File Name : figs1_beta_sensitivity_v2.pdf

Caption : sensitivity of the optical response to the empirical exponent beta. (a) representative signal trajectories for beta = 1.0, 1.5, and 2.0. (b) signal separation for the 100 vs 200 mg dl−1 pair. (c) minimum classification error averaged over three representative glucose pairs (black), with individual pairs shown as dashed lines. the average error approaches a broad low-error plateau for intermediate-to-large beta, so that further increases beyond this regime provide little improvement in discrimination. (d) optimal read-out time topt (mean and range across pairs) versus beta. the green band marks the practical 30–50 min window. small beta values shift topt to later times, whereas larger beta values compress the response toward earlier read-out. in this sense, beta ≈ 1.5 represents a practical intermediate regime that combines near-minimal classification error with an operationally convenient read-out window for the representative cases explored here.

File Name : figs2_transport_sweep.pdf

Caption : transport-regime sweep over effective peroxide diffusivity dh and peroxide-consumption rate kcons. the map identifies the transition from strongly surface-localised degradation to the nearly homogeneous limit, and clarifies when a 1d transport-aware treatment becomes necessary.

File Name : figs3_kinetic_robustness.pdf

Caption : robustness of the predicted read-out window and discrimination behaviour to moderate perturbations in the nominal kinetic parameters. panels a/c perturb the basal degradation rate as k0(1+δ) with kg fixed, whereas panels b/d perturb the glucose-to-degradation coupling as kg(1+δ) with k0 fixed, using δ = ±10%, ±20%, and ±30%. the practical operating window remains in the same approximate range under these parameter variations.

File Name : figs4_mm_vs_linear.pdf

Caption : comparison between the linearised glucose-to-peroxide mapping and the full michaelis–menten production law across representative km values. nonlinear saturation modifies the transient dynamics and generally reduces degradation severity relative to the linearised approximation, while preserving monotonic signal separation across the explored range.

File Name : figs5_mm_discrimination.pdf

Caption : impact of michaelis–menten enzymatic saturation on discrimination performance, using the gox-mediated model area-matched at gref = 200 mg dl−1, with σs = 2% and tread ≤ 60 min. (a) minimum classification error as a function of km for representative glucose pairs, compared against the linear surrogate reference. (b) corresponding optimal read-out time topt as a function of km. (c) steady-state peroxide contrast ratio versus km, showing saturation-induced compression relative to the linear limit. (d) representative signal traces for the 100 vs 140 mg dl−1 pair at nominal (km = 5 mm) and more strongly saturated (km = 2 mm) conditions. nonlinear saturation increases classification error quantitatively, especially for closer glucose pairs, while preserving monotonic signal separation and a practical read-out window in the explored range.