Table 3.
Description of the parameters related to the development of the model.
| Parameter | Description |
|---|---|
| Handling of missing values | 26/81 (32.1) |
| Compatible with model | 4/26 (15.4) |
| Imputation | 17/26 (65.4) |
| Not reported | 5/26 (19.2) |
| Preprocessing | 78/81 (96.3) |
| One-hot encoding | 45/78 (57.7) |
| Embedding | 42/78 (53.8) |
| Time window aggregation | 26/78 (33.3) |
| Simultaneous preprocessing techniques (n) | [0, 5], 1.9 (1), 2 [1-3], 100 |
| Feature selection | 6 (7.4) |
| Varying-length sequence handling | |
| Preprocessing | 31/81 (38.3) |
| Zero-padding | 22/81 (27.2) |
| Not reported | 19/81 (23.5) |
| Not needed | 5/81 (6.2) |
| Number of layers | [0,10], 2.9 (2), 2 [2-4], 60.5 |
| Use of attention mechanism | 45/81 (55.6) |
| Use of static variables | 27/81 (33.3) |
| Hyperparameter tuning | |
| Performed | 36/81 (44.4) |
| Not performed | 34/81 (42) |
| Not reported | 11/81 (13.6) |
| Hyperparameter tuning method | |
| Fine-tuning | 13/36 (36.1) |
| Grid search | 7/36 (19.4) |
| Others | 10/36 (27.8) |
| Not reported | 6/36 (16.7) |
| Hyperparameters tuned | |
| Number of neurons per layer | 16/36 (44.4) |
| Learning rate | 10/36 (27.8) |
| Dropout rate | 8/36 (22.2) |
| Simultaneous hyperparameters being tuned (n) | [1, 8], 2.6 (1.9), 2 [1-3.3] (80.6) |
Categorical parameters are described as N (%), while quantitative parameters as [min, max], mean (SD), median [Q1-Q3], (% studies reported).