Marcel writes in his mail of  9 Jun 2020 13:01:00 +0000:

In simrisc.cc the sensitivity of mammography is written as (line 305)

if (rndPoolSensitivity[ screeningRound*(modality+1) ] <= (0.9*((exp(beta1 +
(beta2*tumor.diameter()) + (beta3*m) +
(beta4*((m)/(pow(tumor.diameter(),2)))))) / (1 + exp(beta1 +
(beta2*tumor.diameter()) + (beta3*m) +
(beta4*((m)/(pow(tumor.diameter(),2))))))))) 
    //Found a tumor with the screening

This function, called the beta-function, indeed applies to mammography (as he
wrote, which I then overlooked)

This function implements W.A. van Veldhuizen (2019) formula [2], where the
factor 'se' is set to 0.9 in the above formula.

If the above condition is not true then the screening didn't find a tumor, and
a false negative *may* be encountered.

Whether a false negative is encountered is determined by a random
value from the uniform distribution: if that value exceeds the bi-rads 
sensitivity of that age then a false negative is encountered. 

E.g., for mammography the sensitivity values for the four bi-rad categrories
could be 0.87 0.84 0.73 and 0.65. If the (0-based) bi-rad categorie is 2 then
a false negative is encountered if the encountered random value exceeds 0.73.


For tomosynthesis and MRI the beta-function is not used, but other than that 
false negatives are determined as with mammography. 

For MRI the sensitivity value is fixed and doesn't depend on bi-rad
categories. E.g., when configured as 0.94 false negatives occur if the random
values exceed 0.94.

