From: Detecting inpatient falls by using natural language processing of electronic medical records
Data sources | Results of NLP analysis | Fall-positive | Fall-negative | Total | Sensitivity (Recall) | PPV (Precision) | Specificity | F-measure |
---|---|---|---|---|---|---|---|---|
 | Positive | 223 | 3,351 | 3,574 |  |  |  |  |
Progress notes | Negative | 0 | 209,064 | 209,064 | 1.00 | 0.06 | 0.98 | 0.12 |
 | Total | 223 | 212,415 | 212,638 |  |  |  |  |
 | Positive | 52 | 0 | 52 |  |  |  |  |
Incident reports | Negative | 0 | 476 | 476 | 1.00 | 1.00 | 1.00 | 1.00 |
 | Total | 52 | 476 | 528 |  |  |  |  |
 | Positive | 10 | 0 | 10 |  |  |  |  |
Image order entries | Negative | 2 | 19,672 | 19,674 | 0.83 | 1.00 | 1.00 | 0.91 |
 | Total | 12 | 19,672 | 19,684 |  |  |  |  |
 | Positive | 2 | 13 | 15 |  |  |  |  |
Discharge summaries | Negative | 0 | 1,149 | 1,149 | 1.00 | 0.13 | 0.99 | 0.24 |
 | Total | 2 | 1,162 | 1,164 |  |  |  |  |