Using ECxEstimator

Issues that occurred during ECxEstimator testing

• If the considered exposure profiles differ from those in calibration data in ECx calculation, use x_in. x_in needs to be an x-array that can be generated with: x_in = PymobSimulator._exposure_data_to_xarray(exposure_data=exposure_data_dict, dim="substance"), exposure profile in x_in is different from exposure profiles in sim-object
• sim-object can be modified after calibration
• time needs to be defined in the ECxEstimator function; it should be a day that is part of the time coordinate (from the exposure profile), it is best to set the end time here (last measurement): ECxEstimator(sim=sim, effect='survival', x=0.5, time=int(x_in['time'][-1]), id='sub', x_in = x_in)
• warnings should appear if (1) no mortality can be observed until time that was set (no exposure?) and/or (2) background mortality will lead to mortality and is the dominant factor across the observed time (long observation time can lead to extinction)
• recommendation: if 'EC_50 calculation did not terminate successfully': increase accept_tolerance until calculation works, then look at survival plot (plt_profile_and_effect()) to find possible explanations (e.g. more than 50% mortality at investigated time point or no mortality at all, see warnings), if reason for failed calculation is known: adapt starting value log_x0 (default: 0)