Practical implementation of multivariate calibration models has been limited in several areas due to the requirement of appropriate development and validation to prove their performance to standardization agencies. Herein, a detailed description of the application of multivariate calibration based on partial least-squares regression models (PLSR) for the determination of soluble solids (BRIX), polarizable sugars (POL), and reducing sugars (RS) in sugar cane juice, based on near infrared spectroscopy (NIR), for the alcohol industries is presented. The development of the models, including variable selection and outlier elimination, and their validation by determination of figures of merit, such as accuracy, precision, sensitivity, analytical sensitivity, prediction intervals, and limits of detection and quantification, are described for a representative data set of 1381 sugar cane samples. Values estimated by PLSR are compared with appropriate reference methods, where the results indicated that the PLSR models can be used in the alcohol industry as an alternative to refractometry and lead clarification before polarization measurements (standard methods for BRIX and POL, respectively). For RS, the results of a titration reference method were compared with the PLSR estimates and also with an estimate based on BRIX and POL values, as actually used in the alcohol industry. The PLSR method presented a better agreement with the titration method. However, the results indicated that the RS estimates from both PLSR and those based on the BRIX and POL values, actually used, should be improved to a safe determination of RS.

No abstract

Recebido em 25/6/08; aceito em 15/12/08; publicado na web em 28/5/09 STATE OF THE ART OF FIGURES OF MERIT IN MULTIVARIATE CALIBRATION. The validation of an analytical procedure must be certified through the determination of parameters known as figures of merit. For first order data, the acuracy, precision, robustness and bias is similar to the methods of univariate calibration. Linearity, sensitivity, signal to noise ratio, adjustment, selectivity and confidence intervals need different approaches, specific for multivariate data. Selectivity and signal to noise ratio are more critical and they only can be estimated by means of the calculation of the net analyte signal. In second order calibration, some differentes approaches are necessary due to data structure.

No abstract

Os teores de sólidos solúveis (BRIX), açúcares polarizáveis (POL) e açúcares redutores (RS) em caldo de cana foram determinados utilizando espectroscopia na região do infravermelho próximo (NIR) e calibração multivariada, a qual foi validada pelo cálculo das figuras de mérito. Devido à heterogeneidade das amostras, foi necessário, como primeira etapa do trabalho, a otimização dos conjutos de calibração e validação através da eliminação das amostras anômalas (outliers). As figuras de mérito como: sensibilidade, sensibilidade analítica, seletividade, limites de confiança, precisão (média, repetibilidade), exatidão e razão sinal-ruído foram calculadas. Resultados viáveis foram obtidos para BRIX e POL apresentando resultados de RMSEP de 0,28 e 0,42% de caldo, respectivamente. Os coeficientes de correlação para ambos os parâmetros foram de 0,99. Estes resultados indicam que os modelos desenvolvidos para BRIX e POL podem ser seguramente utilizados como uma alternativa em relação ao método padrão utilizado na indústria alcooleira.The determination of soluble solids (BRIX), polarizable sugars (POL) and reducing sugars (RS) sugar cane juice by using near infrared spectroscopy (NIR) and multivariate calibration was developed and validated by calculation of figures of merit. Due to the heterogeneity of the samples, first it was necessary to optimize the calibration set by outlier elimination. The figures of merit such as: sensitivity, analytical sensitivity, selectivity, confidence limit, precision (mean, repeatability), accuracy and signal-to-noise ratio were calculated. Feasible results were obtained for BRIX and POL with RMSEP values of 0.28 and 0.42% of juice and precision of 0.02 and 0.08% of juice, respectively. For both BRIX and POL goodness of fit showed correlation coefficients of 0.99. These results indicate that the models developed for BRIX and POL can be used as an alternative to standard procedures in the sugar cane industry.