Table 1. Summary of training and test datasets.
Datasets | Measures | IH | UPMC |
---|---|---|---|
Training | Encounters datesa | 1/2008 to 5/2010 | 1/2008 to 5/2010 |
# of encounters | 47,504 | 41,189 | |
# of influenza encounters | 1,858 | 915 | |
# of NI-ILI encounters | 15,989 | 3,040 | |
# of other encounters | 29,657 | 37,234 | |
# of clinical notes | 60,344 (1.2 per encounter) | 76,467 (1.9 per encounter) | |
# of finding extracted by UPMC parser | 877,377 (18 per encounter) | 1,031,134 (25 per encounter) | |
# of finding extracted by IH parser | 934,414 (20 per encounter) | 849,932 (21 per encounter) | |
Test | Encounters dates | 6/2010 to 5/2011 | 6/2010 to 5/2011 |
# of encounters | 182,386 | 238,722 | |
# of influenza encounters | 661 | 339 | |
# of NI-ILI encounters | 5,722 | 1,567 | |
# of other encounters | 176,003 | 236,816 | |
# of clinical notes | 220,276 (1.2 per encounter) | 480,059 (2 per encounter) | |
# of findings extracted by UPMC parser | 2,822,282 (15 per encounter) | 6,305,782 (26 per encounter) | |
# of findings extracted by IH parser | 2,950,928 (16 per encounter) | 5,361,241 (22 per encounter) |
aFor training purposes, we only used other encounters during the summer period from July 1, 2009 to August 31, 2009.