Optimising Sensory Evaluation in Oral Healthcare Products Using Machine Learning
Case Study
Sensory panels are integral to the development of consumer healthcare products, particularly oral healthcare products like toothpaste and mouthwash. These panels consist of trained individuals who assess various sensory attributes such as taste, aroma, texture, and appearance. Sensory panel evaluation is crucial for determining consumer acceptance and preference, ensuring products achieve the desired sensory performance. Sensory perception in oral healthcare products involves a complex interplay of taste buds and olfactory receptors, creating a multisensory experience influenced by texture, consistency, and emotional response.
The challenge
Lucideon's client wanted to link objective analytical product performance data with the data derived from sensory panel evaluation aiming to reduce the number of formulations requiring sensory panel assessment. Specifically, this entailed matching the physical characteristics of the product's foaming characteristics with the data obtained when those same products were evaluated by a trained sensory panel. Lucideon was tasked with developing a data-led model that could reduce the time and resource constraint on the sensory panels involved in product development, thereby reducing the time and cost required to launch new products.
What we delivered
Lucideon collaborated with the client to implement a machine-learning model that could predict sensory panel results based on analytical data. The project involved collecting extensive data from sensory panels evaluating various oral healthcare products. In parallel, characteristic foam data was collected in-house for the same set of products evaluated by the sensory panels.
Using this comprehensive dataset, a machine learning model was developed to identify patterns and correlations between the analytical data and the sensory panel results. The model was rigorously validated against additional sensory panel evaluations to ensure accuracy and reliability. Through iterative refinements, the model's predictive capabilities were continuously improved. Ultimately, Lucideon developed a robust tool that could reliably forecast sensory outcomes from objective data, significantly enhancing the efficiency of the product development process.
Value to the client
Lucideon's expertise and holistic approach supported the client in screening products for sensory panel assessment allowing for better informed decisions during the formulation development stages. The machine learning model could potentially reduce the reliance on resource-intensive sensory panels which can save time and costs associated with product development. Lucideon's expertise in machine learning and data analytics, combined with deep domain knowledge of product formulations, enabled the successful delivery of a data-led model that enhanced and sped up the client's product development process.