Skip to main content

Table 4 Program evaluation and review technique estimate of time for a human to evaluate the digital media data collection

From: Sorting Through the Safety Data Haystack: Using Machine Learning to Identify Individual Case Safety Reports in Social-Digital Media

Variables Minimum Maximum Mode Exp SD L90%CI* U90%CI*
Posts (n) 311,189 311,189 311,189 311,189 0.0 311,189 311,189
Human reading speed (wpm) 136 232 178 180 16 153 206
Post length (words) 10 10,000 316 1879 1665 48 2101
Read/ID speed (min/post) 0.07 43.10 1.78 8.38 7.17 0.31 10.22
Total evaluation time (min) 22,882 13,413,319 554,001 2,608,701 2231,701 96,517 3,179,934
Total evaluation time (h) 381 223,555 9233 43,478 37,196 1609 52,999
  1. Exp exponential, ICSR individual case safety report, L90%CI lower 90% confidence interval, SD standard deviation, U90%CI upper 90% confidence interval, wpm words per minute
  2. * PERT Beta Function Approximate Confidence Intervals