Prediction of flavonoid components and antioxidant activity in tangerine peel based on machine learning

Tangerine peel is a valuable citrus by-product enriched with flavonoids and antioxidant compounds, possessing considerable potential for applications in the functional food and nutraceutical industries. Conventional analytical techniques for flavonoid determination are often labor-intensive, time-consuming, and unsuitable for high-throughput quality evaluation. This study developed a machine learning-based prediction framework for the rapid assessment of major flavonoid components and antioxidant activity in tangerine peel using physicochemical characteristics and near-infrared spectral information. A total of 320 tangerine peel samples were analyzed, incorporating moisture content, soluble solids, color parameters, total phenolic content, and spectral variables as predictors. Hesperidin, nobiletin, and tangeretin concentrations were quantified by high-performance liquid chromatography, while antioxidant activity was determined using the DPPH radical scavenging assay. Three machine learning algorithms, including Support Vector Regression (SVR), Random Forest (RF), and Extreme Gradient Boosting (XGBoost), were evaluated and compared. The results demonstrated that XGBoost exhibited the highest predictive accuracy, achieving coefficient of determination (R²) values of 0.945, 0.921, and 0.907 for hesperidin, nobiletin, and tangeretin, respectively. Antioxidant activity prediction achieved an R² of 0.959, with a root mean square error (RMSE) of 2.73%, which demonstrates the reliability and generalization capability of the model. Feature importance analysis revealed that total phenolic content (24.8%), absorbance at 1450 nm (18.6%), and peel color parameter a* (12.9%) were the most influential predictors. Compared with conventional laboratory analysis, the proposed framework reduced analytical time from approximately 51.6 min per sample to less than 2.5 s while maintaining prediction accuracy above 95%. These findings confirm that machine learning integrated with spectral sensing technologies provides an efficient, accurate, and non-destructive strategy for predicting flavonoid composition and antioxidant capacity, offering substantial potential for intelligent quality control and high-value utilization of citrus processing by-products.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-11-15
DOI
https://doi.org/10.1038/s41598-026-74716-9
Primary Topic
Spectroscopy and Chemometric Analyses
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Prediction of flavonoid components and antioxidant activity in tangerine peel based on machine learning

Chao Chi, Dongyang Cui, Chunmin Yang
Scientific Reports
Spectroscopy and Chemometric Analyses
article

Prediction of flavonoid components and antioxidant activity in tangerine peel based on machine learning

Chao Chi, Dongyang Cui, Chunmin Yang
article en

Abstract

Tangerine peel is a valuable citrus by-product enriched with flavonoids and antioxidant compounds, possessing considerable potential for applications in the functional food and nutraceutical industries. Conventional analytical techniques for flavonoid determination are often labor-intensive, time-consuming, and unsuitable for high-throughput quality evaluation. This study developed a machine learning-based prediction framework for the rapid assessment of major flavonoid components and antioxidant activity in tangerine peel using physicochemical characteristics and near-infrared spectral information. A total of 320 tangerine peel samples were analyzed, incorporating moisture content, soluble solids, color parameters, total phenolic content, and spectral variables as predictors. Hesperidin, nobiletin, and tangeretin concentrations were quantified by high-performance liquid chromatography, while antioxidant activity was determined using the DPPH radical scavenging assay. Three machine learning algorithms, including Support Vector Regression (SVR), Random Forest (RF), and Extreme Gradient Boosting (XGBoost), were evaluated and compared. The results demonstrated that XGBoost exhibited the highest predictive accuracy, achieving coefficient of determination (R²) values of 0.945, 0.921, and 0.907 for hesperidin, nobiletin, and tangeretin, respectively. Antioxidant activity prediction achieved an R² of 0.959, with a root mean square error (RMSE) of 2.73%, which demonstrates the reliability and generalization capability of the model. Feature importance analysis revealed that total phenolic content (24.8%), absorbance at 1450 nm (18.6%), and peel color parameter a* (12.9%) were the most influential predictors. Compared with conventional laboratory analysis, the proposed framework reduced analytical time from approximately 51.6 min per sample to less than 2.5 s while maintaining prediction accuracy above 95%. These findings confirm that machine learning integrated with spectral sensing technologies provides an efficient, accurate, and non-destructive strategy for predicting flavonoid composition and antioxidant capacity, offering substantial potential for intelligent quality control and high-value utilization of citrus processing by-products.

Scientific Reports
Guangzhou College of Technology and Business (CN)
Spectroscopy and Chemometric Analyses
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon