Artificial Neural Networks (ANN) Modeling for Ethanolic Propolis Extracts

Total phenolic content (TPC) is a critical quality parameter evaluating the bioactive properties of ethanolic propolis extracts. This study aimed to investigate the relationship between color parameters (Hunter Lab), Brix% (dry matter), and total phenolic content, antioxidant capacity in the ethanolic propolis extracts. Four different percentages of ethanol (96%, 90%, 80% and 70%) and four different propolis concentrations (40%, 30%, 20% and 10%) were used in the study. Total phenolic substance amounts, and antioxidant values ​​of the extracts were measured according to the ferric reducing power (FRAP) and 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging activity assays. Pearson correlation analysis and artificial neural network (ANN) modeling were utilized to examine these interactions. The study results showed that both ethanol percentage and propolis amount affected the amount of TPC in the extracts and accordingly the antioxidant capacity. A strong correlation between TPC and the Hunter L* color parameter, as well as Brix%, was identified through ANN modeling, yielding the predictive equation: TPC (mg GAE/mL)=[− 0.07×L + 0.87×Dry Matter − 0.0130]. The ANN-based model developed to predict total phenolic content (TPC) showed about 85% agreement with experimentally obtained values. However, it is predicted that the prediction accuracy of the model will improve with the addition of a larger and more diverse data set. In conclusion, ANN modeling offers a promising alternative for faster and economical evaluation of the quality of ethanolic propolis extracts. • Total phenolic content (TPC) indicates propolis extract quality. • High TPC correlates with increased color intensity and dry matter (Brix)%. • ANN modeling showed strong links between TPC, L* value, and Brix%. • TPC can be predicted from color and Brix% via ANN models.