analiza aiqbench
BEAT-Net advances medical AI by introducing a biomimetic architecture that explicitly encodes cardiac physiology, achieving high diagnostic accuracy with significantly reduced model parameters and improved data efficiency. The approach demonstrates that aligning deep learning representations with clinical heuristics yields more interpretable decision boundaries than implicit signal processing alone. This work highlights a promising direction for deploying efficient, trustworthy AI systems in resource-constrained healthcare environments.
BEAT-Net: Injecting Biomimetic Spatio-Temporal Priors for Interpretable ECG Diagnosis
arXiv:2601.07316v2 Announce Type: replace-cross Abstract: Automated electrocardiogram diagnosis using deep learning remains limited by signal-agnostic representations that treat multi-lead recordings as undifferentiated time-series or images, forcing models to rediscover physiological structure implicitly. This leads to data inefficiency, poor generalization, and opaque decision boundaries misaligned with clinical reasoning. We present BEAT-Net, a supervised biomimetic framework that integrates QRS-centered biological tokenization with a hierarchical architecture mirroring the cardiologist's workflow. A QRS tokenizer converts continuous signals into semantically complete heartbeat sequences, which are processed through four specialized stages: morphological feature extraction via a Word Encoder, lead-invariant normalization through a Spatial Operator, temporal context injection by a Temporal Operator, and global reasoning using a Transformer-based Sentence Encoder. Evaluated across three large-scale benchmarks including PTB-XL, CPSC2018, and CSN, BEAT-Net achieves diagnostic accuracy of 0.924 AUC, comparable to dominant CNN baselines at 0.925 AUC, while reducing parameters by 95 percent from 2.06 million to 0.7 million. Critically, BEAT-Net surpasses the 39.5-million-parameter foundation model HeartLang on morphological Form classification, reaching 0.901 AUC compared to HeartLang's 0.832 AUC, while attaining full CNN-level performance using only 35 percent of training data and exhibiting superior cross-dataset generalization. Learned attention patterns spontaneously align with established clinical heuristics, demonstrating that explicit physiological structure provides a more efficient and interpretable alternative to massive pre-training for clinical deployment.
Abstract:Automated electrocardiogram diagnosis using deep learning remains limited by signal-agnostic representations that treat multi-lead recordings as undifferentiated time-series or images, forcing models to rediscover physiological structure implicitly. This leads to data inefficiency, poor generalization, and opaque decision boundaries misaligned with clinical reasoning. We present BEAT-Net, a supervised biomimetic framework that integrates QRS-centered biological tokenization with a hierarchical architecture mirroring the cardiologist's workflow. A QRS tokenizer converts continuous signals into semantically complete heartbeat sequences, which are processed through four specialized stages: morphological feature extraction via a Word Encoder, lead-invariant normalization through a Spatial Operator, temporal context injection by a Temporal Operator, and global reasoning using a Transformer-based Sentence Encoder. Evaluated across three large-scale benchmarks including PTB-XL, CPSC2018, and CSN, BEAT-Net achieves diagnostic accuracy of 0.924 AUC, comparable to dominant CNN baselines at 0.925 AUC, while reducing parameters by 95 percent from 2.06 million to 0.7 million. Critically, BEAT-Net surpasses the 39.5-million-parameter foundation model HeartLang on morphological Form classification, reaching 0.901 AUC compared to HeartLang's 0.832 AUC, while attaining full CNN-level performance using only 35 percent of training data and exhibiting superior cross-dataset generalization. Learned attention patterns spontaneously align with established clinical heuristics, demonstrating that explicit physiological structure provides a more efficient and interpretable alternative to massive pre-training for clinical deployment.
10 pages, 6 figures and 2 tables. Revised version of the manuscript submitted to the IEEE Journal of Biomedical and Health Informatics. Title updated from "Interpretable ECG Classification" to "Interpretable ECG Diagnosis"; author list expanded to match the submitted version
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2601.07316 [cs.LG]
(or arXiv:2601.07316v2 [cs.LG] for this version)
Text integral sub licența deschisă a publicației · 263 cuvinte
Subiecte conexe
[AINews] Meta Muse Wonder & Wonder Full: MSL’s Claude 5.5-tier Frontier Model
Latent Space · 13 oct. 2026
NeuralDMD: Interpretable Neural Representation of Dynamics from Sparse and Noisy Measurements
arXiv · 19 oct. 2026
PhysMLLMs: Spatial Priors for Unified Referring Segmentation and Grounded Reasoning of Images and Videos
arXiv · 27 aug. 2026
Links #5: 2026/07
LessWrong · 31 iul. 2026