TF-MambaNet: A Temporal and Frequency Fused Bidirectional Mamba Architecture for PPG Foundation Model

Zhanpeng Bao (Institute of Acoustics CAS; UCAS; ImViA Université Bourgogne Europe), Yannick Benezeth (ImViA), Fan Yang (Université Bourgogne Europe), Yonglin Zhang, Haibin Wang, Chao Li (Institute of Acoustics, CAS)

IEEE ICASSP 2026· 2026

PPGFoundation ModelDOI

Paper

Summary

Averages: regression MAE 9.29 vs 10.54 / 11.86 / 10.17. Binary classification accuracy 0.74 vs 0.72 / 0.70 / 0.73.

Reported benchmarks

AFib / arrhythmia

MIMIC PERform · Accuracy ↑

#1 of 1

0.83

MIMIC 8 s AF, frozen features; accuracy only (no AUROC) — not comparable to AUROC rows; Chronos 0.81 / NormWear 0.78 / PaPaGei 0.66

Cuffless blood pressure

PPG-BP Database (Guilin / Liang 2018) · MAE (mmHg) ↓

#1 of 1

12.89 / 8.71

PPG-BP (Liang 2018) 2 s segments, frozen features (paper Table 1); finger clinical PPG; DBP Chronos 8.37 better

Heart rate estimation

PPG-DaLiA · MAE (bpm) ↓

#2 of 6

7.27

8 s windows, frozen features (paper Table 1); split unstated; same-paper re-runs NormWear 10.16 / PaPaGei 12.25 / Chronos 12.34

Mental workload / stress

WESAD · Accuracy ↑

#2 of 2

0.81

Binary stress, 8 s, frozen features (paper Table 2); accuracy — not comparable to AUROC rows; PaPaGei 0.80 / Chronos 0.77 / NormWear 0.77

Updated: Oct 8, 2026