Table 1.
Comparison of studies using machine learning and deep learning (DL) models for blood pressure estimation based solely on ECG data.
| Data | SBP (mmHg) | DBP (mmHg) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Study | # Records | # Subjects | ECG len. | # Features | ME | STD | MAE | STD | ρ | ME | STD | MAE | STD | ρ |
| (Simjanoska et al 2018) 2018 | 3129 | 51 | 30 s | 7 | — | — | 7.7 | 10.2 | — | — | — | 9.5 | 10.0 | — |
| (Mousavi et al 2018) 2018 | 660 | 220 | 15 s | — | — | — | 12.8 | 12.2 | — | — | — | 6.0 | 6.4 | — |
| (Simjanoska et al 2020) 2020 | 3129 | 51 | 30/20/10 s | 7 | — | — | 16.6 | 11.1 | — | — | — | 9.2 | 7.9 | — |
| (Banerjee et al 2022) 2022 | 2000 | — | 16 s | — | 0.8 | 9.1 | 6.0 | — | — | 0.1 | 6.2 | 3.5 | — | — |
| (Wuerich et al 2022) 2022 | — | 942 | 4 s | 34 | 5.9 | 7.2 | 0.9 | — | — | — | — | 3.7 | 5.2 | 0.92 |
| (Syah et al 2023) 2023 | 56 | — | 20 s | 3 | — | — | 2.8 | 1.5 | 0.95 | — | — | 2.9 | 2.1 | 0.93 |
| (Aldein et al 2023) 2023 | 4904 | — | 32 s | — | — | — | 6 | 4.5 | — | — | — | 2.5 | 3.7 | — |
| (Kuzmanov et al 2024) 2024 | 130 461 | — | 30 s | — | — | — | 10.9 | — | — | — | — | 6.6 | — | — |
| (Aldein et al 2025) 2025 | 12 000 | 942 | 32 s | — | — | — | 5.50 | 4.28 | — | — | — | 2.50 | 3.80 | — |
| (Miao et al 2020) 2020 | 897 743 | 1711 | 2.5 s | DL | –0.1 | 10.0 | 7.1 | — | 0.88 | 0.0 | 6.3 | 4.6 | — | 0.71 |
| (Fan et al 2021) 2021 | 21 422 | — | 10 s | DL | 0.1 | 10.8 | 7.7 | — | — | 0.1 | 5.9 | 4.4 | — | — |
| (Fan et al 2020) 2021 | 21 422 | — | 10 s | DL | 0.2 | 10.8 | 7.2 | — | — | 1.2 | 5.9 | 3.9 | — | — |