N Nature Biomedical Engineering · Nov 28, 2025 mRNA engineering of allogeneic mesenchymal stem cells enables coordinated delivery of T cell engagers and immunotherapeutic cues Allogeneic cell therapies can enable off-the-shelf products that address limitations of autologous therapies. Mesenchymal stem cells are a robust allogeneic source, but no bioengineered mesenchymal stem cell-based therapies exist. Here we use mRNA engineering to create an off-the-shelf immunotherapy that we term DC-25. DC-25 consists of a mesenchymal stem cell armed with three designed mRNA constructs encoding CXCR4 to direct migration, a T cell engager specific for B cell maturation antigen to target B cell maturation antigen-expressing plasma cells involved in cancer and autoimmunity, and interleukin-12 to potentiate pro-immune responses. DC-25 allows tunable expression of each gene, supporting a predictable pharmacokinetic profile. In vitro, DC-25 exhibits synergistic killing of target cells, and in a preclinical in vivo myeloma model, this therapy exhibits potent efficacy that surpasses T cell engager protein infusion. In a phase 1 safety study in patients with myeloma, DC-25 appears safe and generates interleukin-12 production after each infusion. This study motivates human cell therapies that exploit mRNA to achieve efficacy through induction of secreted or surface-bound therapeutic elements. Biomedical engineering Cell delivery Immunotherapy Translational research biology mouse experiments
N Nature Biomedical Engineering · Nov 06, 2025 A vision–language pretrained transformer for versatile clinical respiratory disease applications General artificial intelligence models have unique challenges in clinical practice when applied to diverse modalities and complex clinical tasks. Here we present MedMPT, a versatile, clinically oriented pretrained model tailored for respiratory healthcare, trained on 154,274 pairs of chest computed-tomography scans and radiograph reports. MedMPT adopts self-supervised learning to acquire medical insights and is capable of handling multimodal clinical data and supporting various clinical tasks aligned with clinical workflows. We evaluate the performance of MedMPT on a broad spectrum of chest-related pathological conditions, involving common medical modalities such as computed tomography images, radiology reports, laboratory tests and drug relationship graphs. MedMPT consistently outperforms the state-of-the-art multimodal pretrained models in the medical domain, achieving significant improvements in diverse clinical tasks. Extensive analysis indicates that MedMPT effectively harnesses the potential of medical data, showing both data and parameter efficiency and offering explainable insights for decision-making. MedMPT highlights the potential of multimodal pretrained models in the realm of general-purpose artificial intelligence for clinical practice. Biomedical engineering Machine learning Respiratory tract diseases Machine Learning Clinical Human Respiratory Disease
N Nature Biomedical Engineering · Nov 03, 2025 Nonexpansive biodegradable matrix promotes blood vessel organoid development for neurovascular repair and functional recovery in ischaemic stroke Tissue engineering-based vascular reconstruction represents a promising therapeutic strategy for ischaemic stroke. However, in the confined stroke cavity, conventional implants are unable to simultaneously provide swelling-resistant support and growth-permissive internal space, which are crucial for effective revascularization. To address this limitation, we develop a bioinspired, non-expansive biodegradable matrix (NEBM) through covalent–non-covalent assembly of commercially available, clinical-grade natural polymers. We show that NEBM recapitulates key features of brain extracellular matrix—including porous microstructure and tissue-matched stiffness—to deliver structural stability. Moreover, its progressively degradable structure establishes a dynamic remodelling niche that directs cellular behaviour towards promoting angiogenesis. Compared with commercial Matrigel-based matrix, NEBM fosters blood vessel organoid development with higher vascular density, larger vessel diameters and more distinct arterial features. In both subcutaneous and stroke transplantation models, we find that NEBM facilitates the integration of blood vessel organoids with the host vasculature. Strikingly, this revascularization in stroke cavity stimulates neurogenesis, contributing to significant functional recovery. As such, our study provides valuable guidance to design clinically translatable matrices for organ repair and regeneration in confined environments. Bioinspired materials Biomedical engineering Gels and hydrogels Implants biology mouse experiments
N Nature Biomedical Engineering · Nov 03, 2025 Engineered human induced pluripotent stem cell models reveal altered podocytogenesis in congenital heart disease-associatedSMAD2mutations Clinical observations of patients with congenital heart disease carryingSMAD2genetic variants revealed correlations with multi-organ impairments at the developmental and functional levels. Many patients with congenital heart disease present with glomerulosclerosis, periglomerular fibrosis and albuminuria. It remains largely unknown whetherSMAD2variants associated with congenital heart disease can directly alter kidney cell fate, tissue patterning and organ-level function. Here we investigate the role of pathogenicSMAD2variants in podocytogenesis, nephrogenic cell lineage specification and glomerular filtration barrier function using a combination of CRISPR-based disease modelling, stem cell and microfluidic organ-on-a-chip technologies. We show that the abrogation ofSMAD2results in altered patterning of the mesoderm and intermediate mesoderm cell lineages, which give rise to nearly all kidney cell types. Following further differentiation of intermediate mesoderm cells, the mutant podocytes failed to develop arborizations and interdigitations. A reconstituted glomerulus-on-a-chip system showed substantial albumin leakage, as observed in glomerulopathies. This study implicates chronic heart disease-associatedSMAD2mutations in kidney tissue malformation that might inform targeted regenerative therapies. Biomedical engineering Development Paediatric kidney disease Stem-cell differentiation Urological manifestations biology
N Nature Biomedical Engineering · Oct 29, 2025 Iontronic tip-sensing guidewires Plaque accumulation in coronary arteries causes stenoses, reducing blood flow and increasing the risk of cardiovascular disorders such as heart attacks. To assess the physiological impact of blood pressure across a stenosis, commercial pressure guidewires measure the fractional flow reserve using optical, piezoresistive or piezoelectric sensors, which suffer from brittleness, limited manoeuvrability and high costs. Here we report an iontronic tip-sensing guidewire (ITG) that integrates a thin iontronic tip sensor in a commercial workhorse guidewire via iontronic-based signal transmission, leveraging the ionic nature of human tissues. Intravascular pressure changes induce a capacitance difference at the interface of the metal and ionic gel of the ITG, allowing detection of subtle pressure changes in blood flow, substantially outperforming commercial guidewires. The ITG is free of embedded conductive leads needed in other pressure guidewires to ensure an ideal torque ratio for high manoeuvrability, and we validated its effectiveness and sensitivity in rabbit, goat and pig models in vivo. The compatibility of the ITG with commercial horsework guidewires will upgrade the design of interventional medical devices. Biomedical engineering Ischaemia Cardiovascular Iontronic Sensing Guidewire Animal Models
N Nature Biomedical Engineering · Oct 24, 2025 Implanted microelectrode arrays in reinnervated muscles allow separation of neural drives from transferred polyfunctional nerves Targeted muscle reinnervation surgery reroutes residual nerve signals into spare muscles, enabling the recovery of neural information through electromyography (EMG). However, EMG signals are often overlapping, making the interpretation of limb functions complicated. Regenerative peripheral nerve interfaces surgically partition the nerve into individual fascicles that reinnervate specific muscle grafts, isolating distinct neural sources for precise control and interpretation of EMG signals. Here we combine targeted muscle reinnervation surgery of polyvalent nerves with a high-density microelectrode array implanted at a single site within a reinnervated muscle, and via mathematical source separation methods, we separate all neural signals that are redirected into a single muscle. In participants with upper-limb amputation, the deconvolution of EMG signals from four reinnervated muscles into motor unit spike trains revealed distinct clusters of motor neurons associated with diverse functional tasks. Our method enabled the extraction of multiple neural commands within a single reinnervated muscle, eliminating the need for surgical nerve division. This approach holds promises for enhancing control over prosthetic limbs and for understanding how the central nervous system encodes movement after reinnervation. Biomedical engineering Computational neuroscience Machine learning Microarrays Motor neuron biology mouse experiments
N Nature Biomedical Engineering · Sep 24, 2025 Invasive neurophysiology and whole brain connectomics for neural decoding in patients with brain implants Brain–computer interface research can inspire closed-loop neuromodulation therapies, promising spatiotemporal precision for the treatment of brain disorders. Decoding dynamic patient states from brain signals with machine learning is required to leverage this precision, but a standardized framework for invasive brain signal decoding from neural implants does not exist. Here we develop a platform that integrates brain signal decoding with magnetic resonance imaging connectomics and demonstrate its use across 123 h of invasively recorded brain data from 73 neurosurgical patients treated with brain implants for movement disorders, depression and epilepsy. We introduce connectomics-informed movement decoders that generalize across cohorts with Parkinson’s disease and epilepsy from the United States, Europe and China. We reveal network targets for emotion decoding in left prefrontal and cingulate circuits in deep brain stimulation patients with major depression. Finally, we showcase opportunities to improve seizure detection in responsive neurostimulation for epilepsy. Our study highlights the clinical use of brain signal decoding for deep brain stimulation and provides methods that allow for rapid, high-accuracy decoding for precision medicine approaches that can dynamically adapt neurotherapies in response to the individual needs of patients. Biomedical engineering Computational neuroscience Neuroscience Machine Learning Human Clinical
N Nature Biomedical Engineering · Sep 24, 2025 An artificial cilia-based array system for sound frequency decoding and resonance-responsive drug release Hair cells in the human ear contain cilia of varying lengths that sense varied acoustic signals. Here, inspired by this, we report an artificial cilia-based sound-decoding device capable of directly recognizing and responding to sound frequencies without relying on electricity and algorithms. We create 3D-printed micrometre-sized (40–200 μm) artificial cilia-based arrays with varying length-to-diameter ratios (30–100) that can sense and decode sound frequency signals (100–6,000 Hz), including piano music and human voices, on the basis of acoustic resonance. The artificial cilia can also vibrate accordingly in water to initiate subsequent tasks such as controlling drug release profiles of two distinct therapeutics (insulin and glucagon) in an acoustic-frequency-responsive manner to treat type 1 diabetic mice. This cochlear cilia-inspired device holds potential for broad applications such as recognizing complicated physiological sounds and performing various tasks in personalized voice interactions. Biomedical engineering Biomimetics Mechanical engineering Neuroscience Drug Development Mouse
N Nature Biomedical Engineering · Sep 23, 2025 A collaborative large language model for drug analysis Large language models (LLMs), such as ChatGPT, have substantially helped in understanding human inquiries and generating textual content with human-level fluency. However, directly using LLMs in healthcare applications faces several problems. LLMs are prone to produce hallucinations, or fluent content that appears reasonable and genuine but that is factually incorrect. Ideally, the source of the generated content should be easily traced for clinicians to evaluate. We propose a knowledge-grounded collaborative large language model, DrugGPT, to make accurate, evidence-based and faithful recommendations that can be used for clinical decisions. DrugGPT incorporates diverse clinical-standard knowledge bases and introduces a collaborative mechanism that adaptively analyses inquiries, captures relevant knowledge sources and aligns these inquiries and knowledge sources when dealing with different drugs. We evaluate the proposed DrugGPT on drug recommendation, dosage recommendation, identification of adverse reactions, identification of potential drug–drug interactions and answering general pharmacology questions. DrugGPT outperforms a wide range of existing LLMs and achieves state-of-the-art performance across all metrics with fewer parameters than generic LLMs. Biomedical engineering Health care Machine Learning Drug Development Clinical
N Nature Biomedical Engineering · Sep 23, 2025 An immune-competent lung-on-a-chip for modelling the human severe influenza infection response Severe influenza affects 3–5 million people worldwide each year, resulting in more than 300,000 deaths annually. However, standard-of-care antiviral therapeutics have limited effectiveness in these patients. Current preclinical models of severe influenza fail to accurately recapitulate the human immune response to severe viral infection. Here we develop an immune-competent, microvascularized, human lung-on-a-chip device to model the small airways, successfully demonstrating the cytokine storm, immune cell activation, epithelial cell damage, and other cellular- and tissue-level human immune responses to severe H1N1 infection. We find that interleukin-1β and tumour necrosis factor-α play opposing roles in the initiation and regulation of the cytokine storm associated with severe influenza. Furthermore, we discover the critical stromal–immune CXCL12–CXCR4 interaction and its role in immune response to infection. Our results underscore the importance of stromal cells and immune cells in microphysiological models of severe lung disease, describing a scalable model for severe influenza research. We expect the immune-competent human lung-on-a-chip device to enable critical discoveries in respiratory host–pathogen interactions, therapeutic side effects, vaccine potency evaluation, and crosstalk between systemic and mucosal immunity in human lung. Biomedical engineering Influenza virus Tissue engineering Immunology Human Drug Development Clinical
N Nature Biomedical Engineering · Sep 22, 2025 Bead-based approaches for increased sensitivity and multiplexing of CRISPR diagnostics CRISPR-based diagnostics have emerged as a promising tool for fast, accurate and portable pathogen detection. There has been rapid progress in pre-amplification processes and CRISPR-related enzymes used in these approaches, but the development of reporter systems and reaction platforms has lagged behind. In this paper, we develop bead-based techniques to address these gaps. First, we develop a novel bead-based split-luciferase reporter system with up to 20× sensitivity compared with standard fluorescence-based reporter design in CRISPR diagnostics. Second, we develop a highly deployable, bead-based platform capable of detecting nine distinct viral targets in parallelized, droplet-based reactions, with sensitivity reaching as low as 2.5 copies per µl of input RNA. We demonstrate the enhanced performance of both approaches on synthetic and clinical sample sensitivity, speed, multiplexing and deployability. Assay systems Biomedical engineering Biotechnology CRISPR Microbiology Clinical Genomics Drug Development
N Nature Biomedical Engineering · Sep 18, 2025 Systematic functional screening of switchable aptamer beacon probes Immunoassays using affinity binders such as antibodies and aptamers are crucial for molecular biology. However, the advancement of analytical methods based on these affinity probes is often hampered by complex operational steps that can introduce errors, particularly in intricate environments such as intracellular settings and microfluidic systems. There is growing interest in developing molecular probes for wash-free assays that activate signals upon target detection. Here we report a systematic functional screening platform for switchable aptamer beacon probes that can achieve target-responsive detection. A stem–loop, hairpin-shaped beacon library was constructed on microbeads and screened using target-responsive fluorescence-activated sorting. The selected aptamer beacons exhibit strong affinities, triggering fluorescence only upon binding, thus enabling wash-free immunoassays for the detection of intracellular and membrane proteins. Computational modelling offers insights into aptamer binding and structural switching mechanisms, revealing how specific protein–aptamer interactions drive stem–loop unwinding and postbinding conformational changes critical for functional activation. This approach establishes a standardized platform for generating switchable aptameric tools, supporting their potential in advanced diagnostics and research. Biochemical assays Biomedical engineering Immunology Proteomics Machine Learning Human
N Nature Biomedical Engineering · Sep 16, 2025 Label-free metabolic imaging monitors the fitness of chimeric antigen receptor T cells Chimeric antigen receptor (CAR) T cell therapy for solid tumours is challenging because of the immunosuppressive tumour microenvironment and a complex manufacturing process. Cellular manufacturing protocols directly impact CAR T cell yield, phenotype and metabolism, which correlates with in vivo potency and persistence. Although metabolic fitness is a critical quality attribute, how T cell metabolic requirements vary throughout the manufacturing process remains unexplored. Here we use optical metabolic imaging (OMI), a non-invasive, label-free method to evaluate single-cell metabolism. Using OMI, we identified the impacts of media composition on CAR T cell metabolism, activation strength and kinetics, and phenotype. We demonstrate that OMI parameters can indicate cell cycle stage and optimal gene transfer conditions for both viral transduction and electroporation-based CRISPR/Cas9. In a CRISPR-edited anti-GD2 CAR T cell model, OMI measurements allow accurate prediction of an oxidative metabolic phenotype that yields higher in vivo potency against neuroblastoma. Our data support OMI as a robust, sensitive analytical tool to optimize manufacturing conditions and monitor cell metabolism for increased CAR T cell yield and metabolic fitness. Biomedical engineering Translational immunology Immunology Cancer CRISPR Human Drug Development
N Nature Biomedical Engineering · Aug 28, 2025 Distributed battery-free bioelectronic implants with improved network power transfer efficiency via magnetoelectrics Networks of miniature implants could enable simultaneous sensing and stimulation at different locations in the body, such as the heart and central or peripheral nervous system. This capability would support precise disease tracking and treatment or enable prosthetic technologies with many degrees of freedom. However, wireless power and data transfer are often inefficient through biological tissues, particularly as the number of implanted devices increases. Here we show that magnetoelectric wireless data and power transfer supports a network of millimetre-sized bioelectronic implants in which system efficiency improves with additional devices. We demonstrate wireless, battery-free networks ranging from one to six implants, where the total system efficiency increases from 0.2% to 1.3%, with each node receiving 2.2 mW at 1 cm distance. We show proof-of-concept networks of miniature spinal cord stimulators and cardiac pacing devices in large animals via efficient and robust wireless power transfer. These magnetoelectric implants provide a scalable network architecture of bioelectronic implants for next-generation electronic medicine. Biomedical engineering Electrical and electronic engineering Neuroscience Machine Learning Human Drug Development
