Contrastive Learning for Multi-Label ECG Classification with Jaccard Score–Based Sigmoid Loss

Junichiro Takahashi, Masataka Sato, Satoshi Kodeta, Norihiko Takeda · 2026

Read Contrastive Learning for Multi-Label ECG Classification with Jaccard Score–Based Sigmoid Loss by Junichiro Takahashi, Masataka Sato, Satoshi Kodeta, Norihiko Takeda (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-10553.txt.

Description

Junichiro Takahashi and three co-authors build a stronger ECG encoder for mixed medical AI in a 2026 paper. Large medical models handle text-based exam questions well, but their ECG results lags, and some leading models do not support ECG data at all. Interpreting heart traces is hard, and scores vary with the reader's experience.

The team trains an encoder on real hospital data using a contrastive model with a sigmoid loss suited to multi-label prediction. They then present a modified loss tailored to the multi-label nature of ECG diagnosis, where one trace can carry several settings at once. Tests reported in the paper show the approach improves ECG grasp, giving mixed medical models a firmer footing in cardiology.

Catalog record

CREATORJunichiro Takahashi, Masataka Sato, Satoshi Kodeta, Norihiko Takeda
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-10553

Browse the full library · About the Tower of Babel