The Materials Innovation Factory (MIF) at the University of Liverpool is a £100M world-leading centre for accelerating materials discovery via the seamless integration of computation, chemistry automation, data analysis, and world-class expertise. A strategic partnership between the University of Liverpool (UoL), Unilever R&D, Research England, and EPSRC’s Royce Institute, the 11,500 m² facility houses over £20M of open access scientific equipment, a 20 FTE technical support team, over 100 industry, and 250 academic researchers. We will discuss accelerating product development in the MIF, and where we see the future of R&D programs such as Prof. Cooper’s 2020 work on the mobile robotic chemist, a robotic platform that performed 688 autonomous experiments over eight days.

Advances in AI offer powerful opportunities to accelerate cosmetic innovation, yet model performance is limited by sparse datasets, complex formulation effects, and the difficulty of representing ingredient functionality. Physics-based molecular simulations help overcome these challenges by providing mechanistic, high-resolution descriptors that strengthen AI model accuracy a. Building on recent work in hair biophysics, skin–formulation interactions, antioxidant chemistry, packaging migration, and formulation-aware ML, we show how molecular dynamics, coarse-grained models, quantum calculations, and free-energy methods can be integrated into data-driven pipelines. This combined Physics+AI framework delivers rapid, reliable functional ingredient characterization and accelerates the design of high-performance, sustainable cosmetic products.

The concept of “collagen banking” represents an emerging strategy in cosmetic science, focusing on the long-term preservation of dermal structure rather than solely stimulating collagen synthesis. Skin aging is increasingly understood as a multifactorial process driven by the interplay between inflammatory signaling, extracellular matrix (ECM) degradation, and impaired regenerative capacity. Central to this process is the 'inflammation-matrix axis,' involving interleukin-1 alpha (IL-1α), matrix metalloproteinase-1 (MMP-1), and Type I procollagen.

This study evaluates a novel active system combining plant-derived exosome-like vesicles from Centella asiatica callus culture with hydrolyzed Ganoderma lucidum protein-derived peptides. The system is designed to modulate intercellular communication and influence gene-level pathways associated with collagen preservation and matrix integrity.

In vitro investigations were conducted using human keratinocyte cultures subjected to lipopolysaccharide (LPS)-induced inflammatory stress. Biomarker analysis via immunofluorescence demonstrated a significant reduction in IL-1α expression (up to 39%), indicating effective upstream suppression of inflammatory signaling. Concurrently, the active system reduced collagen degradation, with an observed 11% decrease in MMP-1-associated matrix breakdown. Additionally, a marked increase in Type I procollagen synthesis was observed, confirming stimulation of matrix regeneration pathways.

Clinical evaluation at a 2% use level further supported these findings. After 28 days of topical application, subjects exhibited a statistically significant increase in dermal density (mean +16.4%, with 88% of participants showing improvement), alongside measurable improvements in skin firmness and elasticity. Wrinkle analysis revealed a 14.2% reduction in wrinkle depth and expression lines.

These results demonstrate a comprehensive, multi-pathway approach to skin aging, supporting collagen preservation, protection, and regeneration, and reinforcing the potential of “collagen banking” as a framework for next-generation cosmetic innovation.

Objective:
Sebum, a complex mixture of lipids, plays a key role in protecting and lubricating the skin and hair. While sebum acts as a natural barrier against microbes and pollutants, its build-up or contamination can contribute to conditions like acne and dandruff. The aim of this work was to establish an understanding and develop new methods and knowledge about physicochemical properties of artificial sebum (AS) that underpin the cleansing technologies.

Methods:
A novel micromanipulation method was developed and optimised for mechanical characterisation of AS thin films through scratch testing. The effect of temperature and humidity were evaluated using a combination of Differential Scanning Calorimetry (DSC) and Dynamic Vapour Sorption (DVS). Polarised Optical Microscopy (POM) allowed for evaluating polymorphism of AS.

Results:
AS upon ageing reduced the cohesive energy from 0.3 mJ on the day of preparation to 0.04 mJ after 72 hours, which is due to the kinetic transformation of more unstable β’ to β polymorph. DSC and POM analysis showed a change in crystallinity of AS upon ageing due to the change into the thermodynamically stable polymorphs. The addition of model air pollutants to AS resulted in lower mechanical strength and change in the polymorphism of AS, which made AS more solid-like at skin temperature therefore making it less likely to melt and remove. DVS analysis of polluted sebum after 24 h showed reduction in water sorption upon addition of pollutants suggesting that air pollutants affect water sorption and therefore moisturisation properties of sebum.

Conclusion:
This work offers an in-depth understanding of the physicochemical characteristics of AS and its interactions with model air pollutants. It was established that the physicochemical properties of AS are dependent on the crystal structure of lipids, which also get affected by the addition of model air pollutants.

Objective:

Hyperpigmentation requires a holistic approach targeting melanocytes and their microenvironment. This study identifies a solution that treats dark spots by directly targeting not only melanin but also lipofuscin accumulation, while mitigating pro-pigmenting molecules released by surrounding skin cells and sensory neurons.

Methods:

A specific Zingiber officinale root extract (ZOE) was evaluated in vitro on skin cells and monocytes (for RGMB release). Subsequently, an 84-day double-blind clinical study was conducted. Thirty Asian volunteers with dark spots applied 0.5% ZOE twice daily on a randomized half-face versus placebo. Cross-polarized imaging (ColorFace® and UVCam®) was used to measure dark spot density, skin color contrast, yellowness, luminosity, and lightening at D0, D56, and D84.

Results:

In vitro, ZOE effectively targeted both source pigments, decreasing melanin in melanocytes (-46%) and lipofuscin in fibroblasts (-26%). Furthermore, ZOE reduced the RGMB release (-12%) by monocytes, which led to a secondary reduction of melanin production (-22%). In vivo, clinical imaging confirmed remarkable corrections. At day 56, 0.5% ZOE significantly reduced dark spot density (-181% vs. placebo) and skin color contrast (-142%). By day 84, it reduced skin yellowness (-29%) while drastically increasing luminosity (+100%) and skin lightening (+85%).

Conclusion:

The applied methodology identified a specific ginger extract that directly reduces both melanin and lipofuscin, two pigment sources of dark spots. The study validates that a comprehensive intercellular approach successfully translates into strong clinical efficacy, significantly correcting dark spots and boosting overall skin radiance. To complete this holistic strategy in formulation, integrating an upcycled Persea gratissima oil (PGO) delivers complementary biological and sensory benefits, infused with a strong touch of naturality.

Objective:
Methods such as differential scanning calorimetry (DSC) and complementary structural assays are widely used to evaluate protein and peptide-based haircare actives. Still, they are applied late in new ingredient development and require significant resources. A predictive, high-throughput strategy for early-stage screening of candidate materials for selective hair keratin binding remains needed. Additionally, identifying the specific keratin regions targeted by these actives can strengthen mechanistic understanding and guide rational product design.

Methods:
We present a high-throughput screening platform for assessing selective binding of proteins, peptides, and hydrolysates to human hair keratins. Fluorescently labelled materials are incubated in 96-well plates coated with defined keratin peptide fragments representing key structural domains. After controlled wash steps, designed to mimic pre-damaged, post-damaged or healthy hair, retained fluorescence is quantified using a plate reader to determine relative binding affinity and selectivity. Differential wash conditions are applied to explore potential interaction mechanisms, including covalent and electrostatic contributions.

Results:
Distinct binding profiles were observed across the tested materials, including selective affinity for regions of Keratin 32 and Keratin 82, which are abundant in the hair cuticle. Wash-dependent differences suggested multiple interaction modes, including disulphide-mediated and charge-driven associations.

Conclusion:
This scalable, cost-effective platform enables rapid, function-led prioritisation of protein and peptide actives, supporting precision-driven haircare development and accelerating innovation from discovery to formulation.

Understanding the cellular mechanisms underlying variation in human hair fibre traits remains a key challenge in cosmetic science and trichology. Large-scale genetic and sequencing studies have identified numerous loci associated with traits such as hair pigmentation, fibre shape, and thickness; however, these approaches do not directly reveal which hair follicle cell types mediate these effects. To address this, we developed Hair Score, a computational pipeline that integrates diverse genetic dataset such as human genome-wide association studies (GWAS) and other relevant sequencing data from other species with transcriptomic profiles of hair follicle cell populations.

The Hair Score pipeline maps trait-associated genes onto transcriptomic datasets representing distinct follicular cell types. Gene expression within each population is quantified and aggregated to generate a cumulative score reflecting the relative contribution of each cell type to the biology of a given hair fibre trait. This enables systematic identification of follicular compartments most likely to influence fibre properties.

Application of Hair Score to datasets relating to hair pigmentation, strand shape, and thickness revealed distinct cell-type associations across traits. For pigmentation, melanocytes emerged as the highest-scoring contributors when using cheetah RNAseq data, consistent with their central role in pigment production. In addition, signals from adjacent follicular cell populations were observed, highlighting intercellular communication within the follicle. EDN3 was prioritised as a secreted protein, supporting a role for signalling between follicular cells that regulate melanocyte activity and ultimately hair colour.

For hair shape, the inner root sheath was the strongest-associated cell population, consistent with its role in guiding the emerging fibre.

Together, these findings demonstrate that integrating genetic and transcriptomic data can reveal the cellular context underlying human hair fibre traits. The Hair Score provides a scalable framework for linking genetic discoveries to specific follicular cell types and signalling pathways, the identification of novel targets for cosmetic science.

Objective:
Research on textured hair is accelerating as the understanding of curl behaviour and targeted care needs increasingly come into focus. Tensile testing of textured hair provides valuable information about elasticity, tensile strength and extensibility. However, as curl levels increase, so does the inter-fibre variability, complicating the analysis and the interpretation of influences of curl type on fibre behaviour. In particular, the ‘toe in’ effect makes it difficult to consistently determine tensile parameters across different curl patterns, as the elastic region deviates from a straight line. Given the importance of the modulus of elasticity (E Modulus), we investigate whether the first derivative of the stress-strain curve is an alternative and consistent method for modulus determination.

Methods:
Human tresses were classified by professional stylists using the Walker Scale (Types 1-4C). Hair fibres were taken from the tresses and measured for curl count and stretched length. The curl values for fibres provide the foundation for a detailed comparison across curl categories. This metric is the basis to investigate how fibre curl patterns impact tensile properties. All tensile tests were carried out in wet conditions.

Results:
Using the 1st derivative approach, the influence of fibre ‘toe-in’, machine bias and human error on the modulus was minimised, improving the accuracy of measurements at high curl levels. In particular, it could be confirmed and quantified that as curl value increases, the E modulus drops.

Conclusion:
This study examines critical challenges in textured-hair research and introduces methodologies that substantially enhance the precision and interpretability of tensile testing for curly hair. These advancements streamline scientific communication and offer valuable implications for both the cosmetic industry and for salon professionals.