TABLE 3.
Safety and appropriateness of triage recommendations for doctors and the Babylon Triage and Diagnostic System (Babylon AI) against a range of acceptable recommendations provided by GPs.
| GP-1 | GP-2 | GP-3 | ||||
|---|---|---|---|---|---|---|
| Safety (%) | Appr. (%) | Safety (%) | Appr. (%) | Safety (%) | Appr. (%) | |
| (95% CI) | (95% CI) | (95% CI) | ||||
| Doctor A | 97.9 | 89.4 | 91.5 | 83.0 | 95.7 | 89.4 |
| Doctor B | 79.5 | 75.6 | 60.3 | 59.0 | 75.6 | 74.4 |
| Doctor C | 97.9 | 89.6 | 93.8 | 89.6 | 95.8 | 93.8 |
| Doctor D | 80.4 | 76.5 | 64.7 | 62.8 | 86.3 | 84.3 |
| Doctor E | 84.3 | 78.6 | 70.0 | 67.1 | 80.0 | 78.6 |
| Doctor F | 92.2 | 86.3 | 74.5 | 68.6 | 92.2 | 84.3 |
| Doctor G | 92.2 | 88.2 | 72.6 | 70.6 | 84.3 | 80.4 |
| Doctor average | 89.2 | 83.5 | 75.3 | 71.5 | 87.1 | 83.6 |
| — | (82.5–95.9) | (78.1–88.8) | (64.4–86.3) | (62.1–80.9) | (80.4–93.8) | (78.0–89.2) |
| Babylon AI | 90.0 | 74.0 | 81.0 | 75.0 | 90.0 | 81.0 |
The AI powered System gives safer triage recommendations than the doctors on average, at the expense of a slightly lower appropriateness.