Calculated average EC50 does not seem correct. Also, the standard deviation is difficult to interpret.
The database is calculating a geometric mean and associated standard deviation for EC50 averages. Since EC50s are log-based values, an arithmetic mean would be misleading. The standard deviation for geometric mean is actually a factor and should not be interpreted as +/- but rather multiply/divide by to get the range.
You can see the geometric mean setting if you navigate to the EC50 protocol readout definition for that assay and click edit. The averaging setting is in the middle of the form.