ECG-RSI: Recursive Self-Improvement through Targeted Electrocardiogram Waveform Evolution
Abstract
Electrocardiogram (ECG) models need supervision that targets evolving weaknesses and remains faithful to waveform evidence. We introduce ECG-RSI, a recursive self-improvement framework combining verified waveform synthesis with feedback-guided intervention search. Signal remeasurement verifies waveform edits; construction errors guide generator repair, while learner errors motivate interventions compared on a separate selection set. Across seven benchmarks, three rounds raise Qwen3.5-35B-A3B's mean question-answering accuracy from 34.57% to 56.04% and mean classification F1 from 24.27 to 34.03. Across the retained trajectory, the preferred supervision shifts from direct reinforcement to confusion-focused contrasts. Continued feedback correction yields diminishing external gains. These results motivate separating error diagnosis from intervention selection and checking whether gains on the feedback set transfer to independent benchmarks.
Related
- ScienceRSI: Recursive Scientific Self-Improvement
- Machine learning-assisted directed evolution yields dramatic improvement on novel AAV engineering task
- NeuroGate: waveform translation between brain
- UniWave-2: A Hybrid Model for Nucleic Acid Waveform Feature Extraction Enhanced by Fourier and Wavelet Transforms
- biHash: bisulfite-aware recursive nucleotide hashing
BibTeX
@misc{wang2026ecgrsi,
title = {{ECG-RSI: Recursive Self-Improvement through Targeted Electrocardiogram Waveform Evolution}},
author = {Xiaoliang Wang and Zhuo Chen and Chuanyang Zheng and Jiangwei Lao and Quanbin Wang and Honglin Lin and Jiajia Liu and Zhihong Lu and Hao Jin and Jian Wang and Jinjie Gu and Lijun Wu},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.17718},
eprint = {2610.17718},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.17718},
}
RIS
TY - UNPB
TI - ECG-RSI: Recursive Self-Improvement through Targeted Electrocardiogram Waveform Evolution
AU - Xiaoliang Wang
AU - Zhuo Chen
AU - Chuanyang Zheng
AU - Jiangwei Lao
AU - Quanbin Wang
AU - Honglin Lin
AU - Jiajia Liu
AU - Zhihong Lu
AU - Hao Jin
AU - Jian Wang
AU - Jinjie Gu
AU - Lijun Wu
PY - 2026
DA - 2026/10/26
PB - arXiv
DO - 10.48550/arxiv.2610.17718
UR - https://arxiv.org/abs/2610.17718
AB - Electrocardiogram (ECG) models need supervision that targets evolving weaknesses and remains faithful to waveform evidence. We introduce ECG-RSI, a recursive self-improvement framework combining verified waveform synthesis with feedback-guided intervention search. Signal remeasurement verifies waveform edits; construction errors guide generator repair, while learner errors motivate interventions compared on a separate selection set. Across seven benchmarks, three rounds raise Qwen3.5-35B-A3B's mean question-answering accuracy from 34.57% to 56.04% and mean classification F1 from 24.27 to 34.03. Across the retained trajectory, the preferred supervision shifts from direct reinforcement to confusion-focused contrasts. Continued feedback correction yields diminishing external gains. These results motivate separating error diagnosis from intervention selection and checking whether gains on the feedback set transfer to independent benchmarks.
ER -