The development of mechanically tunable hydrogels that replicate the dynamic mechanoelastic properties of native extracellular matrices (ECMs) is essential for advancing 3D tissue engineering. DNA, with its precise, programmable architecture and exceptional control at the nanometre scale, offers a valuable platform for designing ECM-mimicking scaffolds. This study presents stiffness-tuneable DNA supramolecular hydrogels with different branching architectures for programming cellular and organellar states. Utilizing precise DNA motifs-including DX (Double Crossovers), PX (Paranemic Crossovers), and Tensegrity architectures-we engineer hydrogels with widely adjustable mechanical properties (50-185 kPa) without chemical additives or enzymatic crosslinking. These hydrogels exhibit excellent strain-bearing and load-bearing capacity, making them suitable for biomedical applications. Additionally, these DNA hydrogels influence cellular behaviour in retinal pigment epithelial (RPE1) cells by enhancing cellular adhesion, encouraging elongation (a 3-8-fold increase in area compared to control), and improving viability (dependent on concentration, 1-8-fold increase vs. control), while also maintaining organellar homeostasis, including mitochondrial fragmentation and ER stress reduction. This work presents a framework for automating the production of stiffness-tunable DNA hydrogel scaffolds, aligning with the mechanical needs of various cells and tissues, thereby advancing personalized, high-throughput tissue engineering platforms.
Low back pain (LBP) is the leading cause of disability worldwide, yet clinical imaging remains largely limited to anatomical assessment, providing little insight into the spinal tissue mechanics underlying most idiopathic cases. This review highlights emerging noninvasive imaging technologies that enable in vivo quantification of intervertebral disc and spinal muscle mechanics, including radiography, ultrasound imaging, ultrasound elastography, magnetic resonance imaging, and magnetic resonance elastography. These approaches move beyond static morphology to capture spinal kinematics, load-dependent deformation, and tissue material properties under physiologically relevant conditions. Despite substantial technical progress, translation is hindered by inter-individual variability, limited symptomatic cohorts, and challenges in separating age-related changes from pathology. We discuss opportunities to accelerate clinical impact through development of normative mechanical datasets, dynamic and load-dependent imaging paradigms, and integration of imaging-derived mechanical biomarkers with computational modeling and machine learning. Together, these innovations position mechanics-based imaging to enable objective diagnosis, improved patient stratification, and mechanism-driven treatment of low back pain.
Considerations around model retraining are standard practice in industry and non-healthcare sectors; however, this is much less well explored in medical artificial intelligence (AI). The problem is not only that models often fail to generalise, but that academia in particular does not have a systematic science of retraining to address this gap. This matters for building trustworthy models capable of making a lasting impact, rather than compounding as research waste. In this Perspective, we highlight three common challenges that constrain model retraining in medicine, and argue that academia must evolve beyond a focus on developing proofs-of-concept and world-first innovations to also recognise model retraining as scholarship. Drawing from case examples in ophthalmology, we call on stakeholders to consider not just how we build AI models, but how we should retrain, maintain, and share them.
In this review, we systematically categorize diverse organoid engineering strategies-including cellular programming, material engineering, and platform- or system-level innovations-according to their impact on reproducibility and scalability, and highlight representative applications and emerging directions. By reframing organoid generation as a manufacturing process, these technological advances pave the way toward standardized and high-fidelity organoid production for both fundamental research and translational applications.
The human endometrium is a vital component of the female reproductive system that is essential for fertilization, embryo development and female health. However, due to significant ethical concerns and practical limitations associated with human subject research and species differences in animal models, it is highly required to develop in vitro biomimetic human models to facilitate the understanding of physiology and pathology of endometrium in biomedical research. In this review, we highlight recent progress in bioengineered technologies, including organs-on-chips, organoids, advanced biomaterials and bioprinting that enable the reconstruction of functional endometrial models in vitro. We summarize various bioengineering strategies developed to recapitulate key features of the human endometrium in both healthy and diseased states. Furthermore, we introduce the application of these in vitro models in studies of reproductive biology, pregnancy processes and disease mechanisms. Finally, we discuss current challenges and future opportunities in the development of more sophisticated in vitro human endometrial models for biomedical research.
Osteoarthritis (OA) is a leading cause of chronic pain and disability, and lacks disease-modifying therapies. Unlike conventional hydrogels, which suffer from poor mechanical robustness and limited retention, cryogels overcome existing limitations through their interconnected pore architecture, shape-memory behavior, and fatigue-resistant mechanics. Here, we review how cryogels are being engineered as platforms for stem cell delivery, bioactive molecule release, localized gene activation, and immunomodulation, and discuss key translation challenges.
Digital health interventions (DHIs) are increasingly used to strengthen patient engagement. However, despite its rapid growth, DHIs remain unevenly evaluated and poorly standardized. Six databases were searched to generate an evidence gap map of interventions and research gaps for DHIs targeting patient engagement. A total of 160 systematic reviews (including 42 meta-analyses) comprising 3974 primary studies were mapped with most (92%) conducted in high-income countries. Evidence was concentrated around mHealth, eHealth, telehealth, and messaging technologies. Commonly reported outcomes included medication adherence, quality of life, implementation outcomes, and self-management. Overall, 61% of reviews reported positive conclusions, although most were rated as low or critically low methodological quality. Priority areas include strengthening evidence from LMICs, evaluating long-term and emerging DHIs (e.g., wearables and gamified platforms), and improving methodological rigor for systematic reviews. The findings highlight disparities in the global DHI evidence base and identify priority gaps for future research and implementation.
Chronic and acute skin wounds affect more than six million people in the United States each year. Many heal with basic care, but others do not and become infected, scarred, or chronic. These wounds reduce quality of life and increase healthcare costs. Bioelectronic integrated wound dressings and smart bandages can improve healing by delivering treatments locally and on demand, including electric field therapy and the release of pharmacological compounds. Localized bioelectronic delivery improves healing outcomes and reduces off-target effects. Here, we demonstrate a flexible bioelectronic wound dressing that combines electric field therapy and drug delivery, employing integrated microfluidics to switch between therapeutic modalities. Treatment with the bioelectronic dressing in a pilot porcine wound study showed promising results, including increased wound closure rates, improved tissue maturity, and reduced inflammatory response, compared with standard of care.
Stroke often causes persistent upper limb and hand motor dysfunction due to disrupted neural reorganization. To address this, we developed the Magnetic NeuroRing: a portable brain-computer interface integrating real-time electroencephalogram (EEG) with closed-loop continuous theta burst stimulation (cTBS) for adaptive transcranial magnetic stimulation (TMS). A multi-channel EEG array over motor cortical regions (FC3, FC4, CP3, CP4, FT7, FT8, TP7, TP8) detects event-related desynchronization (ERD), indicating motor intent. When ERD/ERS falls below a threshold (ERD/ERS < 0 over five consecutive activations), the system delivers inhibitory cTBS to hyperactive regions, aiming to rebalance stroke-impaired interhemispheric dynamics. The lightweight, patient-specific headgear uses magnetic levitation for precise targeting and EEG-TMS synchronization. In healthy subjects, adaptive cTBS significantly modulated resting-state and task-related neural metrics, aligning with prior large-device findings and demonstrating feasibility for inducing neuroplastic changes. By bridging real-time diagnostics with targeted neuromodulation, the Magnetic NeuroRing enables dynamic, data-driven rehabilitation across clinical and home settings.
Computed tomography pulmonary angiography (CTPA) is the gold standard for pulmonary embolism (PE) diagnosis, but patients with iodinated contrast allergies or renal insufficiency are often ineligible. CT-derived perfusion (CTP) is a novel, non-contrast method to quantify pulmonary perfusion from an inhale/exhale CT image pair (4DCT). The resulting CT-P information can be used to identify hypo-perfused regions associated with PE. This pilot study introduces a thresholding approach that estimates the number of lung lobes with perfusion deficits according to optimally selected CTP thresholds. The number of lobes indicated as low-functioning provides a score to categorize patients as PE-positive, negative, or inconclusive. We trained and validated the model on a retrospective dataset of 123 suspected PE patients, achieving 72% accuracy, 75% sensitivity, and 69% specificity, with 17% of cases inconclusive. To our knowledge, this is the first PE diagnostic model from non-contrast 4DCT, showing the feasibility of non-contrast PE diagnosis strategies.
Personalized exoskeleton assistance has substantial potential to enhance human locomotion performance. However, current human-in-the-loop optimization methods for generating personalized assistance are cumbersome and time-consuming. Since humans can perceive locomotion through internal sensory feedback, user preference-based self-tuning may facilitate the individualization of exoskeleton assistance to meet individual needs. Here, we explore a user-driven human-in-the-loop tuning approach for walking assistance, hypothesizing that individuals can quickly find their preferred personalized assistance through subjective perception. We conducted experiments with 11 healthy participants, who were instructed to tune four control parameters while wearing a hip exoskeleton and walking on a treadmill. The tuning procedure concluded when participants indicated that they had found their preferred assistance. Then we surveyed the sense of agency to assess the user experience. We evaluated the effort of walking with the preferred setting and explored the metabolic cost landscape around it. Participants identified their preference in 10.9 ± 0.9 min, while testing 30.5 settings and spending 18.7 s per setting on average. Preferred assistance profiles varied widely between participants, with timing differences of up to 22.5% of the stride time. The metabolic cost of walking with the preferred assistance was reduced by 16.6 ± 1.1% compared to walking with the exoskeleton in a zero-torque condition. Timing deviations of up to ±8% of the stride time did not significantly affect metabolic cost reduction, indicating the robustness of the preferred assistance profiles. Significant changes in the sense of agency between unassisted and assisted walking demonstrate its sensitivity to partial exoskeleton assistance. The results highlight the potential of preference-based user-tuning while suggesting that additional guidance throughout the user-tuning procedure may support a systematic exploration, thereby advancing the preference-based individualization of exoskeleton assistance.
Pathway enrichment analysis (PEA) of omics data identifies significant pathway-molecule associations, yet delivers results as tabular lists in which complex systems-biology insights remain inaccessible. Hyperpathway is an open-access network-based webtool that addresses this limitation through three original innovations: (1) conversion of a PEA results table into a pathway-molecule bipartite network; (2) a minimal artificial linking strategy to resolve structural disconnections; (3) a leaf removal and post-hoc reinsertion pipeline that accelerates coalescent embedding without any loss of geometric fidelity. The resulting network is visualized in a two-dimensional hyperbolic disk with flexible coloring schemes encoding hierarchical relevance, connectivity similarity, statistical significance, or user-defined annotations; revealing latent functional modules that are invisible in conventional tabular outputs. Validated on genomic, metabolomic, and lipidomic datasets, Hyperpathway enables a deeper, systems-level understanding of the interplay between pathways and their molecular components, providing insights that go beyond p-value-based significance testing. Beyond PEA, Hyperpathway can be used as a general-purpose open webtool for fast hyperbolic embedding and interactive visualization of any bipartite network.
Microphysiological models recapitulating complex vasculature have enabled in vitro evaluation of intricate cellular crosstalk and microenvironmental mechanical cues and stressors. Here, we present a comprehensive review of time and architecture as two under-explored facets required for physiological mimetic vascular models. We conclude with our perspective that incorporating longevity and architecture will improve complex vascular modeling and provide a platform to support chronic disease modeling, personalized medicine, drug testing, and tissue engineering.
The HeMonitor study evaluated the feasibility and accuracy of non-invasive hemoglobin (Hb) assessment using image-based techniques and machine learning in patients with hematologic malignancies. A total of 367 patients with hematologic malignancies and 184 healśśthy donors were enrolled, with fingernail and eyelid photographs collected and analyzed using Light Gradient-Boosting Machine (LightGBM) regression models. The best-performing model achieved a residual standard deviation of ±1.02 mmol/L for Hb prediction. Our framework further explored a two-stage concept combining (i) a non-invasive image-based Hb predictor and (ii) a post hoc, rule-basśed corridor aggregation layer integrating EORTC Global Health and Fatigue categories. This exploratory layer was designed to contextualize Hb estimates with patient-reported symptom burden and well-being. Visual analyses suggested that lower Hb levels were generally associated with impaired quality-of-life measures, consistent with the known clinical burden of anemia. Within the QoL subset, the integrated framework showed encouraging concordance with clinician assessments, particularly in borderline Hb ranges. These findings support the feasibility of combining digital biomarkers with patient-reported outcomes for future patient-centered home monitoring strategies, while prospective validation remains necessary.
The electrical stimulation of the nervous system has shown great clinical potential in injury and pathology, yet experimentally driven practice makes it challenging to identify effective design choices and personalized stimulation protocols. This review outlines emerging model-based optimization frameworks that address these challenges by leveraging biophysical digital twins of neural interfaces. Enabling acceleration strategies and complementary data-driven approaches are also highlighted, along with key factors that currently limit clinical translation.
Brain-computer interfaces (BCIs) are evolving from research prototypes into clinical, assistive, and performance enhancement technologies. Despite the rapid rise and promise of implantable technologies, there is a need for better and more capable wearable and non-invasive approaches whilst also minimising hardware requirements. We present a non-invasive BCI for iterative selection-based mind-drawing that infers a subject's internal visual intent through iterative selection of adaptive visual probes presented on a screen encoded at different flicker-frequencies and analyses the steady-state visual evoked potentials (SSVEPs). Gabor-inspired or machine-learned policies dynamically update the spatial placement of the visual probes on the screen to explore the image space and reconstruct simple imagined shapes within approximately two minutes or less using just single-channel EEG data. Additionally, by leveraging stable diffusion models, reconstructed mental images can be transformed into realistic and detailed visual representations. Whilst we expect that similar results might be achievable with e.g. eye-tracking techniques, our work shows that symbiotic human-AI interaction can increase BCI bit-rates by more than a factor 5x, providing a platform for future development of AI-augmented BCI.
In vitro studies of intestinal fibrosis are confounded by spontaneous fibroblast activation on tissue culture polystyrene, hindering the investigation of early events that initiate fibrosis. Here, we present a statistically optimized culture protocol that suppresses fibroblast activation while preserving cell viability under standard culture conditions. Using a design of experiments (DOE) framework, we systematically evaluated combinations of extracellular matrix proteins and soluble factors to identify conditions that reduce myofibroblastic marker expression and extracellular matrix production. Fibroblasts cultured under optimized conditions remained spindle-shaped, exhibited low myofibroblastic marker expression, and showed reduced collagen and fibronectin secretion without evidence of cytotoxicity. The protocol was further validated in primary human colonic fibroblasts from both male and female donors, yielding consistent suppression of activation markers with high viability. This accessible and scalable approach provides a reproducible baseline for studying fibroblast activation and supports the use of DOE as a powerful strategy for defining microenvironmental conditions that regulate fibroblast behavior in vitro.
Clinical treatment of inflammatory bowel diseases (IBD) remains challenging due to the complex interplay between the epithelial barrier, immune system, and gut microbiota. While in vitro models are pivotal for studying barrier dysfunction, developing a standardized and functionally relevant system for IBD remains challenging. To overcome this, we established an immunocompetent murine colon epithelium monolayer to model IBD-like conditions. Colons from wild-type mice were digested into single cells and seeded onto Matrigel-coated transwells. Within seven days, monolayers showed strong barrier properties and displayed epithelial cell lineage, including goblet, stem, and enteroendocrine cells. However, exposure to pro-inflammatory cytokines as well as infection with pathogenic bacteria, including Clostridium rodentium and Salmonella Typhimurium, disrupted epithelial integrity. To better reflect the in vivo state, polarized T cells and macrophages were co-cultured with the epithelium. Pro-inflammatory Th1 and Th17 cells impaired barrier function, while M0 and M2 macrophages maintained it, representing both homeostatic and disrupted conditions of the gut. Upon Salmonella Typhimurium infection, M1 macrophages produced IFN-γ, and M2 macrophages secreted IL-10 and enhanced ZO-1 expression. Overall, our model presents a promising platform to study epithelial barrier dysfunction, immune-epithelial cross-talk, and host-pathogen interactions, offering valuable insights into IBD mechanisms and potential therapeutic approaches.
Kidney organoids derived from human pluripotent stem cells have emerged as promising models for studying kidney disease and therapeutic development. However, the lack of a scalable production system has limited their industrial applications in regenerative medicine. Here, we have developed a cost-effective mass-production method for manufacturing vascularized kidney organoids, which has improved production efficiency by more than 50 times compared to conventional culture systems. The incorporation of a dynamic culture environment in delta-wing stirred bioreactors has significantly enhanced the glomerular vascularization of kidney organoids via mechanosensory integrin α2β1. Single-cell RNA sequencing and functional analyses demonstrated the enhanced maturation in STR nephron epithelia. The large quantities of vascularized kidney organoids enabled the fabrication of a nephron sheet with nephron numbers equivalent to those found in two rat kidneys. Intravital imaging of a nephron sheet implanted in a dorsal skinfold chamber of mice revealed filtration function with size selectability in the organoid glomeruli vascularized with human endothelia. This work may represent a significant step towards bridging the gap between basic research and commercial products, paving the way towards developing bioengineered kidneys for kidney replacement therapy.
Engineered bacterial therapeutics are emerging as promising candidates for the treatment of solid cancers. However, despite effectively reducing tumour burden in different pre-clinical models, these "smart bugs" are yet to convincingly translate this efficacy to cancer treatment in humans. Here, we highlight key features of in vitro co-culture methods and how they can be used to investigate bacterial behaviour and guide the rational engineering of more effective anti-cancer bacterial therapeutics.