TABLE 1.
Summary characteristics of algorithm developments, external validations, and clinical utility assessments
| Characteristic | Algorithm development (n = 433) | External validations (n = 481) | Clinical utility assessments (n = 52) |
|---|---|---|---|
| Publication year, n (%) | |||
| 2000 and before | 7 (1.6) | ‐ | ‐ |
| 2001 to 2005 | 12 (2.8) | 3 (0.6) | 2 (3.8) |
| 2006 to 2010 | 75 (17.3) | 81 (16.8) | 12 (23.1) |
| 2011 to 2015 | 175 (40.4) | 224 (46.6) | 18 (34.6) |
| `2016 to 2020 | 164 (37.9) | 173 (36.0) | 17 (32.7) |
| Sample size, median (range) | 229 (18–10,673) | 125 (28–2,181) | 234 (10–2,343) |
| Participants (included), n (%) | |||
| ≥5% white | 186 (43.0) | 205 (42.6) | 36 (69.2) |
| ≥5% Asian | 210 (48.5) | 277 (57.6) | 17 (32.7) |
| ≥5% black | 121 (27.9) | 115 (23.9) | 16 (30.8) |
| ≥5% mixed/other | 77 (17.8) | 62 (12.9) | 2 (3.8) |
| Adults | 422 (97.5) | 455 (94.6) | 49 (94.2) |
| Children | 11 (2.5) | 26 (5.4) | 3 (5.8) |
| Location, n (%) | |||
| Africa a | 2 (0.5) | 2 (0.4) | ‐ |
| Asia b | 175 (40.4) | 208 (43.2) | 14 (26.9) |
| Europe | 34 (7.9) 121 | 55 (11.4) | 11 (21.2) |
| North America | 136 (31.4) | 121 (25.2) | 25 (48.1) |
| South America | 15 (3.5) | 21 (4.4) | ‐ |
| Middle East | 30 (6.9) | 25 (5.2) | 2 (3.8) |
| Oceania | ‐ | 8 (1.7) | ‐ |
| Multiple | 41 (9.5) | 41 (8.5) | ‐ |
| Covariates included, n (%) | |||
| Clinical c only | 87 (20.1) | 49 (10.2) | 11 (21.2) |
| Genetic only d | 2 (0.5) | ‐ | ‐ |
| Clinical c and genetic | 344 (79.4) | 432 (89.8) | 41 (78.8) |
| Application time, n (%) | |||
| Dose initiation | 373 (86.1) | 443 (92.1) | 40 (76.9) |
| Dose revision | 41 (9.5) | 31 (6.4) | 10 (19.2) |
| Both initiation and revision e | 19 (4.4) | 7 (1.5) | 2 (3.8) |
| Modelling techniques, n (%) | |||
| Artificial neural network | 32 (7.4) | 2 (0.4) | 1 (1.9) |
| Multiple linear regression | 280 (64.7) | 458 (95.2) | 47 (90.4) |
| Nonlinear mixed effects f | 14 (3.2) | 7 (1.5) | 3 (5.8) |
| Support vector regression | 27 (6.2) | 2 (0.4) | ‐ |
| Other g | 66 (15.2) | 9 (1.9) | ‐ |
| Unclear | 10 (2.3) | 3 (0.6) | 1 (1.9) |
| Algorithm presentation, n (%) | |||
| Computer program h | 10 (2.3) | 4 (0.8) | 4 (7.7) |
| Nomogram/table | 9 (2.1) | 3 (0.6) | ‐ |
| Regression formula | 239 (55.2) | 453 (94.2) | 47 (90.4) |
| None | 175 (40.4) | 21 (4.4) | 1 (1.9) |
a
Excludes Egypt, which is under Middle East.
b
Mostly China (131 algorithm developments, 120 external validations and 11 clinical utility assessments). This was followed by South Korea (16 algorithm developments, 59 external validations and 1 clinical utility assessment) and Japan (10 algorithm developments and 14 external validations).
c
Clinical includes clinical, demographic, and environmental variables.
d
Clinical factors also considered during the modelling.
e
All incorporate pharmacokinetic and/or pharmacodynamic techniques.
f
Used to fit pharmacokinetic/pharmacodynamic‐based algorithms.
g
See Table S6 for details.
h
Or online tool.