Beta >0 in Tables

[amarkelkar]

I'm working on a CEA and running into the issue of Amua telling me that the analysis won't run because the b-value in a Beta distribution must >0 [Beta(a,b,~)]. However, I've double-checked my tables and none of the listed values are 0. I also ran individual tables and identified 2 tables that were the issue, but I can't figure out why those two are having issues while the others are not. I've double-checked my code and there's nothing different about how I'm referencing those two tables: if (n_Tx_line==3, Beta(p_Progression_3rd_line_CAR_Table[n_CAR_months,6], p_Progression_3rd_line_CAR_Table[n_CAR_months,5], ~),0)

I'm not sure if this is an issue with interpolation creating some "0" values or something else ... has anyone else experienced this or have any advice?

[zward]

It sounds like it could be an interpolation issue. What type of interpolation are you using for the tables? If cubic splines, there may be some oscillations that dip below 0 (usually near the bounds). You could try calling/plotting different values returned from the table to see if this is the case. Another approach (more of a hack) could be to use abs() or max(), or add a small constant to ensure the values are positive, or interpolate in log-space and exponentiate to ensure b>0.

[amarkelkar]

Thanks for the suggestions. I've tried switching to Linear interpolation without any improvement. I also tried putting max() and abs() limits both inside and outside my Beta() statements without resolution either.

[zward]

If you send me a copy of the model (zward@hsph.harvard.edu) - or another reproducible example - I can try to debug to see what is going on.

[amarkelkar]

Thanks so much Zach! I just sent it to you.