding promoter interference in next-generation translational skin substitutes.

Some chimeric antigen receptor (CAR) T cell therapies have shown strong clinical efficacy, yet systematic screening of new CAR designs remains constrained by labor-intensive, low-throughput evaluation methods. To address this limitation, we developed a cytotoxicity-centered, high-throughput screening platform that integrates single-cell pooled screening with fully automated arrayed screening enabling both large-scale library handling and quantitative functional resolution for systematic CAR design exploration. Using a mutation-based CAR design approach guided by protein fitness prediction, we generated a 4-1BB-based CAR library with approximately 10 theoretical variants while minimizing the prevalence of low-activity designs. In pooled screening, CAR T cells were evaluated at the single-cell level based on cytotoxicity and proliferation, enabling rapid enrichment of high-performing variants from a highly diverse library. Subsequent automated arrayed screening quantitatively measured cytotoxicity with high reproducibility, providing high-resolution functional data suitable for comparative ranking. Selected CAR variants demonstrated superior antitumor efficacy in a leukemia xenograft model compared with a template CAR. Furthermore, systematic analysis of mutation sites from an enhanced CAR variant identified essential mutation combinations underlying functional enhancement. Together, this study establishes a cytotoxicity-focused screening framework that provides a robust approach for optimizing CAR architectures and accelerating the development of CAR T-cell therapies.

Longitudinal electronic health records (EHRs) form irregular event sequences that mix multiple clinical coding systems and care settings. Learning transferable patient representations requires modeling both within-encounter code composition and long-range temporal dependencies. We aim to develop a pretraining framework that preserves event structure and explicitly uses elapsed time, while remaining straightforward to fine-tune for new supervised endpoints without task-specific feature engineering. We propose HealthFormer, a dual-level Transformer for event-centric EHR modeling. An Intra-Event Encoder aggregates heterogeneous domain tokens within each typed clinical event into an event embedding via code-specific embedding modules and attention pooling. Event embeddings are combined with a Date Encoder and a continuous-time attention bias based on attention with linear biases (ALiBI) inside an Inter-Event Encoder. We pretrain on Hungarian national administrative health records from a large-scale nationwide longitudinal cohort (spanning millions of individuals over a decade) using multi-task self-supervision with (i) per-domain masked token prediction (masked language modeling, MLM), (ii) event-type prediction under full-event masking (Event-level MLM), (iii) next-event type prediction, and (iv) time-to-next-event ({Delta}t) regression. Pretraining induces hierarchy-consistent organization in learned diagnosis (ICD-10) embedding geometry conducive to analysis and interpretation. On incident cancer prediction, end-to-end fine-tuning achieves test AUCs of 0.81/0.75/0.73 for colorectal cancer (CRC) and 0.94/0.87/0.84 for prostate cancer across 30/60/90-day horizons on balanced cohorts, outperforming logistic-regression baselines, including time-decayed bag-of-codes. HealthFormer provides an event-centric, time-aware representation that transfers via standard fine-tuning without endpoint-specific designs. Using ICD-10 diagnoses and ATC codes can facilitate adoption beyond Hungary. Learned diagnosis embeddings align with the hierarchy, enabling clinical inspection. Broader benchmarking across endpoints remains needed.

Gut microbiome studies in Parkinsons disease (PD) are challenged by high dimensionality, sparsity, compositionality, and substantial between-cohort heterogeneity, all of which complicate robust community typing and disease-status classification. Here, we developed a variational autoencoder (VAE)-based methodology for deep enterotyping and PD diagnosis prediction (i.e., predicting diseased vs. control status) using a harmonized multi-cohort gut microbiome compendium comprising 1,957 16S rRNA samples from six PD case-control cohorts and an independent shotgun metagenomic validation cohort of 725 samples. Compared with conventional enterotyping approaches such as partitioning around medoids (PAM) and Dirichlet multinomial mixture (DMM) modelling, the VAE-derived latent space supported a clearer and more reproducible three-cluster solution. These three enterotype-like community states were biologically interpretable and were annotated as Enterococcus-type, Bacteroides-type, and Ruminococcus-type configurations. The same broad three-enterotype structure was independently recapitulated in the metagenomic dataset, supporting cross-platform robustness. Across the three inferred types, the proportion of PD samples was similar, and both the primary generalized linear mixed-effects model and sensitivity model showed that enterotype assignment was not a significant differentiating factor for PD status and that the lack of association was not dependent on a single modelling strategy. In the supervised branch, VAE-derived representations supported PD case-control classification while also providing a shared latent representation for clustering, enterotype transfer, and downstream interpretation. Collectively, these findings show that deep representation learning can improve the resolution, reproducibility, and interpretability of enterotype inference in heterogeneous microbiome datasets, and provide a practical methodology for organizing broad community structure in PD. In this setting, the main advantage of the VAE method lies in its ability to link unsupervised community typing with supervised prediction through a shared latent representation, even when broad community types do not function as stand-alone disease biomarkers.

Accurately resolving the full spectrum of somatic alterations remains a major barrier in cancer genomics. Current short-read sequencing methods often prioritize SNVs and copy-number changes while overlooking SVs, haplotype-specific events, and epigenetic dysregulation. To bridge this gap, we present TumorLens, the first unified long-read framework that jointly detects SNVs, indels, SVs, large CNVs, loss-of-heterozygosity, and CpG methylation in a single assay. TumorLens introduces purity-aware long-read CNV/LoH modeling and personalized HLA-locus reconstruction, enabling the mechanistic interpretation of immune escape through allele-specific methylation profiling. Benchmarked across GIAB standards and clinical cohorts, TumorLens accurately recovered key somatic events, including interferon locus disruptions and HLA loss. Furthermore, it revealed pervasive global hypomethylation alongside focal hypermethylation in critical oncogenic pathways. By consolidating multi-omic layers into an end-to-end analytic pipeline, TumorLens establishes a new standard for comprehensive tumor profiling, accelerating the translation of long-read sequencing into precision oncology.

Chimeric antigen receptor (CAR) T cells have transformed cancer treatment, yet challenges for achieving broader clinical success remain, including overcoming tumor antigen heterogeneity and limited T cell fitness. To address these challenges and enhance CAR T cell functionality, we leveraged meditope technology, a lock-and-key platform where Fab regions of antibodies are modified to bind a small cyclic peptide termed meditope (meP). We developed a panel of meditope-enabled Fab-based CARs (meCARs), which show selective binding to the meP and comparable activity to traditional single-chain variable fragment (scFv)-based CARs. Focusing on HER2-targeted meCARs for evaluating platform utility, we exploited the modularity of the meditope platform to detect meCAR T cells using meP-fused fluorescent agents, promote meCAR T cell expansion via meP-fused IL-15 cytokine, and broaden tumor antigen targeting through meP-fused antibodies to address tumor heterogeneity. These findings establish the meditope technology as a versatile strategy to augment CAR T cell functionality and overcome key limitations of current CAR-based therapies.

Matrix stiffness serves as a pivotal biophysical cue that profoundly dictates exosome biogenesis and cellular internalization, yet often creates a functional trade-off that impedes clinical translation. Herein, we developed a mechano-chemo-transductive strategy to engineer mesenchymal stem cell (MSC) exosomes endowed with robust biogenesis and superior delivery potency. Specifically, we revealed that MSCs cultured on soft matrices secreted a significantly elevated exosome yield and demonstrated enhanced competence to drive macrophage towards anti-inflammatory M2 polarization. Conversely, stiff matrices upregulated ATP-binding cassette transporter A1 (ABCA1) expression, enriching exosomal membrane cholesterol and facilitating cellular internalization by recipient cells. By taking advantages of these unique mechano-responses, we engineered MSCs via substrate softening combined with ABCA1 modulation to generate mechanochemically reprogrammed exosomes with concurrently enhanced yield and internalization efficiency. In a murine model of pulmonary fibrosis characterized by restrictive biological barriers, inhaled mechanochemically reprogrammed exosomes treatment demonstrated superior lung retention and deep tissue penetration. Furthermore, they effectively orchestrated immune homeostasis by repolarizing alveolar macrophages to reverse fibrotic remodeling and restore lung function. Collectively, by reconciling the intrinsic trade-off between biogenesis and cellular uptake, this strategy represents a paradigm shift in exosome engineering and paves the way for next-generation therapeutics against refractory fibrotic diseases.

Epidermal development and homeostasis require precise coordination between keratinocyte differentiation and mechanics. Still, the mechanisms integrating these processes remain poorly understood in part due to limitations of existing experimental systems. Here, we introduce StrataChip, a tractable microphysiological system that enables dynamic, multimodal interrogation of human epidermal morphogenesis. The platform integrates a media perfused dermal tissue with human epidermal keratinocytes within a microfluidic device and supports rapid epidermal stratification following establishment of an air-liquid interface. High-resolution confocal imaging and single-cell RNA-sequencing demonstrate that the StrataChip recapitulates key architectural and molecular features of human epidermis, including distinct basal, spinous, and granular layers defined by canonical differentiation markers and adhesion molecule organization. Single-cell profiling reveals transcriptionally distinct basal and spinous subpopulations, including transitional states associated with suprabasal commitment. Live 3D imaging in situ captures keratinocyte morphodynamics including basal cell delamination and asymmetric division, linking dynamic cellular behaviors to defined differentiation fates and stratification. Altogether, StrataChip provides a robust platform for a dynamic and mechanistic interrogation of how gene regulation and cell mechanics are coupled during epidermal morphogenesis.

How distinct regional identities emerge within a single developing brain remains poorly understood. Current in vitro models address this by fusing independently generated organoids, but this introduces variability in size, maturation state, and connectivity, confounding the study of regionalization itself. Here, we present a microfluidic platform that supports the co-development of different tissue identities within a single, continuous 3D culture domain. The device integrates controlled microfluidic flow with real-time fluorescence imaging, providing stable perfusion and high-resolution tracking of molecular transport without the need for embedded sensors or disruptive sampling. By delivering SAG, a Sonic hedgehog pathway agonist, to one surface of mouse forebrain organoids, we induced spatially segregated ventral (Nkx2.1+) and dorsal (Pax6+) domains within a unified tissue architecture. Controlled morphogen delivery is sufficient to drive region-specific fate specification without organoid fusion, offering a practical, scalable alternative for studying tissue regionalization in vitro.

Cardiac surgery patients experience rapidly evolving hemodynamics in early post-operative period requiring intensive support. Identifying hemodynamic subphenotypes from these data can inform personalized management. Using 24-hour high-resolution physiologic and treatment data from 6,630 MIMIC-IV and 1,963 SICdb patients, we trained a transformer encoder with a reconstruction-contrastive objective to derive patient-level embeddings capturing multivariate temporal dynamics within first 24h of ICU stay and compared them against those generated by dynamic time warping (DTW). Spectral clustering uncovered three reproducible hemodynamic subphenotypes. Compared with subphenotype 1, subphenotype 3 received more IV fluids, vasopressors, inotropes, and exhibited higher in-hospital mortality (OR 5.85, 95 % CI 2.43-14.13), longer ICU stay (7.12 days, 95% CI: 5.52-8.73) and hospitalization (8.86 days, 95% CI: 6.57-11.16). DTW derived subphenotypes had weaker prognostic separation. Thus, contrastive-transformer framework identified more clinically meaningful temporal hemodynamic subphenotypes that may optimize post-operative risk stratification and inform personalized management.

Persistent infection with high-risk human papillomavirus (HPV) is the primary cause of cervical cancer and other HPV-related malignancies. Effective screening and early detection of HPV, particularly in point-of-care (POC) settings, can reduce disease progression and associated mortality. Although PCR-based assays provide high sensitivity, their dependence on centralized laboratory infrastructure limits accessibility in POC settings. CRISPR-Cas diagnostics enable programmable, isothermal detection of HPV with lateral flow assay (LFA) readouts; however, visual interpretation of faint bands can be subjective and inconsistent. Our objective was to develop a machine learning (ML)-enhanced, smartphone-native CRISPR-LFA platform for highly sensitive and reliable detection of HPV DNA in plasma. A smartphone-based diagnostic system integrating CRISPR-LFA with a ML framework was developed using standardized image acquisition within a light-controlled enclosure. Radiomics-inspired strip features were extracted and analyzed using a multivariable logistic regression model. A total of 150 plasma samples were used for model development and 60 independent samples for validation. An optimized model was developed that had 96.7% sensitivity and 100% specificity for detection of HPV DNA. The smartphone-enabled CRISPR platform demonstrated higher sensitivity than visual interpretation, particularly for faint-band results, and reduced false positives. Validation in the independent cohort confirmed the robustness of the assay. Performance remained stable across smartphone models, lighting conditions, and operators, and on-device inference enabled reliable operation. In sum, the smartphone-integrated CRISPR-LFA platform can facilitate accurate and reliable detection of plasma HPV DNA in POC settings and has the potential to enhance early detection, prevention, and treatment of cervical cancer.

High-plex spectral imaging has the potential to transform the analysis of spatial organization in cells and tissues, yet its practical implementation remains limited by challenges in panel design, sample preparation, signal balancing, and experimental validation. While cyclic imaging approaches are widely used in spatial omics, spectral imaging across the full fluorescence spectrum and computational unmixing remain underutilized due to these challenges. Here, we present a generalizable framework for high-plex spectral imaging that leverages DNA-barcoded labeling and programmable signal amplification to provide precise control over fluorescence signal composition. Orthogonal DNA barcodes decouple target labeling from fluorophore detection, enabling reversible fluorophore application and systematic panel optimization directly on the same sample. Programmable DNA-based amplification further enables independent and quantitative tuning of fluorescence intensities across targets, overcoming a key limitation of spectral unmixing, namely imbalanced signal contributions in overlapping channels, and thereby improving accuracy and robustness. The framework also supports the generation of experiment-specific ground truth datasets and systematic evaluation of unmixing algorithms, providing a quantitative basis for panel validation and performance assessment. We demonstrate the practical implementation of this framework by developing a panel for simultaneous imaging of 15 subcellular structures without fluidic cycling and using the optimized panel to profile the effects of chemical perturbations on subcellular organization. We quantitatively evaluate panel compilation and provide a rigorous assessment unmixing performance using both linear and reference-free unmixing methods. Importantly, we leverage foundation models trained on standard fluorescence data, for segmentation-free, high-dimensional analysis of spectrally unmixed images without needing large datasets or model retraining. Together, we establish a practical and tunable framework for high-plex spectral imaging that lowers experimental barriers and enables broader adoption of spectral unmixing for biological and biomedical applications.

Our understanding of human mucosal T cell clonotype distribution in health and disease has centered on immunodominant antigens. We performed single cell T cell receptor (TCR) and RNA sequencing as an untargeted approach to define distributions of T cell clonal groups in health and ulcerative colitis (UC) across 333,088 T cells in colon and peripheral blood. Healthy donor-specific TCR repertoires had limited blood-colon clonal sharing, which was highest in cytotoxic T effector memory (Tem) populations and lowest in regulatory T cells (Tregs), reflecting tissue-based compartmentalization. Within healthy colon, TCR repertoires showed high T cell clonal sharing independent of anatomic distance, associated with high intra-clonal phenotypic diversity. Colon cytotoxic and Th17 populations showed high dispersion across sites, while Tregs were compartmentalized. Clonal lineages dispersed across blood and colon upregulated trafficking markers, suggesting active movement between tissues, while those dispersed across colon sites upregulated residency markers, suggesting intra-colon repertoire sharing is mediated by long-term, slow moving clonal groups. In UC, Tregs were expanded across inflamed sites, and increased CD8 Tem clonal groups showed increased dispersion regardless of inflammation. These findings reveal principles of T cell clonal organization in the human colon during health and disease, identifying opposing patterns of clonal dispersion among Treg and Th17 clonal groups, high phenotypic diversity within dispersed clonal groups, and elevated cross-colon dispersion of CD8 Tem clonotypes in UC.

We previously developed synthetic Notch (synNotch)-chimeric antigen receptor (CAR)-T cells to improve the safety and efficacy of CAR-T therapy for glioblastoma. In this system, an anti-EphA2/IL13R2-dual-CAR is expressed only upon recognition of tumor- or brain-specific "priming" antigens, EGFRvIII (termed E-SYNC cells) or brevican (B-SYNC), respectively, with E-SYNC currently under phase I clinical evaluation (NCT06186401). However, tracking and profiling these engineered cells in vivo remain challenging, limiting our understanding of their activity and therapeutic potential. To address this gap, we developed a single-cell RNA-sequencing (scRNA-seq) workflow with custom spike-in probes for synNotch-CAR transcripts, enabling simultaneous detection of engineered cells and transcriptomic profiling. In vitro, integration of multiple probes using machine-learning-assisted classifiers detected 78.2% of E-SYNC cells and 60.0% of B-SYNC cells with 98.0% specificity. In a xenograft model, synNotch-positive cells were detected across the spleen, lung, and brain, with the highest frequency and most robust priming and activation observed in the brain. Single-cell transcriptomic analyses revealed tissue-specific differentiation programs, including cytotoxicity, proliferation, metabolic activity, and acquisition of tissue-resident memory phenotypes, shaped by both environmental cues and synNotch-mediated antigen recognition. In summary, this spike-in probe-enhanced scRNA-seq workflow enables robust detection and high-resolution characterization of synNotch-CAR-T cell dynamics and provides a broadly applicable platform for monitoring engineered immune cells in diverse clinical contexts. One Sentence SummaryOur spike-in probe-enhanced single-cell RNA-sequencing method enables analysis of tissue-dependent activation and transcriptional states of synNotch-CAR-T cells, providing a robust and scalable platform for in vivo tracking and transcriptomic profiling of engineered cell therapies.

Three-dimensional (3D) handheld photoacoustic tomography typically relies on bulky and expensive external positioning trackers to correct motion artifacts, which severely limits its clinical flexibility and accessibility. To address this challenge, we present PA-SfM, a tracker-free framework that leverages exclusively single-modality photoacoustic data for both sensor pose recovery and high-fidelity 3D reconstruction via differentiable acoustic radiation modeling. Unlike traditional Structure-from-Motion (SfM) methods that formulate pose estimation as a geometry-driven optimization over visual features, PA-SfM integrates the acoustic wave equation into a differentiable programming pipeline. By leveraging a high-performance, GPU-accelerated acoustic radiation kernel, the framework simultaneously optimizes the 3D photoacoustic source distribution and the sensor array pose via gradient descent. To ensure robust convergence in freehand scenarios, we introduce a coarse-to-fine optimization strategy that incorporates geometric consistency checks and rigid-body constraints to eliminate motion outliers. We validated the proposed method through both numerical simulations and in-vivo rat experiments. The results demonstrate that PA-SfM achieves sub-millimeter positioning accuracy and restores high-resolution 3D vascular structures comparable to ground-truth benchmarks, offering a low-cost, softwaredefined solution for clinical freehand photoacoustic imaging. The source code is publicly available at https://github.com/JaegerCQ/PA-SfM.

Vascular flow delivers nutrients and imposes hemodynamic forces that govern vessel behavior in health and disease, yet fully human systems that recapitulate and tune physiological intraluminal flow in three-dimensional (3D) tissues are lacking. We developed VIVOS (Vascularized In Vitro Organ Systems), a platform that couples perfused human vascular beds to tunable pumps, generating continuous intraluminal flow through millimetre-scale vessels and 3D tissues at physiological shear stresses and pressures. VIVOS supports integration and perfusion of diverse human organoids and tissues, including lung organoids, cerebral organoids, vascular organoids, breast spheroids, and human retinal explants, as well as enables direct measurement and control of pressure, shear stress, and perfusion-dominant compound transport over extended culture periods. By tuning intraluminal flow and applying single-cell transcriptomics, we uncover a remodeling program in which laminar shear stress acts through a YAP/TAZ-TEAD "switch" to rewire an Apelin ligand-receptor axis and bias tip-stalk endothelial states, reshaping human vascular networks and linking hemodynamic cues to cell state transitions. We further model fast-flow arteriovenous malformations (AVMs) from Hereditary Hemorrhagic Telangiectasia and show that BMP9 constrains vessel caliber and perfusion while antagonizing a VEGF-driven angiogenic program, generating flow-quantified AVM-like lesions in a fully human 3D context. Together, these findings establish VIVOS as a generalizable platform that links physiological intraluminal flow to endothelial state transitions and vessel remodeling, enabling preclinical testing and mechanistic dissection of flow-regulated vascular pathologies in perfused 3D human tissues under defined hemodynamic conditions.