Figure 1.
Reliability of weights of phenotypes during tumerogenesis.
The upper bound on the fractional error in the Lagrange multipliers, , at different successive time points in the WI-38 cancer model of Rotter et al [11]. A constraint is warranted by the data when the fractional error is below unity, see equation (14).
is the tumor signature and it is seen that it is only valid in later times but well before the cell is cancerous that is observed at time point 12. Note that the error in the steady state constraint is minimal.
Figure 2.
Soundness of weights of phenotypes during tumerogenesis.
The bound for the fractional error of the Lagrange multipliers, , for steady state (
) and the next 3 constraints
calculated for the four time points measured in the HPV-16 model [12].
Figure 3.
Reliability of weights of phenotypes for a renal cancer patient.
The error bound in the Lagrange multipliers for steady state () and the next 2 constraints
calculated for the 2nd patient of renal carcinoma, using the data reported measured by Stevanović et al [13] for patient number 2. Quite similar results are obtained for the other two patients.
Figure 4.
The importance of patient variability.
The statistical error bound in the Lagrange multipliers for steady state () and the next 2 constraints
calculated taking into account patient variability by using equation (16). Renal cancer data for 3 patients as reported in Stevanović et al [13]. The error bound for the 2nd constraint is high at all time points.