Table 2.
Description of the registers collection, the prediction task, and the architecture by data context.
| Hospital care (N = 52) |
|||||||
|---|---|---|---|---|---|---|---|
| Parameter | Primary care (N = 7) | Consultation (N = 4) | Hospitalization (N = 20) | ICU (N = 28) | Multiple healthcare settings (N = 11) | Not reported (N = 11) | Total (N = 81) |
| Mean number registers per person | |||||||
| 0-10 | 0 (0.0) | 0 (0.0) | 5 (25.0) | 8 (28.6) | 2 (18.2) | 2 (18.2) | 17 (21.0) |
| 11-50 | 0 (0.0) | 2 (50.0) | 2 (10.0) | 2 (7.1) | 1 (9.1) | 4 (36.4) | 11 (13.6) |
| 51-100 | 3 (42.9) | 0 (0.0) | 3 (15.0) | 2 (7.14) | 0 (0.0) | 0 (0.0) | 8 (9.9) |
| Not reported | 4 (57.1) | 2 (50.0) | 10 (50.0) | 16 (57.1) | 8 (72.7) | 5 (45.5) | 45 (55.6) |
| Frequency of registers (N = 84) | (N = 9) | (N = 4) | (N = 20) | (N = 29) | (N = 11) | (N = 11) | (N = 84) |
| Hourly (every 0.5, 1, 2, … h) | 0 (0.0) | 0 (0.0) | 2 (10.0) | 8 (27.6) | 2 (18.2) | 0 (0.0) | 12 (14.3) |
| Daily | 1 (11.1) | 0 (0.0) | 1 (5.0) | 4 (13.8) | 1 (9.1) | 1 (9.1) | 8 (9.5) |
| Weekly | 1 (11.1) | 1 (25.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (9.1) | 3 (3.6) |
| Monthly (every 1, 6, … months) | 2 (22.2) | 1 (25.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 2 (18.2) | 5 (6.0) |
| Yearly (every 1, 2, … years) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (9.1) | 1 (1.2) |
| At routine follow-up | 5 (55.6) | 2 (50.0) | 15 (75.0) | 16 (55.2) | 8 (72.7) | 6 (54.5) | 52 (61.9) |
| Not fixed | 0 (0.0) | 0 (0.0) | 1 (5.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (1.2) |
| Not reported | 0 (0.0) | 0 (0.0) | 1 (5.0) | 1 (3.45) | 0 (0.0) | 0 (0.0) | 2 (2.4) |
| Prediction task (N = 124) | (N = 9) | (N = 4) | (N = 32) | (N = 48) | (N = 13) | (N = 18) | (N = 124) |
| Clinical predictions | 5 (55.5) | 4 (100) | 13 (40.6) | 9 (18.8) | 8 (61.5) | 15 (83.3) | 54 (43.5) |
| Cancer (eg, colorectal, pancreatic) | 1 (11.1) | 0 (0.0) | 1 (3.12) | 0 (0.0) | 2 (15.4) | 1 (5.6) | 5 (4.0) |
| Cardiovascular system (eg, heart failure) | 3 (33.3) | 2 (50.0) | 4 (12.5) | 2 (4.17) | 2 (15.4) | 3 (16.7) | 16 (12.9) |
| Infections (eg, sheptic shock) | 0 (0.0) | 0 (0.0) | 2 (6.3) | 3 (6.3) | 2 (15.4) | 0 (0.0) | 7 (5.7) |
| Mental health (eg, depression, suicidal ideation) | 0 (0.0) | 1 (25.0) | 2 (6.3) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 3 (2.4) |
| Metabolic (eg, diabetes, obesity) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 6 (33.3) | 6 (4.8) |
| Neurorological system (eg, Alzheimer’s) | 0 (0.0) | 1 (25.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (5.6) | 2 (1.6) |
| Respiratory system (eg, COPD) | 0 (0.0) | 0 (0.0) | 2 (6.3) | 4 (8.3) | 1 (7.7) | 3 (16.7) | 10 (8.1) |
| Urinary system (eg, kidney disease) | 1 (11.1) | 0 (0.0) | 2 (6.3) | 0 (0.0) | 1 (7.7) | 1 (5.6) | 5 (4.0) |
| Disease progression and health status | 1 (11.1) | 0 (0.0) | 4 (12.5) | 12 (25.0) | 2 (15.4) | 0 (0.0) | 19 (15.3) |
| Decompensation | 0 (0.0) | 0 (0.0) | 0 (0.0) | 3 (6.3) | 0 (0.0) | 0 (0.0) | 3 (2.4) |
| Mortality | 1 (11.1) | 0 (0.0) | 4 (12.5) | 9 (18.8) | 2 (15.4) | 0 (0.0) | 16 (12.9) |
| Outcome measures for quality care | 0 (0.0) | 0 (0.0) | 4 (12.5) | 8 (16.7) | 0 (0.0) | 0 (0.0) | 12 (9.7) |
| Hospital (re)admission | 0 (0.0) | 0 (0.0) | 3 (9.4) | 1 (2.1) | 0 (0.0) | 0 (0.0) | 4 (3.2) |
| In-hospital mortality | 0 (0.0) | 0 (0.0) | 0 (0.0) | 4 (8.3) | 0 (0.0) | 0 (0.0) | 4 (3.2) |
| Length of stay | 0 (0.0) | 0 (0.0) | 1 (3.1) | 3 (6.3) | 0 (0.0) | 0 (0.0) | 4 (3.2) |
| Other predictions | 3 (33.3) | 0 (0.0) | 11 (34.3) | 19 (39.5) | 3 (23.1) | 3 (16.7) | 39 (31.5) |
| Next event (eg, diagnose, drug) | 3 (33.3) | 0 (0.0) | 10 (31.2) | 15 (31.2) | 2 (15.4) | 1 (5.6) | 31 (25.0) |
| Others (Freq. <2) | 0 (0.0) | 0 (0.0) | 1 (3.1) | 4 (8.3) | 1 (7.7) | 2 (11.1) | 8 (6.5) |
| Prediction window (N = 117) | |||||||
| Hours (1, 3, 6, 8 h) | 0 (0.0) | 0 (0.0) | 3 (8.3) | 4 (12.1) | 3 (18.8) | 0 (0.0) | 10 (8.6) |
| Days (1, 2, 7, 15 days) | 0 (0.0) | 0 (0.0) | 8 (22.2) | 7 (21.2) | 0 (0.0) | 1 (7.1) | 16 (13.7) |
| Months (1, 2, 3, 6, 9 months) | 2 (14.3) | 0 (0.0) | 12 (33.3) | 4 (12.1) | 3 (18.8) | 4 (28.6) | 25 (21.4) |
| Years (1, 2, 3, 4, 5, 10 years) | 8 (57.1) | 2 (50.0) | 7 (19.4) | 0 (0.0) | 4 (25.0) | 4 (28.6) | 25 (21.4) |
| Any | 4 (28.6) | 2 (50.0) | 6 (16.7) | 18 (54.5) | 6 (37.5) | 5 (35.7) | 41 (35.0) |
| Architecture | |||||||
| RNN-based only | 5 (71.4) | 3 (75.0) | 14 (70.0) | 14 (50.0) | 8 (72.7) | 6 (54.5) | 50 (61.7) |
| RNN/BiRNN | 0 (0.0) | 1 (25.0) | 0 (0.0) | 0 (0.0) | 1 (9.1) | 0 (0.0) | 2 (2.5) |
| GRU/BiGRU | 4 (57.1) | 0 (0.0) | 8 (40.0) | 4 (14.3) | 3 (27.3) | 3 (27.3) | 22 (27.2) |
| LSTM/BiLSTM | 1 (14.3) | 2 (50.0) | 6 (30.0) | 10 (35.7) | 4 (36.4) | 3 (27.3) | 26 (32.1) |
| Transformer-based | |||||||
| BERT-based architectures | 0 (0.0) | 0 (0.0) | 1 (5.0) | 2 (7.1) | 2 (18.2) | 2 (18.2) | 7 (8.6) |
| Combinations | 1 (14.3) | 1 (25.0) | 5 (25) | 11 (39.3) | 1 (9.1) | 3 (27.3) | 22 (27.2) |
| Variational RNN | 0 (0.0) | 0 (0.0) | 1 (5.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (1.2) |
| CNN only or LSTM/GRU+CNN | 0 (0.0) | 1 (25.0) | 4 (20.0) | 4 (14.3) | 0 (0.0) | 3 (27.3) | 12 (14.8) |
| DAG+GRU | 1 (14.3) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (1.2) |
| Dense only or LSTM/GRU+Dense | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (3.6) | 1 (9.1) | 0 (0.0) | 2 (2.5) |
| GAN+LSTM | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (3.6) | 0 (0.0) | 0 (0.0) | 1 (1.2) |
| GNN only or LSTM/GRU+GNN | 0 (0.0) | 0 (0.0) | 0 (0.0) | 4 (14.3) | 0 (0.0) | 0 (0.0) | 4 (4.94) |
| GRU+GCN | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (3.6) | 0 (0.0) | 0 (0.0) | 1 (1.2) |
| Machine learning | 1 (14.3) | 0 (0.0) | 0 (0.0) | 1 (3.6) | 0 (0.0) | 0 (0.0) | 2 (2.4) |
| Lasso-SVM | 1 (14.3) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (1.2) |
| Gradient boosting tree mimic | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1 (3.6) | 0 (0.0) | 0 (0.0) | 1 (1.2) |
| Number of layers | |||||||
| <3 | 6 (85.7) | 2 (50.0) | 6 (30.0) | 6 (21.4) | 4 (36.4) | 3 (27.3) | 27 (33.3) |
| 3-5 | 0 (0.0) | 0 (0.0) | 2 (10.0) | 9 (32.1) | 3 (27.3) | 1 (9.1) | 15 (18.5) |
| 6-10 | 0 (0.0) | 0 (0.0) | 2 (10.0) | 2 (7.14) | 1 (9.1) | 2 (18.2) | 7 (8.6) |
| Not reported | 1 (14.3) | 2 (50.0) | 10 (50.0) | 11 (39.3) | 3 (27.3) | 5 (45.5) | 32 (39.5) |
Categorical parameters are described as N (%). Some categories have been aggregated. Raw parameters are available in the Shiny app. N>81 is due to the same study considering different possibilities for the same parameter.