Smart ice cream formulation optimization using AI and spectroscopy

Overview

This innovative solution leverages advanced spectroscopy, machine learning, and AI-driven models to revolutionize the ice cream formulation process. By integrating Design of Experiments (DoE) with multimodal data collection and hybrid modeling, this approach allows for rapid scenario testing and optimization of ice cream properties such as texture, mouthfeel, and stability. The incorporation of AI-driven ingredient interaction models and sensory prediction models ensures that the final product aligns with consumer preferences, thereby enhancing overall product quality and satisfaction.

Technical specifications

Key features:

• Utilizes Design of Experiments (DoE) to pinpoint critical process variables affecting ice cream properties.
• Employs hybrid modeling, combining multiphysics simulations and machine learning techniques like PLS, SVM, and deep learning, to predict product outcomes.
• Develops AI-driven ingredient interaction models using Graph Neural Networks (GNNs) to capture complex formulation effects.
• Trains sensory prediction models on consumer data to forecast taste and texture perceptions.
• Provides a user-friendly platform for R&D teams to simulate and optimize new product development efficiently.

Technology readiness level

This technology is at Technology Readiness Level 8, indicating that it has been validated in relevant environments and is ready for deployment in commercial applications. The models and methodologies have undergone extensive testing and are prepared for integration into product development pipelines.