Table 1.
Sample sentence pairs and annotations from the clinical semantic textual similarity data set.
| Ground trutha | Score | Observations | ||
| Sentence 1 | Sentence 2 |
|
Domain dependence | Comments |
| “The patient was taken to the PACUb in a stable condition.” | “The patient was taken to the post anesthesia care unit postoperatively for recovery.” | 5.0 | Domain specific | Clinical abbreviations |
| “Albuterol [PROVENTIL/VENTOLIN] 90 mcg/Act HFAc Aerosol 1-2 puffs by inhalation every 4 hours as needed.” | “Ipratropium-Albuterol [COMBIVENT] 18-103 mcg/Actuation Aerosol 2 puffs by inhalation two times a day as needed” | 3.5 | Domain specific | Medication instruction parsing |
| “Cardiovascular assessment findings include heart rate normal, atrial fibrillation with controlled ventricular response.” | “Cardiovascular assessment findings include heart rate, first degree AVd Block.” | 3.0 | Domain specific | Medical concept similarity and medical concept mapping |
| “He was prepped and draped in the standard fashion.” | “The affected shoulder was prepared and draped with the usual sterile technique.” | 3.0 | Domain independent | Alignment |
| “Musculoskeletal: Positive for gait problem, joint swelling and extremity pain.” | “Musculoskeletal: Negative for back pain, myalgias and extremity pain.” | 1.5 | Domain independent | Assertion classification (polarity) |
aItalics indicate the phrases within each sentence which correspond to the observations.
bPACU: post anesthesia care unit.
cHFA: hydrofluoroalkane.
dAV: atrioventricular.