LILLE 2024

The perception and appreciation of food flavor depends on many interacting chemical compounds and external factors, and therefore proves challenging to understand and predict. Here, we combine chemical and sensory analyses of 250 different beers to train machine learning models that predict flavor and consumer appreciation. For each beer, we measured over 200 chemical properties, performed quantitative descriptive sensory analysis with a trained tasting panel and mapped data from over 180,000 consumer reviews to train 10 different machine learning models. This yielded models that significantly outperform predictions based on conventional statistics and accurately predict food features and consumer appreciation from chemical profiles. Model dissection allowed identifying specific and unexpected compounds as drivers of beer flavor and appreciation. Adding these compounds to commercial alcoholic and non-alcoholic beers resulted in improved consumer appreciation. In this talk, we reveal how big data and machine learning uncover complex links between food chemistry, flavor and consumer perception, and lay the foundation to develop novel, tailored foods with superior flavors.

A variety of malts helps brewers to impart distinct taste and smell characteristics to the beer. However, it is not fully understood what sensory characteristics the individual malt varieties have and what makes them different.
In this study, we tried to capture the general characteristics of mainly relatively light specialty malts (15-60 EBC). To achieve this, a congress mash with the same colour was carried out with pilsner malt and about 20 specialty malts each. These congress worts were compared in terms of sensory and chemical properties.
The results showed that several malt varieties with the same variety name and the same malt colour grade have completely different properties. In addition, brewing trials were conducted to determine how the differences between them affect the beer quality and what is essential for the preferred sensory property.
This knowledge would be helpful for brewers to add the preferred properties and to capture the crucial impact on the sensory properties of the special malts in a brewing process.

Besides a different flavour, non-alcoholic beer suffers from an altered texture profile. The lack of production or removal of ethanol introduces a myriad of changes in the beer mouthfeel, often subdivided into carbonation, fullness and afterfeel. The goal of this presentation is to showcase the effect of missing ethanol in beer through five key experiments we performed for each of these aspects of mouthfeel. We observed that non-alcoholic beers have a higher surface tension affecting bubble formation. This impacts the pearling sensation on the tongue as well as the foam head. Moreover, this effect increases during drinking, as a non-alcoholic beer loses CO2 faster compared to an alcoholic one. Regarding fullness, removing ethanol led to decreased viscosity and increased density. The various aspects of mouthfeel even affect each other as the reduced bubble formation additionally impairs the perceived fullness and cleansing of the tongue after drinking. Practical solutions are working with new ingredients, changing carbonation levels or a complete redesign of the brewing process. These insights will provide brewers with the tools to develop better non-alcoholic beers.