How do I use a function containing symsum with lsqcurvefit?

I have the following code, which is to fit a function to some data:

time=[10,15,17,25,30,34,40,43,51,55,68,70,84,104]; fraction=[1/18,2/18,6/18,7/17,8/18,10/18,11/18,12/18,13/18,14/18,15/18,16/18,17/18,18/18];

P0=[0.02 7]; syms k;

F=@(Q,xdata)1-symsum(((Q(1).*(xdata-Q(2))).^k.*exp(-Q(1)*(xdata-Q(2))))/factorial(k),k,0,4);

[Q,resnorm,~,exitflag,output]=lsqcurvefit(F,P0,time,fraction);

However, I get the following error: Error using lsqcurvefit (line 244) LSQCURVEFIT requires all values returned by user functions to be of data type double.

I think this is because class(F(P0,time))=sym, whereas lsqcurvefit requires it to be double.

Is there anyway to get around this?

Thanks, Caroline