Table 3.
The NLP engine was able to retrieve information from 30,498 full text bladder pathology reports. Because this analysis focused on urothelial carcinoma, data on grade, invasion, presence of muscularis propria, and presence of carcinoma in situ is only reported for the 20,515 reports with urothelial carcinoma
| Variable | Abstracted by NLP engine |
|---|---|
| Value | N (%) |
| Histology (n=30,498) | |
| Not from Bladder | 1,560 (6.3) |
| No Cancer | 5,577 (18.3) |
| Urothelial Carcinoma | 20,515 (67.3) |
| PUNLMP | 213 (0.7) |
| Other Histology* | 717 (2.4) |
| Missing | 1,916 (6.3) |
|
| |
| Grade (n=20,515) | |
| Low | 6,708 (32.7) |
| Intermediate | 1,271 (6.2) |
| High | 9,548 (46.5) |
| Undifferentiated | 194 (1.0) |
| Not stated / missing | 2,794 (13.6) |
|
| |
| Carcinoma in situ (n=20,515) | |
| Present | 2,630 (12.8) |
| Explicitly absent | 224 (1.1) |
| Not mentioned / missing | 17,661 (86.1) |
|
| |
| Invasion presence vs absence (n=20,515) | |
| Non-invasive | 7,869 (38.4) |
| Suspected invasion | 175 (0.9) |
| Invasive | 9,008 (43.9) |
| Not stated / missing | 3,463 (16.9) |
|
| |
| Invasion depth among reports with invasion or suspected invasion by NLP (n=9,183) | |
| Lamina propria | 3,741 (40.8) |
| Muscularis propria | 2,270 (24.7) |
| Perivesical / Other Organ | 283 (3.1) |
| Not stated / missing | 2,889 (31.5) |
|
| |
| Muscularis propria in specimen (n=20,515) | |
| Present | 10,305 (50.2) |
| Not present | 3,617 (17.6) |
| Not stated / missing | 6,593 (32.1) |
*
Other histology included squamous cell carcinoma, adenocarcinoma, small cell carcinoma, undifferentiated carcinoma, unspecified carcinoma, among others. PUNLMP = Papillary Urothelial Neoplasm of Low Malignant Potential.