Sensitivity would be the derivate of the curve fitted. Such nonlinear cases one can give the sensitivity for different regions, or an average value. For your curve, as the decay constant has 50% standard error I woud fit a line and use its slope as average sensitivity.

Thank you very much.

The response appears to be linear between 1 and 4 mg/ml, but this could be a statistical accident.

What is the measurement uncertainty in the data points?  Do you know that the response is described by the same equation when no enzyme is present?

If the wavelength uncertainty is 1 nm or higher, I would fit a straight line through all 4 points, as David suggests.

If the uncertainty is 0.1 nm or smaller, it is more likely that the 0 mg/ml point is an outlier, or is not described by the same underlying process.  In this case a straight line through the remaining 3 points could be more appropriate.

choose the linear parts from the curve and found the slop for each part(average) then togote the relation you can calculate the sensitivity. 

Hi,

How do you define the sensitivity of a curve? This is really necessary before calculating anything.

There are at least two questions pending in your question. What other functions did you try? Did you try a 4th order polynomial that would have given a better fit?

Besides, you certainly need more experimental points to empirically ascertain an exponential curve, unless your experiment is sustained by a theory predicting an exponential behaviour.

Charles