Table 1 Characteristics of the single-group studies
Study, year | n | Age (years)a | Time since strokea | Brain signal for BCI | BCI intervention | Dosage of BCI | Outcome measures | Main results |
|---|---|---|---|---|---|---|---|---|
Buch et al. (2008) [56] | 8 | 58.2 ± 7.0 | 25.2 ± 11.6 (mo) | MEG, mu | BCI-orthosis | 1–2 h/d, 3–5 d/wk., 3–8 wk | MRC | Increased mu rhythm modulation, but no clinical change in MRC. |
Prasad et al. (2010) [26] | 5 | 58.6 ± 8.98 | 28 ± 15.4 (mo) | EEG, mu, beta | BCI-visual feedback | 30 min/d, 2 d/wk., 6 wk | ARAT, MRC, 9-HPT | Positive improvement in at least one outcome in all subjects. |
Tung et al. (2013) [57] | 6 | Unknown | Unknown | EEG | BCI-robot | 1 h/d, 5 d/wk., 2 wk | FMA-UE | Significant improvement in FMA-UE after BCI. |
Ono et al. (2014) [37] | 12 | 57.6 ± 11.8 | 30.8 ± 41.3 (mo) | EEG, mu, beta | BCI-visual feedback/somatosensory feedback | 1 h/d, 12–20 d | SIAS, EMG | BCI training with somatosensory feedback was more effective than that with visual feedback. |
Morone et al. (2015) [58] | 8 | 60 ± 10.9 | 24.4 ± 21.2 (wk) | EEG, beta | Con-rehab + BCI-visual feedback | 30 min/d, 3 d/wk., 4 wk | FMA-UE, NIHSS, BI | Positive improvement in all subjects and half of them had improvements higher than the MCID. |
Kawakami et al. (2016) [59] | 29 | 50.6 ± 10.9 | 48 ± 41.4 (mo) | EEG, mu | 40 min standard training + BCI-orthosis | 45 min/d, 5 d/wk., 2 wk | FMA-UE, MAL, MAS | Significant improvement in FMA-UE and MAL scores after BCI training. |
Kotov et al. (2016) [60] | 5 | 47.0 ± 7.7 | 2 mo-4 yr | EEG | BCI-exoskeleton | 8–10 d | NIHSS, MAS, BI, mRS | All patients showed decreases in neurological deficit after BCI training. |
Bundy et al. (2017) [61] | 10 | 58.6 ± 10.3 | 73.6 ± 104.2 (mo) | EEG, mu, beta | BCI-exoskeleton | 10–120 min/d, 5 d/wk., 12 wk | ARAT, MAS, | Significant improvement in ARAT after BCI training. |
Ibáñez et al. (2017) [62] | 4 | 54.3 ± 11.8 | 4 ± 0.8 (yr) | EEG, 7–30 Hz, Bereitschafts potential | BCI-FES | 10 days in one month | FMA-UE | Improved scores in FMA-UE after BCI training. |
Sullivan et al. (2017) [63] | 6 | 57.5 ± 7.9 | 51.5 ± 41.9 (mo) | EEG, MRCP | BCI-exoskeleton | 12 d in 5 wk | FMA-UE | Significant improvement in FMA-UE after BCI training. |
Nishimoto et al. (2018) [64] | 26 | 50.2 ± 11.1 | 47.4 ± 43.9 (mo) | EEG, mu | BCI-exoskeleton + FES | 40 min/d, 10 d | FMA-UE, MAL | Significant improvement in FMA-UE and MAL after BCI training. |
Chowdhury et al. (2018) [65] | 4 | 44.75 ± 15.69 | 7 ± 1.15 (mo) | EEG, mu, low beta | BCI-exoskeleton | 2–3 d/wk., 6 wk | ARAT, GS | The group mean changes from baseline in GS and ARAT were + 6.38 kg and + 5.66, respectively. |
Norman et al. (2018) [66] | 8 | 59.5 ± 11.8 | At least 6 (mo) | EEG, mu, beta | BCI- visual feedback | 3 d/wk., 4 wk | BBT | Hand function, measured by BBT improved by 7.3 ± 7.5 versus 3.5 ± 3.1 in those with and without SMR control. |
Remsik et al. (2018) [67] | 21 | 61.6 ± 15 | 1127 ± 1327 (d) | EEG | BCI- visual feedback, FES | 2 h/d, 15 d | ARAT, 9-HPT, SIS | Significant improvement in ARAT after BCI training. |
Tabernig et al. (2018) [68] | 8 | 61.2 ± 19.0 | 36.8 ± 24.2 (mo) | EEG, beta | BCI-FES | 1 h/d, 4 d/wk., 5 wk | Modified FMA-UE | Significant improvement in modified FMA-UE after BCI training. |
Carino-Escobar et al. (2019) [69] | 9 | 58.1 ± 12.1 | 158 ± 74 (d) | EEG, mu, beta | BCI-orthosis | 3 d/wk., 4 wk | FMA-UE | Six out of nine subjects had higher scores in FMA-UE after BCI training. |
Foong et al. (2019) [70] | 11 | 55.2 ± 11.0 | 333.7 ± 179.6 (d) | EEG | Standard arm therapy + BCI-visual feedback | 1 h/d, 2 d/wk., 6 wk | FMA-UE, ARAT | Significant improvement in FMA-UE after BCI training. |
Rathee et al. (2019) [71] | 4 | 62.5 ± 5.7 | 23 ± 4.2 (mo) | EEG, EMG | BCI-exoskeleton | 6 wk | ARAT, GS | Significant improvement in ARAT and GS after BCI training. |