A hybrid of wavelet analysis and intelligent fuzzy inference system based on aggregation of gold nanoparticles for the rapid concurrent colorimetric sensing of trace anti-benign prostatic hyperplasia drugs in biological samples

This study introduces a colorimetric method for the detection of prazosin (PRZ) and terazosin (TER) simultaneously in binary mixtures, pharmaceutical formulations, and biological sample based on gold nanoparticles (AuNPs) using discrete wavelet transform (DWT) combined with principal component analysis (PCA) and fuzzy inference system (FIS). Due to the aggregation, the surface plasmon resonance (SPR) band of AuNPs reduced to 524 nm and a new shifted band at about 690 nm appeared which increases gradually by increasing concentration of drugs. The shape and size of the prepared NPs before and after aggregation were appraised through valid identification methods. The effect of experimental factors (pH, buffer type, buffer volume, concentration of AuNPs, interaction time, electrolyte type, and electrolyte concentration) on absorption ratio was investigated. The linearity was obtained in the range of 25–200 and 20–200 μg/L with R² of 0.9905 and 0.9958 for PRZ and TER, respectively. The LOD of PRZ and TER were 25.38 and 18.70 μg/L, respectively, while the LOQ were reported 76.92 and 56.66 μg/L for PRZ and TER, respectively. In the colorimetric method integrated with DWT, the absorption of the mixtures was analyzed using Daubechies wavelet with second and fourth order (db2, db4) and Harr wavelet. The outputs of each wavelet family were separately dimensioned using the PAC method and considered as FIS input. Db4 wavelet with R² higher than 0.999, RMSE of lower than 1.4, and mean recovery of higher than 99.00 % was chosen as the best wavelet for both components. The proposed method was compared with HPLC as a reference method through one-way analysis of variance (ANOVA) test. The colorimetric method in combination with DWT, PCA, and FIS was successfully used to determine PRZ and TER in urine samples with relative standard deviation (RSD) less than 0.8 % and less than 0.7 %, respectively. This method showed good precision and accuracy with simplicity and high speed in the simultaneous analysis of drugs without the need for complex and expensive devices. Also, it can be an environmentally friendly alternative to HPLC for the concurrent estimation of drugs.