Leveraging machine learning techniques to analyze nutritional content in processed foods
Abstract The global shift towards plant-based diets, particularly in India, is driven by environmental and ethical considerations. While plant foods are often regarded as more sustainable, concerns persist regarding protein quality, especially after processing. With protein deficiencies being preval...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2024-12-01
|
| Series: | Discover Food |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44187-024-00253-x |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract The global shift towards plant-based diets, particularly in India, is driven by environmental and ethical considerations. While plant foods are often regarded as more sustainable, concerns persist regarding protein quality, especially after processing. With protein deficiencies being prevalent among Indians, it is crucial to understand the impact of food processing on nutrient retention. This research integrates machine learning with food science to develop a comprehensive AI framework for forecasting the protein content of various plant-based sources following both traditional and non-conventional processing methods. A robust database was compiled using sources such as Web of Science, Scopus, PubMed, and Google Scholar, covering a wide range of plant-based foods and their protein content before and after processing. After data preprocessing, two primary machine learning algorithms were employed: Support Vector Regression (SVR) and Random Forest (RF), both implemented using Scikit-learn. The SVR model was optimized to identify the best-fitting hyperplane in high-dimensional space, while the RF model utilized GridSearchCV for hyperparameter tuning and performed a “Feature Importance Analysis” to identify key factors influencing the outcomes. Model performance was evaluated using Normalized Mean Squared Error (NMSE) as the evaluation metric. The results indicated that the RF model achieved an NMSE of approximately 0.35, reflecting a moderate level of prediction error relative to data variance. In contrast, the SVR model significantly outperformed the RF model, with an NMSE of approximately 0.03, demonstrating superior accuracy and efficiency in predicting nutrient retention. This study leverages machine learning to bridge a critical gap in understanding nutrient retention in plant-based foods during processing. The findings reveal that the SVR model is particularly effective in predicting nutrient retention, outperforming the RF model. This novel approach holds significant potential to optimize nutrient retention in plant-based food products, offering important implications for public health and food quality. |
|---|---|
| ISSN: | 2731-4286 |