Advanced Science, EarlyView.An eight‐stream 2.5D Transformer integrates complementary PET metabolic and CT morphologic information to estimate spread‐through‐air‐spaces risk before surgery in lung adenocarcinoma. The model generalizes across independent centers, while its continuous risk score independently stratifies progression‐free survival, supporting further prospective ...
Xin‐Yu Zhu +7 more
wiley +1 more source
An Integrated Artificial Intelligence Framework for Nanomedicine Engineering
Advanced Science, EarlyView.This review outlines a closed‐loop AI ecosystem for nanomedicine engineering. By synergizing multimodal data harmonization, objective‐driven AI architectures, and automated high‐throughput validation, this “Design‐Make‐Test‐Analyze” (DMTA) workflow transforms traditional heuristic formulation into a predictive, self‐optimizing paradigm, accelerating ...
Meichen Dong +10 more
wiley +1 more source
ReSNNance: an HDL generation framework for spiking neural networks. [PDF]
Front NeurosciLópez-Asunción S +3 more
europepmc +1 more source
Enhancing Specific Stiffness of Nano‐Architected Materials Through Generative AI and Active Learning
Advanced Science, EarlyView.A structure‐centric generative framework searches voxelated design space directly, coupling a variational autoencoder with multi‐objective optimization and active learning to progressively expand the validated design domain. The discovered nano‐architectures, realized as pyrolytic carbon by two‐photon lithography, deliver 18–23% higher stiffness and 26–
Jinwook Yeo +4 more
wiley +1 more source
Going deeper with morphologically detailed neural networks by simulation-based gradient propagation. [PDF]
Front Comput NeurosciHe G, Du K, Huang T.
europepmc +1 more source
Flexible Integration of Structurally Diverse Spatial Multi‐Omics With AgaeSMO
Advanced Science, EarlyView.AgaeSMO is a versatile and robust deep learning framework for flexible cross‐integration of spatial multi‐omics data. Leveraging a GATs network with a graph attention module, it learns joint embedding across diverse modalities, enabling integration of both structurally paired (e.g., spatial transcriptome‐epigenome, transcriptome‐proteome, transcriptome‐
Kunpeng Liu +9 more
wiley +1 more source
Spatiotemporal Computational Meta-Imager for Broadband Dual-Layer Optical Convolution. [PDF]
Adv Sci (Weinh)Wang M +6 more
europepmc +1 more source
A Biomimetic Solvatochromic Chip With Optimized Chemical Diversity for On‐Site Liquid Identification
Advanced Science, EarlyView.A compact biomimetic photonic‐gustatory chip uses Hansen‐solubility‐parameter‐guided solvatochromic disks to create distinct color patterns in organic liquids. An optimized four‐disk array distinguishes solvent types, ratios, homologues, isomers, and alcoholic beverages. Coupled with smartphone imaging and deep learning, it identifies 80 liquid classes
Jintao Zhang +9 more
wiley +1 more source
Rolling bearing fault diagnosis method based on dual attention multi-scale convolutional network. [PDF]
PLoS OneZhang Y, Liu X.
europepmc +1 more source
Massively Multiplexed Photonic Chip With Cross‐Channel Fusion For Harmonized Biosensing
Advanced Science, EarlyView.We introduce a transformative sensing platform integrating photonic multiplexing with over 10 000 sensors and a deep‐learning framework that combines domain adaptation and cross‐channel fusion for harmonized operation across heterogeneous chips. The platform achieves 99.3% identification accuracy and relative prediction errors of 10−4$^{-4}$ for ...
Ivan Saetchnikov +3 more
wiley +1 more source