Swarm of digital particles trains medical AI without sharing patient data
A particle swarm optimization strategy that dynamically selects the most informative clients in federated learning matches state-of-the-art methods for medical ...
Tag: deep learning
AI Cuts the Cost of Labeling Scrap Steel Images with Pseudo-Labels
A semi-supervised YOLO11 pipeline using confidence filtering, DIoU-NMS deduplication, and light manual checks outperforms fully supervised scrap steel detection while ...
AI Meets the Blood Drop: Scientists Map the Rules for Merging Liquid Biopsy Data in Cancer Research
A new methods article outlines how artificial intelligence should be used to integrate liquid biopsy biomarkers in cancer research, finding ...
Detectors, not association tricks, drive multi-object tracking success
A systematic experimental study in Artificial Intelligence Review finds detector improvements deliver more than 10 percent tracking gains while refined ...
Two-Stage Feedback Network Aims to Sharpen Multimodal Medical Image Fusion
A newly published two-stage deep network called FADFNet uses parallel feature extraction, source-image compensation, and a decomposition feedback loop to ...
AI research on sports injury prediction has exploded, a new map shows
A new bibliometric analysis of 809 publications shows that research on artificial intelligence for predicting and preventing sports injuries grew ...
Hybrid AI Fuses Deep and Handcrafted Features for Near-Perfect Cervical Cancer Screening
A hybrid machine learning framework fusing autoencoder-learned representations with handcrafted texture features achieved approximately 99.85% test accuracy for cervical cancer ...
Ontology-aware model improves external brain-metastasis MRI segmentation
A three-dimensional framework using ontology-aware tissue representation and training-only counterfactual supervision improved brain-metastasis MRI segmentation in 38 cases from a ...
CauTrigger models biological state transitions through hierarchical causal decomposition
A hierarchical deep latent-variable model outperformed existing methods at identifying causal triggers and predicting perturbation responses in synthetic and real ...
Airway AI needs more than high segmentation scores
A narrative review finds that airway AI must preserve branching anatomy, generalize across institutions and demonstrate clinical utility rather than ...
Small-cohort nnU-Net study shows promise for pediatric Wilms tumor MRI segmentation
A retrospective Dutch proof-of-concept study found that nnU-Net trained from scratch on 10 patients reached a median Dice score of ...
African infectious-disease AI review finds 43 studies, uneven coverage and barriers
A systematic review found 43 studies applying artificial intelligence to infectious-disease prediction, management, surveillance or outbreak prediction in Africa between ...
Deep learning improves subject-adaptive P300 EEG decoding in computational study
A computational study found that confidence-based routing between subject-specific and pooled decoders improved P300 EEG decoding in two public datasets.
Neural-aided inertial tracking cuts dog trajectory error in small study
Neural-aided inertial tracking reduced mean position error to 2.91 metres for two walking dogs, while the best configuration differed between ...
MRI fusion model classifies endometrial cancer risk in 246-patient study
A retrospective study of 246 patients found that a model combining multiparametric MRI radiomics, deep learning and clinical data achieved ...
Dual-stage motor imagery boosts brain-computer interface decoding
An offline EEG study of 97 healthy adults linked motor imagery to brief limb execution and used a dual-branch neural ...
Bibliometric map shows AI research surge in bone metastasis
A bibliometric analysis of 540 English-language AI and bone-metastasis publications through June 2026 found rapid growth after 2020, persistent imaging ...
Plantar pressure patterns offer a partial window into skiing experience
A deep-learning classifier partially separated beginner and novice skiers from plantar-pressure images, but limited accuracy, small samples and self-reported experience ...
BiGRU-CRF system reaches 96.52% accuracy in power-grid text
A 2026 computational study found that a domain-specific BiGRU-CRF pipeline achieved 96.52% entity-level exact-match accuracy across 6,000 Chinese power-grid text ...
Research report: Deep learning based detection of active pulmonary tuberculosis from chest radiographs in a retrospective prospective multicenter study
A lightweight ShuffleNetV2 model distinguished active from non-active pulmonary tuberculosis on chest radiographs with a prospective AUC of 0.896 and ...
POPULAR NEWS
-
Case Report Explores Caplacizumab in Late-Diagnosed Congenital Thrombotic Thrombocytopenic Purpura
-
Multitask Cambrian Model Improves Protein Melting Temperature Prediction
-
Vasopressors Exert Divergent Immunomodulatory Effects Across a Spectrum in Septic Shock
-
Study Identifies Key Drivers Shaping Salmonid Gill Microbiomes in Aquaculture