华西医院

Antinuclear antibody (ANA) testing is a critical method for diagnosing autoimmune disorders such as Lupus, Sjögren’s syndrome, and scleroderma. Despite its importance, manual ANA detection is slow, labor-intensive, and demands years of training. ANA detection is complicated by over 100 coexisting antibody types, resulting in vast fluorescent pattern combinations. Although machine learning and deep learning have enabled automation, ANA detection in real-world clinical settings presents unique challenges as it involves multi-instance, multi-label (MIML) learning. In this paper, a novel framework for ANA detection is proposed that handles the complexities of MIML tasks using unaltered microscope images without manual preprocessing. Inspired by human labeling logic, it identifies consistent ANA sub-regions and assigns aggregated labels accordingly. These steps are implemented using three task-specific components: an instance sampler, a probabilistic pseudo-label dispatcher, and self-paced weight learning rate coefficients. The instance sampler suppresses low-confidence instances by modeling pattern confidence, while the dispatcher adaptively assigns labels based on instance distinguishability. Self-paced learning adjusts training according to empirical label observations. Our framework overcomes limitations of traditional MIML methods and supports end-to-end optimization. Extensive experiments on one ANA dataset and three public medical MIML benchmarks demonstrate the superiority of our framework. On the ANA dataset, our model achieves up to +7.0% F1-Macro and +12.6% mAP gains over the best prior method, setting new state-of-the-art results. It also ranks top-2 across all key metrics on public datasets, reducing Hamming loss and one-error by up to 18.2% and 26.9%, respectively. The source code can be accessed at https://github.com/fletcherjiang/ANA-SelfPacedLearning.

Physics-Informed Neural Networks (PINNs) offer a powerful framework for solving PDEs by embedding physical laws into the learning process. However, when applied to domains with irregular boundaries, PINNs often suffer from instability and slow convergence, which stems from (1) inconsistent normalization due to geometric anisotropy, (2) inaccurate boundary enforcement, and (3) imbalanced loss term competition. A common workaround is to map the domain to a regular space. Yet, conventional mapping methods rely on case-specific meshes, define Jacobians at pre-specified fixed nodes, reformulate PDEs via the chain rule-making them incompatible with modern automatic differentiation, tensor-based frameworks. To bridge this gap, we propose JacobiNet, a learning-based coordinate-transformed PINN framework that unifies domain mapping and PDE solving within an end-to-end differentiable architecture. JacobiNet enables direct Jacobian computation via autograd, shares computation graph with downstream PINNs, thereby avoiding case-specific meshing, explicit Jacobian computation/storage, and manual PDE reformulation while unlocking geometric-editing operations. Separating physical modeling from geometric complexity, JacobiNet (1) addresses normalization challenges in the original anisotropic coordinates, (2) facilitates the hard enforcement of boundary conditions, and (3) mitigates the long-standing imbalance among loss terms. Evaluated on various PDEs, JacobiNet reduces the L2,rel error from 0.11 to 0.73 to 0.01-0.09, achieving an average 15.6 & times; improvement in accuracy. In vessel-like domains with varying shapes, JacobiNet enables millisecond-level mapping inference for unseen geometries, improves prediction accuracy by an average of 3.65 & times;, while delivering over 10 & times; speedup-demonstrating strong generalization, accuracy, and efficiency.

Vaccination is an important approach for cancer immunotherapy, but is limited by the inefficient delivery of immunogenic antigens to dendritic cells (DCs) to stimulate tumor-specific immune responses. Herein, we engineered glucosylated lipid nanoparticles (GluLNPs) to deliver tumor antigens and form personalized vaccines in situ, efficiently targeting DCs to stimulate robust antitumor immunity for the immunotherapy of low-immunogenic breast cancer. Through intratumoral injection, GluLNPs can efficiently capture and deliver tumor antigens, including neoantigens released from eradicated primary triple-negative breast tumors after mild phototherapy with photosensitizer-loaded nanoparticles (i.e., ICGNPs), to DCs, thereby triggering DC maturation and robust antitumor immune responses. Moreover, GluLNPs can effectively elicit strong systemic antitumor immunity against distant and metastatic triple-negative breast tumors when combined with immune checkpoint blockade. This study presents a practical strategy for developing effective DC-targeting personalized nanovaccines and in situ vaccination for precise and effective tumor immunotherapy.

Adults with coexistent diabetes mellitus and hypertension (DM-HTN) have a substantially elevated risk of all-cause mortality. Physical activity (PA) is a key modifiable factor, yet conventional self-reported or categorical measures may fail to capture clinically relevant variation. We investigated associations between sensor–derived continuous PA dimensions and all-cause mortality in DM-HTN. We analysed 4,294 DM-HTN and 1,482 diabetes-alone adults from the UK Biobank (median follow-up 85.2 months; 731 deaths) and 530 DM-HTN and 235 diabetes-alone participants from NHANES 2011–2014 (mean follow-up 63.5 months; 123 deaths), all with device-measured wrist accelerometry. Continuous sensor-derived metrics included Daily MET/MIMS (volume) and Peak-30 MET/MIMS (peak intensity). Associations with all-cause mortality were examined using multivariable Cox regression, restricted cubic spline modelling, and GBDT models with SHAP attribution. Both continuous PA dimensions were independently associated with all-cause mortality in the DM-HTN group across cohorts. In the UK Biobank, fully adjusted hazard ratios were 0.66 for Daily MET and 0.43 for Peak-30 MET. In NHANES, Peak-30 MIMS showed robust associations (HR = 0.57, p < 0.001), while Daily MIMS attenuated to non-significance (HR = 0.95, p = 0.116) after full adjustment. Non-linear dose–response modelling identified lower inflection points for peak intensity in DM-HTN compared with diabetes-alone individuals. SHAP analyses confirmed that continuous peak intensity maintained stable protective contributions across comorbidity strata. Wearable sensor-derived continuous PA dimensions capture mortality risk gradients in DM-HTN that are inaccessible to conventional approaches. Population-specific, clinically attainable inflection thresholds identified through continuous monitoring support incorporating sensor-derived PA metrics as digital biomarkers in comorbidity-specific risk stratification frameworks.