Extended Data Fig. 6: Detection time points of all classes by EmbryoNet-Prime.
From: EmbryoNet: using deep learning to link embryonic phenotypes to signaling pathways
Plots of the number of detected phenotypes (dotted lines) for each class over time for EmbryoNet (pink diamonds), EmbryoNet-Prime (blue triangles) and human experts (green dots). The error envelopes show standard error of the mean. Solid lines show the fit of a sigmoid curve to the data. Gray boxes show major developmental periods. Different classes could be detected at different time points. –BMP, + RA, –Wnt, –Nodal and –Shh could be classified earlier by EmbryoNet-Prime than by humans. n[Normal]: 74, n[–BMP]: 119, n[+RA]: = 66, n[–Wnt]: 70, n[–FGF]: 74, n[–Nodal]: 110, n[–Shh]: 63, n[–PCP]: 57. Data also shown in Fig. 2g.
