Table 3. Studies of sustained attention with machine learning models.
| Author | Type of subjects | Number of subjects | Research method | Data sources | Average accuracy | Classification |
|---|---|---|---|---|---|---|
| Borhani et al. (2018) | Healthy, age 21–24 years | N = 38 (27 males) | CNN | EEG | 73% | 2 classes attentional state |
| Hosseini & Guo (2019) | Healthy, age 25,31 years | N = 2 (1 male) | CNN | EEG | 91.78% | Focusing state and mind wandering |
| Ho et al. (2019) | Healthy | N = 16 (8 males) | CNN | fNIRS | 65.43% | 3 classes mental workload |
| Khullar et al. (2021) | ADHD, non-ADHD, age 7–21 years | ADHD (N = 285), non-ADHD (N = 491) | 2D CNN–LSTM | fMRI | 98.12% | ADHD and non-ADHD |
| Zou et al. (2017) | ADHD, non-ADHD, age 7–21 years | ADHD (N = 286), non-ADHD (N = 340) | CNN | MRI | 69.15% | ADHD and non-ADHD |
| Mao et al (2019) | ADHD, non-ADHD, age 7–21 years | ADHD (N = 285), non-ADHD (N = 491) | 4-D CNN | fMRI | 71.3% | ADHD and non-ADHD |
| Chen et al. (2019) | ADHD age 10.44 ± 0.75 years, non-ADHD age 10.92 ± 0.69 years | ADHD (N = 50) (41 males). non-ADHD (N = 57) (53 males) | CNN | EEG | 90.29% | ADHD and non-ADHD |
| Fouladvand et al. (2018) | ADHD, age 13–20 years | N = 11,624 | LSTM | – | 84% | Temporal medication features |
| Phan et al. (2018) | Healthy | N = 20 | RNN | EEG | 82.50% | 5 sleep stages |
| Liu & Liu (2017) | Healthy | N = 27 | RNN | EEG | 74.50% | Arousal and valence |
| Huve, Takahashi & Hashimoto (2018) | Healthy | – | RNN | fNIRS | 63% | Driving under clear weather and driving under rainy weather |
| Rasyid & Djamal (2019) | Healthy | – | RNN | EEG | 89.73% | Emotion and attention |
| Moinnereau et al. (2018) | Healthy | – | RNN | EEG | 83.2% | 3 auditory stimuli |
| Ming et al. (2018) | Healthy | N = 16 | RRN | EEG | 89.3% | 2 classes attentional state |
| Kuang & He (2014) | ADHD | N = 222 | DBN | fMRI | 35.19% | 4 classes attentional state |
| Kuang et al. (2014) | ADHD | N = 94 | DBN | fMRI | 61.90% | 4 classes attentional state |
| Vahid et al. (2019) | ADHD, non-ADHD | N = 144 | EEGNet | EEG | 83% | ADHD and non-ADHD |
| Hao & Yin (2015) | ADHD, non-ADHD | N = 873 | – | fMRI | 64.70% | ADHD and non-ADHD |
| Dubreuil-Vall, Ruffini & Camprodon (2019) | ADHD age 43.85 ± 14.78 years. non-ADHD age 29.90 ± 10.77 years | ADHD (N = 20) (10 males). non-ADHD (N = 20) (10 males) | Deep CNN | EEG | 88% | ADHD and non-ADHD |
Notes.
- CNN
- convolutional neural networks
- EEG
- electroencephalogram
- fNIRS
- functional nearinfrared spectroscopy
- ADHD
- attention-deficit/hyperactivity disorder
- LSTM
- Long short-term memory
- fMRI
- functional magnetic resonance imaging
- RNN
- recurrent neural networks
- DBN
- deep belief network