Table 1.
Number identified by algorithm | Number identified by clinician | Percent completeness of algorithm compared to clinician reviewa | Percent agreement between algorithm and clinicianb |
Clinicians’ average confidence score
(−1, 0, 1) |
Standard deviation of clinicians’ confidence scores | |
---|---|---|---|---|---|---|
|
40 | 42 | 92.9 | 97.5 | N/A | N/A |
|
240 | 245 | 88.6 | 90.4 | N/A | N/A |
Outcome | ||||||
Live birth | 80 | 89 | 89.9 | 100.0 | 0.96 | 0.19 |
Stillbirth | 40 | 38 | 100.0 | 95.0 | 0.97 | 0.16 |
Abortion | 40 | 47 | 78.7 | 92.5 | 1 | 0 |
Ectopic pregnancy | 40 | 33 | 100.0 | 82.5 | 0.85 | 0.44 |
No outcome | 40 | 38 | 76.3 | 72.5 | N/A | N/A |
Gestational age at outcome (±1 week) | ||||||
Live birth | N/A | N/A | 98.8 | 0.71 | 0.46 | |
Stillbirth | N/A | N/A | 89.5 | 0.57 | 0.55 | |
Abortion | N/A | N/A | 67.6 | −0.34 | 0.61 | |
Ectopic pregnancy | N/A | N/A | 63.6 | −0.58 | 0.50 |
As these cases were randomly selected for clinician validation, they do not reveal information about the subpopulation and are not subject to the same policy regarding masking small numbers of patients.
Percent completeness of algorithm compared to clinician review is the number of matching algorithmically defined episodes out of all possible clinician-defined episodes for these categories.
Percent agreement between algorithm and clinician is the number of matching clinician-defined episodes out of all possible algorithmically defined episodes for these categories.