Skip to main content
. 2022 Jun 21;24(6):e32867. doi: 10.2196/32867

Table 2.

Classification performance of the disease identification algorithma.

Disease category Campaigns in the reference set that mention disease category, n Precision (95% CI) Recall (95% CI) F1 score Accuracy (95% CI)
Cardiovascular diseases 82 0.92 (0.86-0.99) 0.74 (0.65-0.84) 0.82 0.94 (0.91-0.96)
Endocrine diseases 19 0.75 (0.54-0.96) 0.63 (0.41-0.85) 0.69 0.97 (0.96-0.99)
Gastrointestinal diseases 18 0.56 (0.33-0.79) 0.56 (0.33-0.79) 0.56 0.96 (0.94-0.98)
Genitourinary diseases 35 0.97 (0.90-1.03) 0.8 (0.67-0.93) 0.88 0.98 (0.97-0.99)
Infections 30 0.56 (0.41-0.71) 0.77 (0.62-0.92) 0.65 0.94 (0.91-0.96)
Injuries and external causes 53 0.69 (0.58-0.80) 0.92 (0.85-1.00) 0.79 0.94 (0.91-0.96)
Mental health disorders 20 0.48 (0.30-0.66) 0.7 (0.50-0.90) 0.57 0.95 (0.93-0.97)
Musculoskeletal diseases 45 0.64 (0.48-0.80) 0.51 (0.37-0.66) 0.57 0.91 (0.88-0.94)
Neoplasms 162 0.95 (0.91-0.98) 0.98 0.96-1.00) 0.96 0.97 (0.95-0.99)
Nervous system diseases 66 0.88 (0.76-0.99) 0.42 (0.31-0.54) 0.57 0.90 (0.86-0.93)
Respiratory diseases 29 0.92 (0.81-1.03) 0.76 (0.60-0.91) 0.83 0.98 (0.96-0.99)

aThe average precision, recall, F1 score, and accuracy values are 0.83, 0.77, 0.78, and 0.95, respectively. Classification performance is based on a comparison to 400 campaigns that were annotated by a team of expert coders. The averages are weighted by the number of campaigns in the reference set that mention each disease category.