Quantitative analysis of tramadol and paracetamol in tablets by spectroscopic and chemometric methods

Razvijeni su novi analitički postupci temeljeni na spektroskopskim metodama i kemometričkim izračunima za istovremeno kvantitativno određivanje tramadola i paracetamola u tabletama. Spektroskopski podaci dobiveni analizom tableta primjenom UV-Vis spektrofotometrije, spektroskopije u bliskom infracrvenom području (NIR) te Ramanove spektroskopije obrađeni su upotrebom metode djelomičnih najmanjih kvadrata (PLS), genetičkih algoritama spregnutih s PLS-om (GA-PLS) i umjetnih neuronskih mreža (ANN). Rezultati dobiveni kemometričkom obradom spektroskopskih podataka statistički su uspoređeni s onima određenima validiranom metodom tekućinske kromatografije ultravisoke djelotvornosti (UHPLC). Dobiveno je dobro slaganje između količina paracetamola određenih u tabletama primjenom UV spektrofotometrije te PLS, GA-PLS i ANN algoritama i referente UHPLC metode. U slučaju tramadola GA-PLS rezultati pokazali su se pouzdaniji u odnosu na one dobivene PLS-om i ANN-om. Najjednostavniji i najpouzdaniji modeli konstruirani su korištenjem PLS metode za paracetamol i GA-PLS metode za tramadol. Maseni udjeli paracetamola u tabletama određeni primjenom NIR spektroskopije i kalibracijskih kemometričkih modela izrađenih pomoću PLS, GA-PLS i ANN metoda vrlo se dobro slažu s onima dobivenima UHPLC-om. Rezultati dobiveni primjenom ANN-a i GA-PLS-a međusobno se ne razlikuju znatno, te su pouzdaniji od onih izračunanih upotrebom PLS metode. S druge strane, udjeli tramadola određeni NIR metodom spregnutom s GA-PLS algoritmom znatno su pouzdaniji od onih izračunanih upotrebom druge dvije metode. GA-PLS pristupom obradi podataka pojednostavio se je kalibracijski model te poboljšala točnost i preciznost dobivenih rezultata. Ramanova spektroskopija spregnuta s PLS, GA-PLS i ANN kemometričkim metodama uspješno je upotrijebljena za kvantitativnu analizu paracetamola, pri čemu su rezultati dobiveni tim metodama usporedivi s referentnim vrijednostima. Algoritam GA-PLS još se jednom pokazao superiornijim u odnosu na PLS i ANN. U slučaju analize tramadola, nijednom uporabljenom metodom nisu dobiveni rezultati usporedivi s referentnim vrijednostima. (112Several new analytical procedures based on spectroscopic methods and chemometric calculations were developed for quantitative determination of tramadol and paracetamol in tablets. Spectroscopic data obtained by analyses of tablets using UV-Vis spectroscopy, spectroscopy in near infrared region (NIR), and Raman spectroscopy were processed by means of partial least squares (PLS), genetic algorithm coupled with PLS (GA-PLS) and artificial neural networks (ANN). Results gained by chemometric processing of the spectroscopic data were statistically compared with those obtained by means of validated ultra-high performance liquid chromatographic (UHPLC) method. A good agreement between the amounts of paracetamol determined in the tablets using UV spectrophotometry and PLS, GA-PLS, as well as ANN algorithms, and that obtained by referent UHPLC method was found. In the case of tramadol, GA-PLS results were proven to be more reliable compared to those of PLS and ANN. The simplest and the most reliable models were constructed by using PLS method for paracetamol and GA-PLS method for tramadol. Mass fractions of paracetamol in the tablets determined by means of NIR spectroscopy and calibration chemometric models constructed using PLS, GA-PLS, and ANN methods were in accordance with those gained by UHPLC. There was no significant difference between the results obtained by using ANN and GA-PLS, and they were more reliable than that obtained by PLS method. On the other hand, fractions of tramadol determined by NIR method coupled with GA-PLS algorithm were considerably more reliable than those calculated using other two methods. GA-PLS-based approach reduced the calibration model complexity and improved the accuracy and precision of the results. Raman spectroscopy coupled with PLS, GA-PLS, and ANN chemometric methods was successfully applied for quantitative analysis of paracetamol, whereby the results obtained by these methods were comparable with the referent values. GA-PLS algorithm was once again proven to be superior over the PLS and ANN. In the case of tramadol analysis, the results gained by applying any of the used method were not comparable with the referent values