It would be nice to get the information about how many data points are used for an effect on symptom. This could help to better distinguish between backed up effects and effects that still need more data to be relevant.

Alternatively or additionally the possibility of setting a minimum data point threshold higher than 3 would also help.