Skip to main content
. Author manuscript; available in PMC: 2022 Apr 26.
Published in final edited form as: Ann Surg. 2020 Oct;272(4):629–636. doi: 10.1097/SLA.0000000000004133

TABLE 4.

False Negative Error Analysis of EasyCIE-SSI in the internal and External Validation Cohorts.

Error Type Cohort Description Example
Internal Validation (n=4) External Validation (n=155)
Documentation Not Present 2 (50%) 88 (56.8%) The documentation needed to identify an SSI was not available electronically. Patient treated at non-index facility post discharge,
Named Entity Recognition 20 (12.9%) The local terminology was not in the dictionary required to identify key terms related to SSI. Misspelling (i.e. “prulent”), Anatomic name not present in Training (i.e. “Morrison’s Pouch”)
Local Inference 1 (25%) 33 (21.3%) Sentence/Paragraph inferencing rules made an incorrect conclusion in aggregating the concepts. Concluded SSI was historical due to phrase “patient has a history of an SSI”
Document Inference 10 (6.5%) Document level inferencing rules made an incorrect conclusion aggregating the sentences and paragraphs. Antibiotics mentioned separately in document and not associated with SSI symptoms to conclude SSI Present
Patient Inference 1 (25%) 4 (2.6%) Patient level inferencing rules made an incorrect conclusion aggregating the documents conclusions Infection PATOS not identified as resolved prior to Organ Space SSI development.