Paperdoi:10.1109/JIOT.2026.3717960

KANFlow: Efficient Shallow-KAN Flow Matching for Wearable PPG-to-ECG Reconstruction

Venue listed2026Added Oct 11, 2026

Authors: Yiming Chen, Qiyu Wang, Yongquan Dong, Zekuan Yu (Fudan, UCL, Jiangsu Normal)

IEEE Internet of Things Journal

TL;DR

Subject-independent conditional flow matching; RBF-KAN refinements only at high-resolution stages, MLP blocks deeper; cross-attentive dual-stream architecture and contrastive flow regularization for unseen subjects. Four public clinical datasets; best overall balance of waveform fidelity and ECG-analytics utility (no numbers in abstract). Deployment analysis favors cloud-assisted inference.

Form factor: any

Results on ChatPPG leaderboards

All leaderboards

No results from this paper on ChatPPG leaderboards yet. Browse the leaderboards.

Tasks & signals

Signals

PPG ECG

Keywords

PPG-to-ECGflow-matchingKANsubject-independentgenerative

How it works

Cite this paper

APA

Chen, Y., Wang, Q., Dong, Y., & Yu, Z. (2026). KANFlow: Efficient Shallow-KAN Flow Matching for Wearable PPG-to-ECG Reconstruction. IEEE Internet of Things Journal. https://doi.org/10.1109/JIOT.2026.3717960

BibTeX

@article{chen2026kanflow,
  title         = {{KANFlow: Efficient Shallow-KAN Flow Matching for Wearable PPG-to-ECG Reconstruction}},
  author        = {Chen, Yiming and Wang, Qiyu and Dong, Yongquan and Yu, Zekuan},
  year          = {2026},
  journal       = {IEEE Internet of Things Journal},
  doi           = {10.1109/JIOT.2026.3717960}
}

Built from the hub record. Check it against the publisher before submitting.

Oct 11, 2026Added to the ChatPPG research hub