1. Table 1 lists dosage levels as −1 = 0.05 g, 0 = 0.010 g, +1 = 0.15 g, while Methods say 0.05–0.015 g and the optimum is 0.052 g. The central level is outside the stated range and likely a typo for 0.10 g. Also, 2 g PW-AC + 4 g KOH is a 1:2 ratio, not the stated 2:1. Please correct/clarify.
2. Table 2 kinetic equations are wrong/duplicated: PSO and IPD are both written as qt=q2k2tqt=q2k2t; IPD is missing t0.5t0.5; fractional power is written as qt=K1t0.5+Ctqt=K1t0.5+Ct. Since kinetic mechanism conclusions depend on these models, please provide the correct nonlinear equations and refit.
3. Table 7 reports quadratic lack-of-fit p<0.0001p<0.0001, but Table 8 reports 0.9291. In Table 8, AD has p=0.0253p=0.0253 (significant at 0.05), yet the text says AD is not significant. Please reconcile.
4. BET parameters are internally inconsistent: With SBET=484.8 m2/gSBET=484.8 m2/g, Vp=0.1756 cm3/gVp=0.1756 cm3/g, and dp=3.751 nmdp=3.751 nm, 4V/S4V/S gives ~1.45 nm, not 3.75 nm. Please state how mean pore diameter was obtained and correct the inconsistency.
5. Table 12 central-point ANN predictions are impossible for a deterministic ANN: Runs 5, 6, 10, 11 and 29 have identical inputs (20 min, pH 6, 0.1 g, 20 mg/L) but ANN predictions differ: 93.67, 92.80, 93.18, 92.53, 93.64. A trained ANN should give one output for identical inputs. This suggests table mislabeling, data leakage, or an incorrect model comparison.
6. Table 11 gives validation MSE = 0.01828, whereas Fig. 17 states best validation performance = 0.0018283 (10× lower). Table 11 also gives testing MSE = 2.2297 versus training 0.16768 and validation 0.01828, despite very high R values. Please reconcile and provide architecture/training details.
7. Table 13 error metrics are mathematically inconsistent: ANN Adj. R2R2 (0.9993) exceeds R2R2 (0.9991), and RSM Adj. R2R2 equals R2R2 exactly. Also, RSM MSE 0.01379 should give RMSE ≈0.117, not 0.03715; ANN MSE 0.001379 should give RMSE ≈0.0371, not 0.003715. Please correct the error-function table.
8. Langmuir gives qmax=123.93 mg/gqmax=123.93 mg/g and has R2=0.9969R2=0.9969, higher than Freundlich R2=0.9771R2=0.9771, yet Freundlich is declared “best” and used to support multilayer adsorption. Moreover, PSO gives qe=183.13 mg/gqe=183.13 mg/g, which exceeds the Langmuir qmaxqmax. Please reconcile model selection and capacity values.
9. Table 10 desirability values are not credible: Ten solutions are listed with desirability = 1.000 despite removal efficiencies ranging from 32.806% to 93.955%. If the response is removal efficiency, desirability cannot be 1 for both near-optimum and poor solutions. Please explain the desirability function and identify the actual optimum.
10. The text says removal increases with increasing pH, but also says acidic conditions are better and the optimum pH is 3.84; it says removal increases with decreasing concentration, while the optimum concentration is 29.79 mg/L near the upper bound. Please correct these statements. Also, PXRD cannot establish hydroxyl groups/hydrogen bonding; that claim should be supported by FTIR/XPS, not PXRD.
