Artificial intelligence-based prediction of lycopene content in raw tomatoes using physicochemical attributes.

Area of Science:

• Agricultural Science
• Food Science
• Biotechnology

Background:

• Lycopene consumption is linked to reduced cancer and cardiovascular disease risks.
• Tomatoes are a primary source of lycopene, a key phytochemical.
• Estimating lycopene via HPLC is costly and time-intensive.

Purpose of the Study:

• To develop AI models for predicting lycopene in raw tomatoes.
• Utilize 14 physicochemical parameters for prediction.
• Explore salinity, TDS, EC, firmness, pH, TSS, TA, Hunter color values, TPC, TFC, and AOA.

Main Methods:

• Collected post-harvest data from over 100 raw tomatoes.
• Developed Linear Multivariate Regression (LMVR), Principal Component Regression (PCR), and Partial Least Squares Regression (PLSR) models.
• Employed 10-fold cross-validation (CV) for model training and evaluation.

Main Results:

• Principal Component Analysis revealed a strong positive association between lycopene, color value 'a', TPC, TFC, and AOA.
• The best LMVR model achieved R² (CV) of 0.70, RMSE (CV) of 8.48, and RMSE (Test) of 9.69.
• PCR and PLSR models showed comparable performance, indicating the potential of AI in lycopene estimation.

Conclusions:

• AI models provide a viable alternative for rapid lycopene estimation in tomatoes.
• Physicochemical parameters can effectively predict lycopene content.
• This approach supports identifying tomatoes rich in lycopene for health benefits.