Table 4.
Subset of CHECK periods selected by the conventional clinical criteria and the ML-L scenario.
| selection | N(1704) | P(358) | S(579) | P+S(160) | not N | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| abs | rel | recall | abs | rel | recall | abs | rel | recall | abs | rel | recall | recall | |
| conventional | 88 | 31% | 5% | 103 | 37% | 29% | 40 | 14% | 7% | 49 | 18% | 31% | 18% |
| ML–L | 296 | 38% | 17% | 183 | 24% | 51% | 203 | 26% | 35% | 96 | 12% | 60% | 44% |
The number of total instances of each category is reported next to the class name. For each category we report an absolute and relative number of included instances, and a recall percentage (how many instances of that category have been retrieved). The “not N” column shows the summarised recall percentage for all progressive instances.