Comparative evaluation of hybrid LSTM-Based Models for predicting bioactive compound contents and antioxidant activity in microwave-assisted extraction from carrots using natural deep eutectic solvents

What if the humble carrot could be mined more sustainably with green solvents and smart AI models? A new study shows how hybrid LSTM models predict bioactive extraction with near-perfect accuracy.

Carrots are more than a crunchy snack—they're a treasure trove of bioactive compounds. But extracting those compounds often relies on toxic solvents and long processing times. Enter natural deep eutectic solvents (NADES): green, sustainable alternatives that, when paired with microwave-assisted extraction (MAE), make the process faster and cleaner.

In this study, researchers ran 40 experiments, tweaking five key parameters: microwave power (400–600 W), temperature (50–70°C), time (10–40 min), sample mass (0.5–1.0 g), and lactic acid concentration (1–2 mol/L). They then tested three hybrid LSTM models—LSTM-RSM, LSTM-BRSM, and LSTM-Box Behnken—to predict total phenolic content, flavonoid content, and antioxidant activity.

The results? Sensitivity analysis revealed that microwave power drives phenolic content, sample mass controls antioxidant activity, and temperature governs flavonoid extraction. Among the models, LSTM-Box Behnken stood out with an R² > 0.99, showcasing remarkable predictive power.

This isn't just about carrots. It's a blueprint for combining green chemistry with advanced modeling—reducing toxic solvent use, cutting extraction time, and boosting efficiency. The future of sustainable extraction just got a little smarter.

Key Points

• NADES + MAE boosts carrot extraction efficiency and sustainability
• LSTM-Box Behnken predicts with R² > 0.99
• Microwave power, sample mass, temperature dominate key outcomes