Chocolate makers are training machine learning models to design confectionery that is healthier to eat without sacrificing taste or shelf life. The problem is that improving chocolate’s nutritional profile—by boosting fibre or polyphenol content, for example—often clashes with the processing conditions that give chocolate its familiar texture and stability. Manufacturers lack a systematic way to balance nutrition, function, and cost. This project fills that gap by combining literature reviews, laboratory experiments, and machine learning to map how ingredients and processing steps affect both health benefits and product performance. If successful, the research will produce a set of predictive models that allow manufacturers to optimise recipes and processes before making physical prototypes. That could shorten product development cycles, reduce waste, and enable chocolate bars with demonstrably better nutritional profiles—such as higher fibre or retained polyphenols—while maintaining the qualities consumers expect. The approach also identifies potential new nutritional challenges facing the industry and explores sensor-based monitoring during production, which could improve quality control. The work is applied, not fundamental science. Its immediate value lies in giving the chocolate industry a data-driven tool to make incremental but real improvements to everyday snack foods.
View original technical description
Project aim: Determine a collated portfolio detailing the most important components of chocolate and related ingredients to provide health, wellbeing (e.g., superior nutrition), or functional (e.g., extended shelf-life) interest from literature review. Evaluate the potential benefits using legitimate ingredients and identify optimal processing routes for their respective retention using machine learning methods, whilst being respectful to potential new nutritional challenges. This will be supported by practical studies to: a. Develop a chocolate or confectionery snack with the best nutritional / functional profile. b. Develop machine learning models to 1) predict the nutritional/function profile of chocolate or confectionary snack products; 2) Determine the optimal processing conditions to maximize the nutritional/functional profile. c. Demonstrate at least one beneficial effect based on the main nutritional improvement. As an example, an area of investigation could be interest in cocoa fibres and/or polyphenols. Approach: 1. A literature search will be done to: a. Identify patents to determine which compounds have already been investigated by the academic and industrial communities, Docusign Envelope ID: 76D6F02C-F9BB-41B3-A0E6-7DB5E2AC8547 Mondelez CTP Studentship 13 b. In the free space, identify a full list of compounds of interest associated to the technologies (ingredient and/or process) to make innovative cocoa snack foods with improved nutritional and/or functional profile c. The functional, health and well-being compounds available in chocolate confectionery/cocoa snack foods and associated ingredients d. Create a dataset using values collected from literature to help build machine learning models e. Identify mathematical approaches to model relevant processes in chocolate manufacturing Carry out a design of experiments flexing identified ingredient and process factors to create nutritionally or functionally improved prototypes for evaluation and develop a dataset used to develop machine learning models. Use Bayesian optimisation and active learning to extend the design of experiments methods and guide the selection of laboratory trials. 3. Develop machine learning models using existing and prototype data to predict nutritional and functional performance and determine optimal processing condition. Combine mathematical equations and process simulation with data-driven machine learning methods via hybrid modelling. 4. Selection of the best prototypes based on measured composition, including all macronutrients as well as additional compounds such as polyphenols. Develop a robust methodology to select the best opportunities to maximise nutritional benefit whilst maintaining functionality and economic and environmental viability. 5. Develop the chosen product on a pilot scale and demonstrate the beneficial effect in relation to the improved nutritional or functional profile. 6. Characterise identified ingredients and monitor compounds of interest during chocolate production through the use of sensors. 7. Identify current and potentially new nutritional challenges that may impact the chocolate industry as well as traditional and novel processing steps that could be explored to optimise identified benefits.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know