Finetuning and Quantization of EEG-Based Foundational BioSignal Models on ECG and PPG Data for Blood Pressure Estimation

CEReBrO is a compact EEG foundation model (3.58M, 39.95M, and 85.15M parameters) pretrained on the public TUEG scalp-EEG corpus, more than 20,000 hours from more than 10,000 subjects. It is not a PPG-native foundation model. Tóth et al. fine-tune it on 10-second ECG+PPG windows from PulseDB, which is built from MIMIC-III and VitalDB, for cuffless blood pressure. The large unfrozen model at 200 epochs reports DBP/SBP MAE of 1.57/2.72 mmHg on the PulseDB MIMIC-III split and 1.92/3.14 mmHg on VitalDB (Table II).

• EEG foundation model pretrained on TUEG; not a PPG-native FM
• Sizes in the CEReBrO tables: 3.58M, 39.95M, 85.15M parameters
• Transfer paper: ECG+PPG, 10 s at 125 Hz, two channels, head predicts SBP and DBP
• PulseDB MIMIC-III, large, unfrozen, 200 epochs: DBP MAE 1.57 mmHg, SBP MAE 2.72 mmHg, BHS A, AAMI pass (Table II)
• VitalDB, same setting: DBP MAE 1.92 mmHg, SBP MAE 3.14 mmHg (Table II)
• Small model 13.73 MB; Table IV dynamic INT8 3.85 MB, static INT8 3.83 MB (abstract says 13.73 to 3.83 MB)

Blood pressure (BP) is a key indicator of cardiovascular health. As hypertension remains a global cause of morbidity and mortality, accurate, continuous, and non-invasive BP monitoring is therefore of paramount importance. Photoplethysmography (PPG) and electrocardiography (ECG) can potentially enable continuous BP monitoring, yet training accurate and robust machine learning (ML) models remains challenging due to variability in data quality and patient-specific factors. Recently, multiple research groups explored Electroencephalographic (EEG)--based foundation models and demonstrated their exceptional ability to learn rich temporal resolution. Considering the morphological similarities between different biosignals, the question arises of whether a model pre-trained on one modality can effectively be exploited to improve the accuracy of a different signal type. In this work, we take an initial step towards generalized biosignal foundation models by investigating whether model representations learned from abundant EEG data can effectively be transferred to ECG/PPG data solely with fine-tuning, without the need for large-scale additional pre-training, for the BP estimation task. Evaluations on the MIMIC-III and VitalDB datasets demonstrate that our approach achieves near state-of-the-art accuracy for diastolic BP (mean absolute error of 1.57 mmHg) and surpasses by 1.5x the accuracy of prior works for systolic BP (mean absolute error 2.72 mmHg). Additionally, we perform dynamic INT8 quantization, reducing the smallest model size by over 3.5x (from 13.73 MB down to 3.83 MB) while preserving performance, thereby enabling unobtrusive, real-time BP monitoring on resource-constrained wearable devices.

Verbatim from arXiv (opens in a new tab) · arXiv metadata (abstracts are released under CC0)

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

• PulseDB v2.0cleaned MIMIC-III + VitalDB cuffless-BP benchmark (MIMIC part ODbL; VitalDB part CC BY-NC-SA)5,361 subjects · PPG · ECG · BP
• VitalDBintraoperative multi-parameter vitals (PhysioNet mirror)6,388 subjects · PPG · ECG · BP

• CEReBrOtransformerFoundation modelPPGECGEEG2025