Encouraged by our ability to use an eNose to predict the
pleasantness of odorants within the training set (P,0.05 in 100%
of the 20 runs), we set out to test its performance with novel
odorants, i.e., odorants that were not available during the
algorithm development. We used the eNose to measure 22
essential oil odorant mixtures made of unknown components
(Supporting Table S1 - essential oils). We measured these oils using
the same parameters as in the learning phase, and used the same
previously developed algorithm to predict the pleasantness of these
odorant mixtures. We then asked 14 human participants to rate
twice the pleasantness of these odorants. The average correlation
of 30 runs between the machine prediction ratings and the
human’s median ratings was r = 0.6460.02 (P,0.0001 in all 30
runs; Figure 2A). We then calculated the correlation between each
human’s ratings and the median human rating. The correlation
was 0.7260.1, thus the machine-human correlation was 88%
(0.64/0.72*100 = 88) of the human to human correlation.
Although these odorants were novel, some of the participants in
this study had participated in the original model-building study as
well. To address the possibility of any bias introduced by this, we
repeated the study again with 17 new participants, and obtained a
similar correlation of r = 0.5960.03, P,0.0001), i.e., a machine-
human correlation that was 82% of the human to human
correlation.