Predictive machine learning model to provide a sun protection factor value (SPF) associated to an ultraviolet (UV) filter composition

Abstract

A computer-implemented method (100) for providing at least one predicted Sun Protection Factor (SPF) value associated to an Ultraviolet (UV) filter composition comprising: - a step of providing (135) an original set of exemplar UV filter composition digital identifiers, said set of exemplar UV filter composition, comprising for at least one UV filter composition digital identifiers: - at least one existing ingredient or at least one combination of existing ingredients, - at least one calculated UV spectrum, and - a value representative of an empirically measured SPF value, to form a training set, - a step of training (140) a predictive machine-learning model using the training set, to predict at least one SPF value, - a step of defining (145) at least one UV filter composition digital identifier, - a step of predicting (150) at least one SPF value using the trained predictive machine-learning model and - a step of providing (155) the predicted SPF value.

Description

[0001]DSM IP Assets B.V. 34717-WO-PCTPREDICTIVE MACHINE LEARNING MODEL TO PROVIDE A SUN PROTECTION FACTOR VALUE (SPF) ASSOCIATED TO AN ULTRAVIOLET (UV) FILTER COMPOSITION TECHNICAL FIELD OF THE INVENTION The present invention relates to a computer-implemented method for providing at least one predicted Sun Protection Factor (SPF) value associated to an Ultraviolet (UV) filter composition digital identifier, representing a materializable UV filter composition, associated with at least one ingredient digital identifier representative of an existing ingredient of the materializable UV filter composition, a computer-implemented method to train a predictive machine learning model to provide at least one Sun Protection Factor value associated to an Ultraviolet (UV) filter composition digital identifier, representing a materializable UV filter composition and a corresponding computer program product, computer-readable storage medium and device. The present invention is applicable to the domain of UV filter products, for example in cosmetics, skincare or sunscreening products. BACKGROUND OF THE INVENTION Sun care application development is expensive and time-consuming because it is mandatory to evaluate the Sun Protection Factor (SPF) in vivo. Today sunscreen developers have several computational tools available, which can predict the sun protection performance of an Ultraviolet (UV) filter composition to some extent. These tools allow developers to quickly pre-evaluate and compare the performance of different UV filter compositions in silico before selecting the most promising candidates for the in vivo SPF study. This saves costs and time during development and helps to find optimal solutions. However, significantly higher cost and time would be saved if these tools allowed determining the SPF rather than making a pre-evaluation. Thus, sufficiently accurate SPF prediction methods allow sunscreen developers to focus their research on other important properties such as sensory, eco-friendliness, or costs. Therefore, the use of such methods will ultimately lead to improved sunscreen products for the consumer, i.e., any product with an SPF claim. Unfortunately, the underlying prediction methods are still not sufficiently accurate to replace the in vivo study because they take only the UV filter composition into account. However, it is known that the distribution of UV filters on the skin and hence the SPF can strongly be influenced by composition ingredients other than UV filters such as emollients, thickeners, emulsifiers, film formers, etc. For instance, it is clear that a highly viscous sunscreen can be less evenly distributed on the skin in comparison to a low viscosity product and therefore leads to a lower SPF even if both products have an identical UV filter composition. Jiyong S. et al. Machine learning for the prediction of sunscreen sun protection factor and protection grade of UVA, Experimental Dermatology, 2019, Volume 28, Issue 7, Pages: 872-874 and WO2022157011A1 both try to overcome this issue by using machine learning methods, which consider, next to the concentration of UV filter substances various additional factors, which may influence the SPF or UVA Protection Factor. The method described in Jiyong S. et al. uses next to concentration of UV filter substances four additional features, such as the presence of pigment, the amount of pigment grade titanium dioxide, the type of formulation and type of product. The method described in WO2022157011A1 distinguishes between UVB and UVA filter and adds seven additional factors, which gives the following feature set: -the viscosity of the composition,- a high polarity emollient, a medium polarity emollient, and a low polarityemollient, -a UVA filter, a UVB filter, and the ratio of UVB vs UVA filters,- the ratio of UV filters in the oil phase vs UV filters in the water phase,- the ratio of absorbing type UV filter vs scattering / reflecting type UVfilters. Both methods use very few manually selected features from the entire possible feature set of UV filter compositions. Manual selection will almost always result in a lower predictive accuracy than automatic selection. Both methods also ignore a possible influence of many other features such as the spreading property or chemical nature of the emollients, the type and amount of the emulsifiers, the presence, type, and amount of film formers etc. Allof the features may have a significant impact on the distribution of UV filters on theskin and hence on the SPF. Consequently, both methods are not sufficiently accurate to replace an in vivo SPF study of the materialized composition. SUMMARY OF THE INVENTION The present invention is intended to remedy all or part of these disadvantages. To this effect, according to a first aspect, the present invention aims at a computer-implemented method for providing at least one predicted Sun Protection Factor (SPF) value associated to an Ultraviolet (UV) filter composition digital identifier, representing a materializable UV filter composition, associated with at least one ingredient digital identifier representative of an existing ingredient of the materializable UV filter composition, comprising: -a step of providing an original set of exemplar UV filter composition digitalidentifiers, said set of exemplar UV filter composition digital identifiers being representative of materialized UV filter compositions, comprising: -at least one value representative of an existing ingredient or at leastone value representative of a combination of existing ingredients, each existing ingredient being associated with an ingredient digital identifier,- at least one value representative of at least one calculated UVspectrum associated with said UV filter composition digital identifiers, and -at least one value representative of an empirically measured SPFvalue associated with said UV filter composition digital identifiers, to form a training set, -a step of training a predictive machine-learning model using the training set,wherein the predictive machine-learning model is trained to predict at least one SPF value, -a step of defining at least one UV filter composition digital identifier,- a step of predicting at least one SPF value using the trained predictivemachine-learning model and -a step of providing the at least one predicted SPF value.Thanks to these provisions, the predicted SPF value is more accurate. The spectra of the UV filter composition and of the individual UV filters of a sunscreen product is by far the most important factor influencing the SPF. Prediction accuracy is dramatically increased by explicitly including information of the spectra of the UV filter composition in the training set. Furthermore, encoding UV filters absorption capacity into UV spectra, and ingredient characteristics into physico-chemical properties, enables the definition of fully informative dense input features, that may not include lists of ingredients. This improves model generalization performance to new sunscreen compositions, even when these compositions contain ingredients that can not be present in the training set. Also, the great flexibility of the method subject of the present invention makes it possible, as the amount of data increases, to achieve accuracy sufficient to replace the in vivo SPF method. In particular embodiments, in the training set, at least one value representative of an existing ingredient is an ingredient quantity. Thanks to these provisions, the UV filter compositions are more precisely defined in the training set. In particular embodiments, in the training set, at least one value representative of an existing ingredient is a value representative of a function of the existing ingredient. In particular embodiments, in the training set, at least one value representative of an existing ingredient or of a combination of existing ingredients is a value representative of a physico-chemical property of the existing ingredient or of the existing combination of ingredients. In particular embodiments, the training set comprises at least one value representative of the proportion of erythemal energy absorbed in oil phase associated with at least one composition digital identifier and / or at least one value representative of the total amount of UV filters in the composition associated with at least one composition digital identifier. Thanks to these provisions, the accuracy of the predicting machine learning model is even more pronounced. In particular embodiments, the computer-implement method subject of the present invention further comprises: -a step of assembling a UV filter composition with at least one ingredient,- a step of calculating at least one UV spectrum of the assembled UV filtercomposition or at least one UV spectrum of a combination of one or more ingredients, -a step of empirically measuring an SPF value of the assembled UV filtercomposition, and -a step of storing, in a database, at least one empirically measured SPFvalue and at least one calculated UV Spectrum in association with at least one UV filter composition digital identifier representative of the assembled UV filter composition, wherein said database is used during the step of providing a training set. Thanks to these provisions, the database constituting the training set is based on real-life samples and measured data from a tangible reality. In particular embodiments, the method further comprises a step of pre- processing the training set, and wherein the pre-processed training set is used during the step of providing a training set. Thanks to these provisions, the number of values of the training set is reduced so as to limit the risk of overfitting. All ingredients can be replaced by their physico-chemical properties and / or ratio within a functional group to generalize the method subject of the invention for SPF value prediction of any given data set. Also, a sparse ingredient amount matrix can be replaced by a dense ingredient property matrix allowing more efficient training of machine learning tools and better generalization performance. In particular embodiments, the trained predictive machine-learning model is a Random Forest Regressor (RF). Thanks to these provisions, predictions of a set of decision trees are combined providing an enhanced performance. In particular embodiments, the trained predictive machine-learning model is a Gradient Boosting Regressor (GB, XGB). Thanks to these provisions, large datasets can be used efficiently with a high accuracy. Ensemble methods exhibit several useful properties for the implementation of the method subject of the invention such as: -achieving higher performance than single machine-learning models,especially in high-dimensional, noisy problems. -being less prone to overfitting, especially when the number of relevantvariables is small compared to the total number of variables. -offering natural metrics to determine feature importance to the predictions.In particular embodiments, the trained predictive machine-learning model is a Neural Network (NN), such as a feed-forward neural network. Thanks to these provisions, the predictive machine learning is particularly scalable. In particular embodiments, the computer-implemented method subject of the invention further comprises a step of sending a digital command representative of an instruction of materialising at least one UV filter composition corresponding to at least one UV filter composition digital identifier provided. Thanks to the provisions, the computer-implemented method can be linked to a device implementing the digital commands to materialize the UV filter composition. In particular embodiments, the computer-implemented method subject ofthe invention further comprises a step of materialising at least one UV filtercomposition corresponding to at least one UV filter composition digital identifier provided. Thanks to these provisions, the UV filter composition is materialized. According to a second aspect, the invention aims at a computer- implemented method to train a predictive machine learning model to provide at least one value associated to an Ultraviolet (UV) filter composition digital identifier, representing a materializable UV filter composition, comprising: -a step of providing an original set of exemplar UV filter composition digitalidentifiers, said set of exemplar UV filter composition digital identifiers being representative of materialized UV filter compositions, comprising: -at least one value representative of an existing ingredient or at leastone value representative of a combination of existing ingredients, each existing ingredient being associated with an ingredient digital identifier, -at least one value representative of at least one calculated UVspectrum associated with said UV filter composition digital identifiers, and -at least one value representative of an empirically measured SPFvalue associated with said UV filter composition digital identifiers, to form a training set and -a step of training a predictive machine-learning model using the training set,wherein the predictive machine-learning model is trained to predict at least one SPF value. According to a third aspect, the invention aims at a computer programproduct comprising instructions which upon execution by a computer cause thecomputer to execute the method subject of the invention. According to a fourth aspect, the invention aims at a computer-readable storage medium storing programming instructions which upon execution by a computer cause the computer to execute the method subject of the invention. According to a fifth aspect, the invention aims at a device for providing at least one predicted Sun Protection Factor (SPF) value associated to an Ultraviolet (UV) filter composition digital identifier, representing a materializable UV filtercomposition, associated with at least one ingredient digital identifier representativeof an existing ingredient of the materializable UV filter composition, comprising: -a means of providing an original set of exemplar UV filter composition digitalidentifiers, said set of exemplar UV filter composition digital identifiers being representative of materialized UV filter compositions, comprising: -at least one value representative of an existing ingredient or at leastone value representative of a combination of existing ingredients, each existing ingredient being associated with an ingredient digital identifier, -at least one value representative of at least one calculated UVspectrum associated with said UV filter composition digital identifiers, and -at least one value representative of an empirically measured SPFvalue associated with said UV filter composition digital identifiers,to form a training set and -a means of training a predictive machine-learning model using the trainingset, wherein the predictive machine-learning model is trained to predict at least one SPF value, -a means of defining at least one UV filter composition digital identifier,- a means of predicting at least one SPF value using the trained predictivemachine-learning model and -a means of providing the at least one predicted SPF value. The second to fifth aspects of the present invention exhibit the same advantages as the related first aspect. BRIEF DESCRIPTION OF THE DRAWINGS Other advantages, purposes and particular characteristics of the invention shall be apparent from the following non-exhaustive description of at least one particular embodiment of the present invention, in relation to the drawings annexed hereto, in which: [Figure 1] represents, schematically, a first succession of steps of aparticular embodiment of the method subject of the present invention, [Figure 2] represents, schematically, a computer system with which anembodiment of the method subject of the present invention can be implemented, [Figure 3] represents, schematically, the correlation between a predictedSPF value and a calculated SPF value, the machine-learning model XGB implemented in the method subject of the invention being trained and evaluated with a first dataset labelled “S1”, [Figure 4] represents, schematically, the correlation between a predictedSPF value and a calculated SPF value, the machine-learning model implemented in the method subject of the invention being trained and evaluated with a second dataset labelled “S2”, [Figure 5] represents, schematically, a first succession of steps of a particular embodiment of pre-processing of the training set of the method subject of the present invention, and [Figure 6] represents a list of features classified by their importance in accurately predicting the SPF value of an ingredient composition. DETAILED DESCRIPTION OF THE INVENTION This description is not exhaustive, as each feature of one embodiment may be combined with any other feature of any other embodiment in an advantageousmanner. Also, various inventive concepts may be embodied as one or moremethods, of which an example has been provided. The acts performed as part of the method may be ordered in any suitable way. Accordingly, embodiments may be constructed in which acts are performed in an order different than illustrated, which may include performing some acts simultaneously, even though shown as sequential acts in illustrative embodiments. The indefinite articles ‘a’ and ‘an’, as used herein in the specification and in the claims, unless clearly indicated to the contrary, should be understood to mean ‘at least one’. The phrase ‘and / or’, as used herein in the specification and in the claims, should be understood to mean ‘either or both’ of the elements so conjoined, i.e., elements that are conjunctively present in some cases and disjunctively present in other cases. Multiple elements listed with ‘and / or’ should be construed in the same fashion, i.e. ‘one or more’ of the elements so conjoined. Other elements may optionally be present other than the elements specifically identified by the ‘and / or’ clause whether related or unrelated to those elements specifically identified. Thus, as a non-limiting example, a reference to ‘A and / or B’, when used in conjunction with open-ended language such as ‘comprising’ can refer, in one embodiment, to A only (optionally including elements other than B); in another embodiment, to B only (optionally including elements other than A); in yet another embodiment, to both A and B (optionally including other elements); etc. As used herein in the specification and in the claims, ‘or’ should be understood to have the same meaning as ‘and / or’ as defined above. For example, when separating items in a list, ‘or’ or ‘and / or’ shall be interpreted as being inclusive, i.e., the inclusion of at least one, but also including more than one, of a number or list of elements, and, optionally, additional unlisted items. Only terms clearly indicated to the contrary, such as ‘only one of’ or ‘exactly one of’, or, when used in the claims, ‘consisting of’, will refer to the inclusion of exactly one element of a number or list of elements. In general, the term ‘or’ as used herein shall only be interpreted as indicating exclusive alternatives (i.e. ‘one or the other but not both’) when preceded by terms of exclusivity, such as ‘either,’ ‘one of,’ ‘only one of’, or ‘exactly one of’. ‘Consisting essentially of,’ when used in the claims, shall have its ordinary meaning as used in the field of patent law. As used herein in the specification and in the claims, the phrase ‘at least one’, in reference to a list of one or more elements, should be understood to mean at least one element selected from any one or more of the elements in the list of elements, but not necessarily including at least one of each and every element specifically listed within the list of elements and not excluding any combinations of elements in the list of elements. This definition also allows that elements may optionally be present other than the elements specifically identified within the list of elements to which the phrase ‘at least one’ refers, whether related or unrelated to those elements specifically identified. Thus, as a non-limiting example, ‘at least one of A and B’ (or, equivalently, ‘at least one of A or B’, or, equivalently ‘at least one of A and / or B’) can refer, in one embodiment, to at least one, optionally including more than one, A, with no B present (and optionally including elements other than B); in another embodiment, to at least one, optionally including more than one, B, with no A present (and optionally including elements other than A); in yet another embodiment, to at least one, optionally including more than one, A, and at least one, optionally including more than one, B (and optionally including other elements); etc. In the claims, as well as in the specification above, all transitional phrases such as ‘comprising,’ ‘including,’ ‘carrying,’ ‘having,’ ‘containing,’ ‘involving,’ ‘holding,’ ‘composed of’, and the like are to be understood to be open-ended, i.e., to mean including but not limited to. Only the transitional phrases ‘consisting of’ and ‘consisting essentially of’ shall be closed or semi-closed transitional phrases, respectively. It should be noted at this point that the figures are not to scale. In the context of the present invention a “UV filter Composition” - commonlyalso referred to as “Sunscreen” or “SPF product” - refers to a compositioncomprising at least one ingredient absorbing UV-light, i.e. an ingredient which absorbs light in the range from 280 to 400 nm. Such ingredients are commonly referred to as “UV-filters” and are often classified as UV-B filters (i.e. ingredients having a lambda max in the range from 280-320nm), UV-A filters (i.e. ingredients having a lambda max in the range from 320-400nm) or broadband filters (ingredients absorbing UV-light in the UV-B and UV-A range). Later on, features listed with a format of “E” concatenated with a numberrefer to components of the UV spectrum associated with the wavelengthcorresponding to the number. Such a feature refers to the fact that an ingredients blocks said wavelength. Later on, features listed with a number only refer to specific ingredients which a person skilled in the art would find with a feature importance determination algorithm, such as disclosed below. Said UV filters used in skin- respectively sun-care products help to preventsunburn, premature skin aging, and / or skin cancer by absorbing, scattering, or reflecting UV radiation. The UV-filters can be either physical (inorganic) or chemical (organic) in nature, or a combination of both, and they may work by different mechanisms to block or mitigate the penetration of UV rays into the skin. Next to one or more UV-filters, the UV-filter Composition generally also comprises a cosmetically acceptable carrier, which is generally composed of usual cosmetic adjuvants and additives, such as preservatives / antioxidants, fattysubstances / oils, water, organic solvents, silicones, thickeners, softeners,emulsifiers, antifoaming agents, moisturizers, aesthetic components such as perfuming ingredients (fragrances), surfactants, fillers, sequestering agents, anionic, cationic, nonionic or amphoteric polymers or mixtures thereof, propellants, acidifying or basifying agents, dyes, colorings / colorants, abrasives, absorbents, essential oils, skin sensates, astringents, antifoaming agents, pigments or nanopigments, or any other ingredients usually formulated into UV filter Compositions. Ingredients (including UV-filters) suitable for use in the UV filter Compositions of the present invention are for example described in the International Cosmetic Ingredient Dictionary & Handbook by Personal Care Product Council (http: / / www.personalcarecouncil.org / ), accessible by the online INFO BASE (http: / / online.personalcarecouncil.org / jsp / Home.jsp), without being limited thereto. Thus, in the context of the present invention, an ingredient refers to any ingredient part of a UV filter composition including but not limited to UV- filters. Particularly preferred UV-filters used in UV filter Compositions are: -Physical (Inorganic) UV Filters, such as:- Zinc oxide: a mineral UV filter that provides broad-spectrum protectionagainst both UVA and UVB rays which sits on the skin's surface and reflects UV radiation; Often such Zinc oxide is surface coated such as with triethoxycaprilylsilane. A particularly suited surface coated zinc oxide is e.g. commercially available as PARSOL® ZX at dsm-firmenich. -Titanium dioxide: another mineral UV filter that works similarly to zincoxide by reflecting UV radiation and is effective against both UVA and UVB rays. Such Titanium dioxides are normally surface coated such as with one or more of silica, dimethicone and triethoxycaprylylsilane. Aparticularly suited surface coated titanium dioxide is e.g. commercially available as PARSOL® TX at dsm-firmenich. -Chemical (Organic) UV Filters such as:- Avobenzone: a commonly used organic UV filter that absorbs UVA rayse.g. commercially available as PARSOL® 1789 at dsm-firmenich. -Diethylamino hydroxybenzoyl hexyl benzoate: a commonly usedorganic UV filter that absorbs UVA rays, e.g. commercially available as PARSOL® DHHB at dsm-firmenich. -Cinnamates such as Octyl methoxycinnamate e.g. commerciallyavailable as PARSOL® MCX at dsm-firmenich and Isoamyl p- Methoxycinnamate, both of which absorb UVB rays. -Octocrylene: which absorbs UVB rays and stabilizes other UV filterssuch as in particular avobenzone, and which is e.g. commercially available as PARSOL® 340 at dsm-firmenich. -Oxybenzone: which absorbs both UVA and UVB rays.- Triazin derivatives such as bis-ethylhexyloxyphenol methoxyphenyltriazine which absorbs UVA and UVB rays (broadband filter), e.g. commercially available as PARSOL® SHIELD at dsm-firmenich, Ethylhexyl Triazone which absorbs UVB rays, e.g. commercially available as PARSOL® EHS at dsm-firmenich, as well as Tris-Biphenyl Triazine and Diethylhexylbutamidotriazon (DBT), both of which absorb UVB rays. -Benzimidazol derivatives such as Phenyl benzimidazole sulfonic acidwhich absorbs UVB rays and is e.g. commercially available as PARSOL® HS at dsm-firmenich as well as Disodium Phenyl Dibenzimidazole Tetrasulfonate. -Salicylates such as Homosalate and Octyl salicylate, both of whichabsorb UVB rays and are e.g. commercially available as PARSOL® HMS respectively PARSOL® EHS at dsm-firmenich. -Methylene Bis-Benzotriazolyl Tetramethylbutylphenol, an organicpigment which absorbs UVA and UVB rays (broadband filter), e.g. commercially available as PARSOL® MAX. -Camphor derivatives such as 4-Methylbenzylidene Camphor which ise.g. commercially available as PARSOL® 5000 at dsm-firmenich, 3-Benzylidene Camphor and Terephtalylidene Dicamphor Sulfonic Acid. -Drometrizole Trisiloxane, which is a broad-spectrum UV absorber withtwo absorption peaks, one at 303 nm (UVB) and one at 344 nm (UVA). -Polysilicone-15, which is a UV-B filter and which is commerciallyavailable as PARSOL® SLX at dsm-firmenich. -Methoxypropylamino Cyclohexenylidene Ethoxyethylcyanoacetate,which absorbs UV-A light, particularly in the range of 360 to 400nm, and -Menthyl anthranilate.Exemplary adjuvants and additives commonly used in UV filter compositions encompass: -Antioxidants: often included in UV filter compositions to help neutralize freeradicals generated by UV radiation and provide additional protection against sun damage, such as Vitamin E or Vitamin C as well as derivatives thereof.- Stabilizers: added to UV filter compositions to enhance the stability andeffectiveness of UV filters over time, e.g. by reducing and / or inhibiting chemical degradation such as Diethylhexyl 2,6-Naphthalate or Diethylhexyl syringylidene malonate.- Moisturizing agents (humectants): compounds or mixtures of compoundswhich give UV filter compositions the quality, after application to or distribution on the skin surface, of reducing the loss of moisture of the stratum corneum (Experimental determination by e.g transepidermal water loss (TEWL)) and / or of positively influencing the hydration of the stratum corneum and / or keeping an actual hydration state such as for example polyols (e.g. glycerol or glycols), lactic acid and / or lactates, especially sodium lactate, biosaccharide gum-1, glycine soya, ethylhexyloxyglycerol, pyrrolidonecarboxylic acid, urea, hyaluronic acid or derivatives thereof, chitosan and / or a fucose-rich polysaccharides, as well as sugars and sugar derivatives such as e.g. saccharide isomerate.- Preservatives / preservative boosters: agents suitable to improve shelf life ofUV filter compositions such as e.g. ethanol, phenoxyethanol, 1,2- pentanediol, 1,2-hexanediol, 1,2-octanediol (caprylyl glycol), 1,2 decanediol, 2-methyl-1,3-propanediol, propanediol, propylene glycol and p- hydroxyacetophenone.- Perfuming ingredients: a compound or compound mixture, which is used ina composition to impart a hedonic effect, in other words, such an ingredient, to be considered as being a perfuming one, must be recognized by a person skilled in the art as being able to impart or modify in a positive or pleasantway the odor of a composition, and not just as having an odor such as e.g. limonene, citral, linalool, alpha-isomethylionone, geraniol, citronellol, 2- isobutyl-4-hydroxy-4-methyltetrahydropyran, 2-tert-pentylcyclohexyl acetate, 3-methyl-5-phenyl-1-pentanol, 7-acetyl-1,1,3,4,4,6- hexamethyltetralin, adipic diester, cinnamal, amyl salicylate, alpha- amylcinnamaldehyde, alpha-methylionone, butylphenylmethylpropional, cinnamal, amylcinnamyl alcohol, anise alcohol, benzoin, benzyl alcohol, benzyl benzoate, benzyl cinnamate, benzyl salicylate, bergamot oil, bitter orange oil, butylphenylmethylpropional, cardamom oil, cedrol, cinnamal, cinnamyl alcohol, citronellyl methylcrotonate, citrus oil, coumarin, diethyl succinate, ethyllinalool, eugenol, Evernia furfuracea extract, Evernia prunastri extract, farnesol, guaiacwood oil, hexylcinnamal, hexyl salicylate, hydroxycitronellal, lavender oil, lemon oil, linalyl acetate, mandarin oil, menthyl PCA, methyl hexadecan, nutmeg oil, rosemary oil, sweet orange oil, terpineol, tonkabean oil, triethyl citrate and vanillin.- Emollients: oils, which are in particular used to solubilize solid UV-filters inUV filter compositions such as dibutyl adipate, diisopropyl sebacates, dibutyl sebacate, isopropyl palmitate, isopropyl myristate, caprylyl capric triglyceride, cetearyl isononanoate, cocoglyceride, isononyl isononanoate, propylene glycol dicaprylate / dicaprate, phenethyl benzoate, C12-C15 alkyl benzoate, C16-17 alkyl benzoates, dicaprylyl carbonate, di-C12-13 alkyl tartrates, hydrogenated castor oil dimerates, triheptanoin, C12-13 alkyl lactates such as lauryl lactate, propylheptyl caprylates, caprylic / capric triglycerides, diethylhexyl 2,6-naphthalate, octyldodecanol, and ethylhexyl cocoatesdibutyl adipate as well as oils which soothe and soften the skin such as silicone (dimethicone, cyclomethicone), vegetable oils (grape seed, sesame seed, jojoba, etc.), butters (cocoa butter, shea butter), and petrolatum derivatives (petroleum jelly, mineral oil).- Emulsifiers / surfactants: compounds suitable for the preparation ofemulsions such as in particular O / W or W / O emulsions such as in particular phosphate ester surfactants (e.g. C8-10 Alkyl Ethyl Phosphate, C9-15 Alkyl Phosphate, Ceteareth-2 Phosphate, Ceteareth-5 Phosphate, Ceteth-8 Phosphate, Ceteth-10 Phosphate, Cetyl Phosphate, C6-10 Pareth-4 Phosphate, C12-15 Pareth-2 Phosphate, C12-15 Pareth-3 Phosphate, DEA-Ceteareth-2 Phosphate, DEA-Cetyl Phosphate, DEA-Oleth-3 Phosphate, Potassium cetyl phosphate, Deceth-4 Phosphate, Deceth-6 Phosphate and Trilaureth-4 Phosphate), PEG 30 Dipolyhydroxystearate, PEG-4 Dilaurate, PEG-8 Dioleate, PEG-40 Sorbitan Peroleate, PEG-7 Glyceryl Cocoate, PEG-20 Almond Glycerides, PEG-25 Hydrogenated Castor Oil, Glyceryl Stearate (and) PEG-100 Stearate , PEG-7 Olivate, PEG-8 Oleate, PEG-8 Laurate, PEG-60 Almond Glycerides, PEG-20 Methyl Glucose Sesquistearate, PEG-40 Stearate, PEG-100 Stearate, PEG-80 Sorbitan Laurate, Steareth-2, Steareth-12, Oleth-2, Ceteth-2, Laureth-4, Oleth-10, Oleth-10 / Polyoxyl 10 Oleyl Ether, Ceteth-10, Isosteareth-20, Ceteareth-20, Oleth-20, Steareth-20, Steareth-21, Ceteth-20, Isoceteth-20, Laureth-23, Steareth-100, glycerylstearatcitrate, glycerylstearate (self- emulsifying), stearic acid, salts of stearic acid, polyglyceryl-3- methylglycosedistearate, sorbitan oleate, sorbitan sesquioleate, sorbitan isostearate, sorbitan trioleate, Lauryl Glucoside, Decyl Glucoside, Sodium Stearoyl Glutamate, Sucrose Polystearate, Hydrated Polyisobuten, synthetic polymers (e.g. eicosene copolymer, acrylates / C10-30 alkyl acrylate crosspolymer, acrylates / steareth-20 methacrylate copolymer, PEG-22 / dodecyl glycol copolymer, PEG-45 / dodecyl glycol copolymer), polymeric emulsifiers (e.g. hydrophobically modified polyacrylic acid such as Acrylates / C10-30 Alkyl Acrylate Crosspolymers), polyalkylenglycolether (e.g. Brij 72 (Polyoxyethylen(2)stearylether) or Brij 721 (Polyoxyethylene (21) Stearyl Ether), Polyglyceryl-3 Methylglucose Distearate, Lauryl Glucoside (and )Polyglyceryl-2 Dipolyhydroxystearate, Glyceryl Sterate Citrate, Sodium Cetearyl Sulfate, Cetearyl Glucoside; Polyglyceryl-6 Stearate (and) Polyglyceryl-6 Behenate, Cetearyl Olivate (and) Sorbitan Olivate, Arachidyl Alcohol (and) Behenyl Alcohol (and) ArachidylGlucosides, Cetearyl Alcohol (and) Coco-Glucoside, Coco-Glucoside (and)Coconut Alcohol, PEG-100 Stearate (and) Glyceryl Stearate, Sodium Stearoyl Glutamate, Steareth-20, Steareth-21, Steareth-25, Steareth-2, Ceteareth-25 and Ceteareth-6. Polyglycerol esters or diesters of fatty acids (e.g. Polyglyceryl-4 Diisostearate / Polyhydroxystearate / Sebacate, PEG-30 dipolyhydroxystearate, polyglyceryl-3 diisostearate), polyglycerol esters of oleic / isostearic acid, polyglyceryl-6 hexaricinolate, polyglyceryl-4-oleate, magnesium stearate, sodium stearate, potassium laurate, potassium ricinoleate, sodium cocoate, sodium tallowate, potassium castorate, sodium oleate, Lauryl Polyglyceryl-3 Polydimethylsiloxyethyl Dimethicone and / or PEG-9 Polydimethylsiloxyethyl Dimethicone and / or Cetyl PEG / PPG-10 / 1 Dimethicone and / or PEG-12 Dimethicone Crosspolymer and / or PEG / PPG- 18 / 18 Dimethicone- Co-surfactants such as compounds selected from the group of mono- anddiglycerides and / or fatty alcohols such as behenyl alcohol, cetyl alcohol, cetearyl alcohol, stearyl alcohol and glyceryl stearate.- Thickeners, generally used to adjust the viscosity of the UV-filtercomposition such as xanthan gum, microcrystalline cellulose, cellulose, cellulose gum, crosslinked acrylate / C10-C30 alkyl acrylate polymer, hydroxyethyl acrylate / sodium acryloyldimethyl taurate copolymer, preferably xanthan gum and hydroxyethyl acrylate / sodium acryloyldimethyl taurate copolymer.- Alkanediols such as 1,3-propandiol, butylene glycol and 1,2-butandiol.- Fillers such as starch and starch derivatives (e.g. tapioca starch, distarchphosphate, aluminum or sodium starch octenylsuccinate), pigments which have neither primarily UV filter effect nor coloring effect (e.g. Valvance Touch 210 or 250 commercially available at dsm-firmenich), Aerosils®, talc, polyethylene, nylon, and silica dimethyl silylate.- Film formers such as selected from the group of the polymers based onpolyvinylpyrrolidone (PVP), Capryloyl Glycerin / Sebacic Acid Copolymer, Diisostearoyl Polyglyceryl-3 Dimer Dilinoleate and Ethyl Cellulose.- Skin active agents such as e.g. ingredients for skin lightening, for tanningprevention, for the treatment of hyperpigmentation, for preventing or reducing acne, wrinkles, lines, as well as atrophy and / or anti-inflammation, anti-cellulites and slimming (e.g. phytanic acid), firming, energizing, self- tanning, and soothing agents, as well as agents to improve elasticity and skin barrier.- Chelators and / or sequestrants. The UV filter composition may be formulated as fluid or creamy oil-in-water emulsion, etc. On application, the composition forms a thin film on the skin surface that affords UV protection. Other UV filter composition types may include, but are not limited to, water- in-oil emulsion, oil-in-water in oil emulsion, water in oil in water emulsion, water in silicone emulsion, silicone-in-water emulsion, polymeric gel cream, lipophilic monophase oil, lipophilic monophase gel, lipophilic monophase stick, lipophilic - alcoholic mixture, hydrophilic monophase fluid, hydrophilic monophase gel, and powder. The UV filter composition may be a spray, a cream, a lotion, a mousse (foam), or a powder, and might be packed accordingly in a bottle, a jar, a pump spray, an aerosol with an appropriate applicator. Examples of UV filter composition encompass in particular light-protective preparations (sun care products, sunscreens), such as sun protection milks, sun protection lotions, sun protection creams, sun blocks or topical’s or day care creams with or without a SPF (sun protection factor) label such as e.g. anti-ageingpreparations, preparations for the treatment of photo-ageing, skin-tanningpreparations (i.e. compositions for the artificial / sunless tanning and / or browning of human skin), for example self-tanning creams as well as skin lightening preparations. An “ingredient function” refers to the purpose of the ingredient within the composition. Ingredients can perform several functions. Ingredient functions can be, for example at least one of the functions of the following, non-limitative list:UVA and / or UVB filters, emollients, emulsifiers, thickener for the aqueous phase,thickeners for the oil phase, preservatives, complexing agents. An ”ingredient physico-chemical property” is an empirically measurable or calculable property that defines the ingredient’s physical and / or chemical nature. Physico-chemical properties of ingredients can, for example, be retrieved from the Handbook of Organic Solvent Properties or the Handbook of Chemistry and Physics. Examples of physico-chemical properties for liquid ingredients such as emollients and solvents are illustrated in the following non-limitative list: viscosity, density, boiling point, molecular weight, freezing point, liquid expansion coefficient, flash point, specific gravity, electrical conductivity, saturated vapor concentration, miscibility with water, solubility parameter, dipole moment, dielectric constant, polarity, refractive index, surface tension, dipole, latent heat, specific heat, Van der Waals’ volume, Van der Waals’ surface area, log Pow. Such physico-chemical properties for liquid ingredients such as emollients, solvents, and liquid UV filters may be (but not limited to): Viscosity, density, boiling point, molecular weight, freezing point, liquid expansion coefficient, flash point, specific gravity, electrical conductivity, saturated vapor concentration, miscibility with water, solubility parameter, dipole moment, dielectric constant, polarity, refractive index, surface tension, dipole, latent heat, specific heat, Van der Waals’ volume, Van der Waals’ surface area, spreading properties, spreading value. Possible properties for surface active agents such as wetting and dispersion agents, emulsifiers, foaming and anti-foaming agents, lubricants, may be (but not limited to): Solubility, surface tension, chemical nature, ionic character (non-ionic, anionic, cationic), HLB (Hydrophilic-Lipophilic Balance), emulsification mechanism, ionization potential, ionic charge, film-forming properties. Additional possible properties specific to UV filters may be (but not limited to): Chemical structure, photostability, solubility, hydrolytic stability. Possible properties of humectants may be (but not limited to): Hygroscopicity, water absorption, hydrophilicity, osmotic pressure. Possible properties of thickening agents may be (but not limited to): Viscosity, solubility, hydration, pH sensitivity, thermal stability, shear thinning / rheology, film-forming properties, ionic charge. Possible properties of polymeric ingredients may be (but not limited to): Molecular weight, number average molecular weight, weight average molecular weight, polydispersity, molecular weight distribution, dielectric constant, modulusof elasticity, tensile strength, crystallinity, resilience, ultimate elongation, yieldstress, yield strain, Rockwell hardness, Notched Izod impact strength, specific gravity, specific heat, thermal expansion, maximum continuous service temperature, clarity, color, refractive index, heat capacity / conductivity, heat of fusion, iodine value, acid value, saponification value, glass transition temperature, degree of branching, degree of cross linking, Possible properties of insoluble, pigmentary ingredients may be (but not limited to): Particle size, particle size distribution, polydispersity, coating chemistry, porosity, particle characteristics. A “value” can be defined relatively to other ingredients of a composition or in absolute units. A “quantity” can be defined relatively to other components or in absolute units, moreover a quantity can refer to a weight or a concentration. A “feature” refers to any data associated with an ingredient digital identifier, or a UV filter composition digital identifier of a set used to train a predictive machine-learning model. The “overall UV spectrum” of a UV filter composition can be calculated using various analytical techniques and instruments designed to measure the absorbance or transmittance of UV radiation across a range of wavelengths. The overall UV spectrum of the composition can be calculated by considering the empirically measured individual absorbance spectra of its ingredients (e.g., UV filters, additives) and their respective concentrations in the UV filter composition. This calculation may involve summing up the absorbance contributions from each component to obtain the overall absorbance spectrum of the UV. The “UV spectrum” of individual filters or of any combination of UV filters may be also considered as input features for the predictive model. The overall UV spectrum is defined as: Where εi, βi, and MWi are the molar extinction (L mol-1 cm-1) determined byempirical measurement, weight concentration (%) and molecular weight (g / mol) of filter i. Similarly, the UV spectrum of a filter i may be defined as a linear function proportional to ^^(^), while the UV spectrum of a combination of UV filters may be defined as a linear combination of one or more ^^(^)’s for multiple values of j. Then a possible way to predict the SPF uses the following approach (Gamma) based on the Beer-Lambert Law: The transmittance (T) is calculated on an irregular film according to: Where d is the average thickness of the applied sunscreen (2mg / cm2=>0.002 cm), and h(F) is the inverse cumulative distribution of the gamma distributiongiven by: As the area under the film profile is normalized to 1, the following relationship must hold: ∫ ^ ^ℎ(^)^^ = 1(4) which is the case when in formula (3) c = 1 / b. It should be noted that the overall UV spectrum of a UV filter composition can also be measured. As used herein the description a “calculated overall UV spectrum” is a value representative of the overall UV spectrum whether measured or calculated. The “calculated UV spectrum” is a value representative of the UV spectrum of one or more individual filters whether measured or calculated. The “Sun Protection Factor” (SPF) is a measure of a sunscreen's ability to protect the skin from harmful ultraviolet (UV) rays, primarily UVB radiation, which causes sunburn. SPF indicates how long a person can stay in the sun without getting sunburned compared to how long they could stay without any protection. SPF is typically measured through in vivo testing, meaning it is tested on human subjects. The SPF can then be calculated according to: Where Ser is the erythemal action spectrum and Ss is the spectrum of the light source. The shape parameter c, which describes the shape of the gamma distribution in formula (3) is given by: Where OFC is the total amount of UV filters given by: ^^^ = ∑^ ^^^ ^^(7) And REAE gives the proportion of the erythemal energy absorbed in the oil phase (o) in relationship to the total absorbed erythemal energy (water (w) and oil phase) and is calculated according to: ^^^ = ∑^^^ ^^^ ^^^^^ ^^,^(^, ^)^^^(^)^^(^)(8) Finally, parameters p1 to p5 can be calibrated by using an optimizer. The “proportion of erythemal energy absorbed in the oil phase” refers to the fraction of ultraviolet (UV) radiation responsible for causing erythema (skin reddening) that is absorbed by the oily components of a sunscreen or other topical product formulation. This measurement is significant in understanding the efficacy of sunscreens in protecting the skin against UV-induced erythema. The” total amount of UV filters” refers to the combined concentration or quantity of UV filtering ingredient present in a UV filter composition. UV filter ingredients are active ingredients that work to absorb, reflect, or scatter UV radiation, thereby protecting the skin from the harmful effects of the sun. A “predictive machine learning model” is a computational algorithm or mathematical framework designed to make predictions or forecasts based on input data. These models learn patterns and relationships from input data to generalize and make predictions about unseen or future data instances. Predictive machine learning models are used across various domains, including finance, healthcare, marketing, and natural language processing, among others. In a general manner, the terms “digital identifier” refer to any bijective digital representation of a physical item, such as an ingredient or a composition. Such a digital identifier may correspond to, for example, an entry in a database. A digital identifier may refer to a label representative of the name, chemical structure or internal reference of an ingredient or a composition, for example. In the context of the present description, the term “materialized” of “assembled” is intended as existing outside of the digital environment of the present invention. “Materialized” or “assembled” may mean, for example, readily found in nature or synthesized in a laboratory or chemical plant. In any event, a materialized or assembled UV filter composition digital identifier presents a tangible reality. The terms ‘to be compounded’ or ‘compounding’ refer to the act of materialization of a UV filter composition digital identifier, whether via extraction and assembly of ingredients or via synthetization and assembly of ingredients. As used herein, the terms “means of inputting” refer to, for example, a keyboard, mouse and / or touchscreen adapted to interact with a computing system in such a way to collect user input. In variants, the means of inputting are logical in nature, such as a network port of a computing system configured to receive an input command transmitted electronically. Such an input means may be associated to a GUI (Graphic User Interface) shown to a user or an API (Application programming interface). In other variants, the means of inputting may be a sensor configured to measure a specified physical parameter relevant for the intended use case. Examples of means of inputting are disclosed in regard to figure 2. As used herein, the terms “computing system”, “computer”, or “computer system” designate any electronic calculation device, whether unitary or distributed, capable of receiving numerical inputs and providing numerical outputs by and to any sort of interface, digital and / or analog. Typically, a computing system designates either a computer executing a software having access to data storage or a client-server architecture wherein the data and / or calculation is performed at the server side while the client side acts as an interface. Examples of such computing systems are disclosed in regard to figure 2. As used herein, the term “database” refers to any physical and / or virtual storage of data. Such a database can refer to a computer memory accessible in a computer network, upon which a database management system is ran. In simpler embodiments, a database can correspond to a table stored in a computer memory. Such a table may be stored in a document, such as a Microsoft Excel document. Figure 1 represents a particular embodiment of the method for providing at least one predicted Sun Protection Factor (SPF) value associated to an Ultraviolet (UV) filter composition digital identifier, representing a materializable UV filter composition, associated with at least one ingredient digital identifier representative of an existing ingredient of the materializable UV filter composition. Figure 1 represents different steps which can be grouped in five different phases: -the first phase relates to the constitution of a database and comprises thefollowing steps: -a step of assembling 105 a UV filter composition with at least oneingredient, -a step of calculating 110 at least one UV spectrum of theassembled UV filter composition, -a step of empirically measuring 115 an SPF value of theassembled UV filter composition, and -a step of storing 120, in a database, at least one empiricallymeasured SPF value, at least one calculated UV spectrum in association with at least one UV filter composition digital identifier representative of the assembled UV filter composition, -the second phase relates to the pre-processing 125 of the training setand will be described with regards to the particular embodiment represented in figure 5, -the third phase relates to the training of a predictive machine-learningmodel and comprises the following steps: -a step of providing 135 an original set of exemplar UV filtercomposition digital identifiers, said set of exemplar UV filter composition digital identifiers being representative of materialized UV filter compositions, comprising: -at least one value representative of an existing ingredient orat least one value representative of a combination of existing ingredients, each existing ingredient being associated with an ingredient digital identifier,- at least one value representative of at least one calculated UVspectrum associated with said UV filter composition digital identifiers, and -at least one value representative of an empirically measuredSPF value associated with said UV filter composition digital identifiers, to form a training set and -a step of training 140 a predictive machine-learning model using thetraining set, wherein the predictive machine-learning model is trained to predict at least one SPF value, -the fourth phase relates to the prediction of at least one SPF valuecomprising the following steps: -a step of defining 145 at least one UV filter composition digitalidentifier, -a step of predicting 150 at least one SPF value using the trainedpredictive machine-learning model and -a step of providing 155 the predicted SPF value,- the fifth phase relates to the materializing of at least one UV filtercomposition and comprises the following steps: -a step of sending 160 a digital command representative of aninstruction of materialising at least one UV filter composition corresponding to at least one UV filter composition digital identifier defined. -a step of materialising 165 at least one UV filter compositioncorresponding to at least one UV filter composition digital identifier defined. The first, second and fifth phases are optional embodiments of the method subject of the invention. The first phase relates to the constitution of a database based on real world experimentation. During the first phase, at least one UV filter composition is assembled 105 in a manner known to one skilled in the art with at least one ingredient. After the UV filter composition is assembled at least one value representative of an UV spectrum is calculated 110, and at least one value representative of an SPF value is empirically measured 115. Such measurements and calculations are known toone skilled in the art and can be based on equation 1 stated above or any othermanner known to one skilled in the art. During the step of storing 115, each composition assembled is associatedwith a UV filter composition digital identifier. The database associates said UV filtercomposition digital identifier with at least one ingredient digital identifier of said UV filter composition and at least one value representative of a calculated UV spectrum, and at least one value representative of an empirically measured SPF value. The database can also store at least one of the following elements: -at least one value representative of an ingredient quantity,- at least one value representative of an ingredient function associated withat least one ingredient digital identifier, -at least one value representative of an ingredient physico-chemical propertyassociated with at least one ingredient digital identifier or with a combination of at least two ingredient digital identifiers, -at least one value representative of the proportion of erythemal energyabsorbed in oil phase associated with at least one UV filter composition digital identifier and / or -at least one value representative of the total amount of UV filters in thecomposition associated with at least one UF filter composition digital identifier. Each at least one value representative of an ingredient is, for example, measured empirically during the step of assembling 105 the UV filter composition. At least one value representative of an ingredient function can be input by a user or automatically input via a user interface for example, based on the ingredient digital identifier. For example, an automatic input can be performed by retrieving data from a database linking at least one ingredient digital identifier to at least one ingredient function. At least one value representative of an ingredient physico-chemical property can be input by a user or automatically input via a user interface for example, based on the ingredient digital identifier. For example, an automatic input can be performed by retrieving data from a database linking at least one ingredient digital identifier to at least one ingredient physico-chemical property. The database linking at least one ingredient digital identifier to at least one ingredient physico-chemical property and the database linking at least one ingredient digital identifier to at least one ingredient function can be the same database. In said database, at least one ingredient physico-chemical property can be related to at least one ingredient function. Examples of such links can be found below in Table 1. [Table 1] Function Physico-chemical propertyEmollient Total weightViscosity of mixture Spreadability (slow / medium / fast) Density of mixture Filter Liquid filters in oil phaseSolid filters in oil phase Filters in water phase Pigments in oil phase Pigments in water phase Viscosity: liquid filters Oil thickener Cetearyl alcoholGlyceryl myristate Cetyl alcohol Stearyl alcohol Oil thickener others Solvents AlcoholHumectants GlycerinHumectant others After the step of storing 115, the database is used during the step of providing 135 a training set of the third phase. The predictive machine-learning model is therefore trained based on at least part of the elements constituting the database stated above. The inclusion of the overall UV spectrum and OFC is motivated by their strong correlation with SPF values. Specifically, for the illustrative datasetconsidered in this invention, the highest correlation is found for E7, with ρ = 0.756,while correlation for OFC is ^ = 0.739. On the other hand, the highest correlationfor any features different from OFC, and extinctions is for one of the ingredients,with ρ = 0.422. These findings are summarized in Table 2, below. Therefore, anypredictive model would greatly benefit from the inclusion of extinction coefficients and OFC, and this is the strategy proposed in the present invention. [Table 2] Sample SPF correlations FEATURE SPF correlationOFC 0.739528 E(290) 0.724637 E(295) 0.736281E(300) 0.742352E(305) 0.741382E(310) 0.734107E(315) 0.738678E(320) 0.7497E(325) 0.756087E(330) 0.735996E(335) 0.711747E(340) 0.693667E(345) 0.675617E(350) 0.662118E(355) 0.64943E(360) 0.640597E(365) 0.634658E(370) 0.625815E(375) 0.604476E(380) 0.567099E(385) 0.557649E(390) 0.550932E(395) 0.408503E(400) 0.249234In particular embodiments, the computer-implemented method 100 further comprises a step of pre-processing 125 the training set, and wherein the pre- processed training set is used during the step of providing 135 a training set. The predictive machine-learning model is therefore trained based on at least part of the elements constituting the reduced training set stated above. In particular embodiments, the training set comprises a large number of data, the predictive machine learning model determines which features have a strong correlation with the SPF value or not. Predictive machine learning models such as random forest regressors, gradient boosting regressors, and neural networks can learn feature importance during training discriminating unimportant features. In particular embodiments, the training set has a limited number of data, arisk of overfitting due to the high dimensionality of the corresponding model is present. In these embodiments, feature engineering becomes particularly important. In particular embodiments, the step of pre-processing 125 comprises: -a step of grouping 505 at least one ingredient digital identifiers dependingon at least one value representative of an ingredient function associated with the at least two ingredient digital identifier, -a step of removing unimportant features by domain expert knowledge orcross validation, -an optional step of determining 510 a value representative of the influenceof at least one groups on the predicted SPF value, -an optional step of selecting 515, by cross validation, the most influential atleast one group and -an optional step of eliminating 520 at least one less influential at least onegroup from the feature set. In particular embodiments, during the step of selecting 515, each group having a determined value representative of the influence on the predicted SPF value above a predefined influence selection value is selected. In particular embodiments, during the step of selecting 515 comprises a step of ranking each group based on the determined value representative of the influence on predicted SPF value from the highest value to the lowest value and a step of choosing a predefined number of at least one group having the highest value representative of the influence on predicted SPF value. The step of pre-processing 125 leads to a drastic reduction of features. Thus, in the illustrative training sets S1 described hereafter, 430 ingredients can be reduced to 130. In particular embodiments, the step of pre-processings125 comprises a step of at least partly replacing 525 an ingredient digital identifier by a value representative of at least one physico-chemical property. Thus, the illustrative training set S2 described hereafter illustrative resulted in the 21 features shown intable 1 without loss of accuracy.In particular embodiments, at least one value representative of a function is associated with at least one value representative of a physico-chemical property in the database. In these embodiments, the step of replacing 525 is prior to the step of grouping and during the step of grouping a step of grouping 505 at least one value representative of a physico-chemical property is grouped depending on at least one value representative of an ingredient function associated with the at least values representative of the same physico-chemical property, The step of pre-processing 125 is optional and depends on the content of the training set used. It is noted here that the predictive machine learning-model can be trained on a training set initially formed using only at least one value representative of an UV spectrum and at least one of -at least one value representative of an ingredient function,- at least one value representative of an ingredient physico-chemical propertyassociated with at least one value representative of an ingredient function. In said embodiments, the training set can also comprise: -at least one value representative of an ingredient physico-chemical propertyassociated with at least one value representative of an ingredient function, -at least one value representative of the proportion of erythemal energyabsorbed in oil phase associated with at least one UV filter composition digital identifier and / or- at least one value representative of the total amount of UV filters in thecomposition associated with at least one UV filter composition digital identifier. In other words, said database is independent of ingredient digital identifiers. Figure 1 also shows a computer-implemented method 130 to train a predictive machine learning model to provide at least one value associated to an Ultraviolet (UV) filter composition digital identifier, representing a materializable UV filter composition, comprising the steps of: -a step of providing 135 an original set of exemplar UV filter compositiondigital identifiers, said set of exemplar UV filter composition digital identifiers being representative of materialized UV filter compositions, comprising: -at least one existing ingredient or at least one value representativeof a combination of existing ingredients, each existing ingredient being associated with an ingredient digital identifier and -said UV filter composition digital identifiers being associated with atleast one value representative of a calculated UV spectrum, and -said UV filter composition digital identifiers being associated with atleast one value representative of an empirically measured SPF value, to form a training set, -a step of training 140 a predictive machine-learning model using the trainingset, wherein the predictive machine-learning model is trained to predict at least one SPF value. Such method 130 for training can be a method object of the present invention in itself, independently from the method 100 for providing at least one predicted SPF value. The step of providing 135 is performed, for example, by a computer program executed by an electronic computation device, such as a microprocessor. During this step of providing 135, at least one UV filter composition digital identifier is provided to the predictive machine-learning model to be trained. Such a step of providing 135 may correspond to the transfer in memory of the exemplar UV filter composition digital identifier from a digital storage location to another, the latter being dedicated to the training set of the predictive machine-learning model. During such a step of providing 135, a larger sample of UV filter composition digital identifiers representative of distinct UV filter compositions may be divided into a training set and into a validation set. During the step of providing, the training set provided can be a pre- processed training set obtained as a result of the step of pre-processing 125 disclosed above or a dataset independent of ingredient digital identifiers as disclosed above. The step of training 140 may use any training method appropriated for training a predictive machine learning model. The fourth phase of the method represented in figure 1, relates to the prediction and provision of the at least one predicted SPF value by the trained predictive machine-learning model. The step of predicting 150 is performed, for example, by a computer program executed upon an electronic computing unit or computing system, such as a computer, for example. During this step, the trained predictive machine learning model is executed to provide, as an output, a predicted SPF value. The step of providing 145 can be performed analogically to the related steps of providing 135 of the method 100. In particular embodiments, the trained predictive machine-learning model is a Random Forest Regressor. A Random Forest Regressor is a bagging ensemble method that enhances performance by combining the predictions of a set of decision trees. The method works by creating numerous bootstrap samples of the training set and by training randomized trees on each sample. Specifically, the random forest algorithm leverages two techniques to reduce variance of the prediction function, known as bagging and de-correlation of trees. The variance of the forest can be expressed as: Where: ^^is the variance of the individual tree, ^is the pairwise correlation of the trees,while ^ is the number of trees.Increasing the number of trees and reducing their pairwise correlations induce a variance reduction. A random forest regressor is constructed by the following algorithm: Firstly, for b=1 to B: (i) Draw a bootstrap sample ^^ of size N from the full dataset(ii) Grow a tree ^^(^, ^^) on ^^ by randomly selecting ^ featuresbefore node splitting, until the minimum node size ^^^^is reached Secondly, define the Random Forest as: Public implementations of random forest regressors are available in the scikit-learn python library, for example. In particular embodiments, the trained predictive machine-learning model is a Gradient Boosting Regressor. Gradient Boosting Regressors belong to a class of machine-learning models that enhance performance by combining the predictions of multiple models, a concept known as ensemble learning. Specifically, a Gradient Boosting Regressor is a sequential ensemble method that builds a series of models, each minimizing the errors of the previous models. An additive model of the following form is constructed: where: x is the input vector, ^^ , ^ = 1,2, … ^ are the expansion coefficients,^^ , ^ = 1,2, … ^ are the basis functions, and^ is a set of parameters. The machine-learning model is then trained by minimizing a loss functionaveraged over the training data (^^ , ^^)^^^^: ^^^, ^(^)^ = ∑^ ^^^ ^^^^ , ^(^^)^(13) In the context of boosting, the minimization of the loss function is done efficiently with an iterative procedure known as Forward Stagewise Additive Modeling. Optimization is performed sequentially, by adding new basis functions to the additive model, and by tuning only the new basis function at each iteration of the training stage: A Gradient Boosting Regressor is constructed by the following algorithm: Firstly, initialize Secondly, for m=1 to M: -Compute -Set The boosted model is ^(^) = ^^(^)The optimization of the second step can be performed by using first or higher order algorithms. A popular variant is Gradient boosting, where at each iteration the gradients, are used to fit a basis learner ^^ to the dataset {^^ , ^^^}^^^^, while the expansion coefficient ^^is fitted by solving the one-dimensional optimizationproblem (17). Higher order variants exist that scale well to very large datasets and parallel environments. In preferred embodiments, the Gradient Boosting Regressor uses a tree function. Public implementations of Gradient Boosting machine-learning models are available in the scikit-learn or the XGBoost libraries. In preferred embodiments, the trained predictive machine-learning model is a Neural Network. Neural networks represent a broad class of machine-learning models. Neural networks present a great scalability and expressiveness, combined with powerful optimization algorithms, which enable Neural Networks to drive many cutting-edge applications across various fields. In the basic form, a Neural Network is composed of multiple layers of interconnected nodes known as neurons. The layers are commonly referred to as input, hidden and output layers. The input layer consists of the input features, the hidden layers apply consecutive (non-linear) transformations to the input layer,while the output layer is the result of these transformations. A network with ^ − 1hidden layers is defined recursively as follows: where: ^(^)is the output of the l-th layer is a non-linear activation function, typically Sigmoid, Sinh, and ReLU ^(^)is the weight matrix the output of the (l-1)-th layer the bias vector with ^(^), ^(^) the input and output vector respectively.The set of ^(^)(^) are commonly referred to as activations.Given training data the Neural Network is trained by minimizing the loss function: Optimization is typically done with first-order optimization algorithms such as gradient descent. When training Neural Networks, regularization techniques are essential to prevent overfitting. Overfitting occurs when the model learns noise and spurious patterns in the data, leading to poor performance on unseen data. This is particularly relevant when the number of training samples is much smaller than the number of parameters of the neural network. This regime is common in the context of SPF prediction, as in-vivo measurements are quite limited, while the effective size of the neural networks is larger. Common regularization techniques include: -L2 Regularization (weight decay): which introduces a penalty term to theloss function proportional to the sum of the squares of the weights (the entries of the weight matrix) and therefore encourages the optimization algorithm to maintain smaller weights, effectively resulting in simpler machine-learning models, -L1 Regularization: which introduces a penalty term proportional to the sumof the absolute values of the model weights and therefore encourages the optimization algorithm to set many weights to zero, effectively leading to simpler models; moreover, it allows the model to ignore irrelevant / redundant features, effectively resulting in feature selection, and -Dropout: which randomly deactivates a fraction of the neurons duringtraining; the random deactivation is repeated at each training epoch and at each iteration, training is effectively applied to a simpler model, preventing the network from fitting spurious and noisy patterns in the data. Aneural network configuration example which can be used for SPFquantification prediction may be characterized as follows. The input layer has number of neurons corresponding to the number of features. The number of hiddenlayers may be set to three, with 64, 32, and 16 neurons respectively. The outputnode corresponds to one neuron quantifying the SPF. An ensemble-based method configuration which can be used for SPFquantification prediction may be characterized as follows. An XGBoost regressorwith input vector dimension corresponding to the number of features. The numberof tree estimators equals 100, and depth of the tree equals 6. The output nodequantifies the SPF. In the following examples, a training set based on 291 (N) UV filter compositions was used. Two training sets of feature vectors can be considered, S1 and S2. The first training set (S1) includes the following features:a) The weighted average of the extinction coefficients at ^^ = 23 wavelengths,defined as: ^^⃗ = ^^ ⋅ ^⃗ (24)where ^^ is a ^^ by ^^ matrix containing the extinction coefficients for the^^ UV filters at ^^ wavelengths, while ^⃗ is a ^^ -vector containing thepercentage concentrations of UV filters in the UV filter composition. b) The OFC:OFC = ∑^^^^^^^(25) c) The REAE, defined as: Where ^^^^is the erythema active extinction of filter ^, while ^^and ^^are the set of indices of oil- and water-soluble filters. the erythema activeextinction is defined as: ^^^ Where ^^^(^, ^) is the specific extinction of filter ^ at wavelength ^, ^^(^) isthe intensity of the light source, while ^^^(^) is the erythemal actionspectrum. d) The list of ingredients of the formulationThe second training set (S2) includes features a)-c) and: e) The list of properties P1, where the function and property names arereported as illustrated in Table 1. For categorical variables, the corresponding categories are reported. Given the set of input features the following pre-processing steps can be applied to the training set: -a step of standardizing each feature to a standard normal distribution- a step of removing highly correlated features with given correlationthreshold ρ^ In this embodiment, the step of standardizing each feature is achieved by computing the mean and the standard deviation of each feature and by applying the following transformation: The step of removing is applied to collinear features and is achieved by computing the pairwise correlations: where ^^, ^^are the standard deviations of ^^, ^^respectively, while ^^^ is the covariance: with N is the number of observations. Given the pairwise correlations, a set of uncorrelated features is generated sequentially, by adding one feature at the time, such that its correlation with any of the previously selected features is no greater than ^^. Features are added until no new feature satisfies the correlation constraint. Both pre-processing steps are standard practice when developing a machine-learning model. Such steps ensure good convergence properties of thetraining method 130, and result in better in- and out-of-sample model performance.Furthermore, additional dimensionality reduction transformations may be applied to the dataset such as t-Distributed Stochastic Neighbor Embedding (t-SNE), Uniform Manifold Approximation and Projection (UMAP), Principal Component Analysis (PCA), Kernel Principal Component Analysis (KPCA), Linear Discriminant Analysis (LDA). In particular embodiments, to obtain a robust estimate of predictive machine learning model’s performance and generalization ability, a method of K-Fold cross validation, known to one skilled in the art, may be adopted. The method relies on dividing the dataset in K equally sized subsets, called folds. In a cross-validation iteration, the model is trained on K-1 folds, while it is validated on the remainingfold (test set). The method is repeated K times, in such a way that each fold is used exactly once for validation, while the remaining K-1 folds are used for training. The corresponding K train set, and test set performance scores are then averaged to produce the final estimate of predictive machine-learning model’s performance.Optionally, it is possible to repeat K-fold cross validation multiple times, byrandomly shuffling the dataset before each repetition. In particular embodiments when a predictive machine learning model includes feature engineering and hyperparameters tuning, an appropriate strategy to estimate model’s generalization ability is a nested cross validation, known to one skilled in the art, was adopted. As for K-fold cross validation, the data set is divided into K folds. In each iteration, K-1 folds are used for hyperparameters tuning, features engineering, and predictive machine-learning model training, while the remaining fold is used for predictive machine-learning model validation (test set). In these embodiments, the choice of model’s hyperparameters and featuresengineering does not rely on any knowledge of the validation data, providing amore accurate estimate of predictive machine-learning model generalization ability. Hereinafter are four tables illustrating the performance of the trained predictive machine-learning models based on the set S1 and S2 and using the method of nested cross validation as stated above. In example 1, the data set used is S1. The table here below illustrates the nested cross validation result ^^, based on the cross-validation training sets and test sets as stated above. Nested Cross validation ^^ - Features S1 (Extinctions, Ingredients, OFC,REAE) Gamma RF XGB GB NNTrain 0.668 0.954 0.964 0.957 0.904Test 0.653 0.722 0.729 0.724 0.688Figure 3 illustrates the correlation of the measured SPF value and thecalculated SPF value of the trained predictive machine-learning model based on the training set S1. Figure 3 shows the accuracy of the trained predictive machine- learning model used in the method of the present invention as well as the corresponding training method. The table here below illustrates the Nested cross validation result ^^based on the cross-validation training sets and test sets having only features c and d of S1. Ablation analysis: Nested Cross validation ^^ - Features S1 (Ingredients,REAE) Gamma RF XGB GB NNTrain 0.668 0.931 0.975 0.971 0.918Test 0.653 0.621 0.663 0.650 0.658As can be seen, the results are far less correlated resulting in a less performant trained predictive machine-learning model. In example 2, the data set used is S2. The table here below illustrates the nested cross validation result ^^based on the cross-validation training sets and test sets. Nested Cross validation ^^ - Features S2 (Extinctions, Properties, OFC,REAE) Gamma RF XGB GB NNTrain 0.668 0.948 0.955 0.948 0.879Test 0.653 0.708 0.714 0.710 0.690Figure 4 illustrates the correlation of the measured SPF value and thecalculated SPF value of the trained predictive machine-learning model based on the training set S2. Figure 4 shows the accuracy of the trained predictive machine- learning model used in the method of the present invention as well as the corresponding training method. The table here below illustrates the nested cross validation result ^^based on the cross-validation training sets and test sets having only feature d and e of S2. Ablation analysis: Nested Cross validation ^^ - Features S2 (Properties,REAE) Gamma RF XGB GB NNTrain 0.668 0.917 0.947 0.954 0.838Test 0.653 0.570 0.567 0.568 0.451As can be seen, the results are far less correlated resulting in a less performant trained predictive machine-learning model. Acomplementary method to analyze the relevance of additional features tothe performance of the models is the computation of feature importance scores.Multiple algorithms for the computation of importance scores exist, such as theSHAP method, importance score based on feature permutation, or tree-basedalgorithms. Figure 6 shows importance scores computed by using the XGBoostlibrary. As can be seen, UV spectra and several additional ingredient and propertyfeatures are significant to the predictions. For example, features that can be used in the training of a machine learning algorithm are at least one of: -the E290 capability of an ingredient or composition,- the liquid filters in oil phase parameter of an ingredient or composition,- the E325 capability of an ingredient or composition,- the E330 capability of an ingredient or composition,- the E335 capability of an ingredient or composition,- the E375 capability of an ingredient or composition,- the OFC of an ingredient or composition,- the viscosity of an ingredient or composition,- the viscosity of liquid filters parameter of an ingredient or composition,- the wos parameter of an ingredient or composition,- the REAE of an ingredient or composition,- the total weight of all filters parameter of an ingredient or composition,- the E395 capability of an ingredient or composition,- the spreadability of an ingredient or composition,- the E400 capability of an ingredient or composition,- the density of an ingredient or composition, and / or- the filters in water phase of an ingredient or composition.The following features may be used: FUNCTION nr feature abbr numberemollient1 total weight of all e_w 1(503)emollients (%)2 viscosity of mixture viscosity 13-5 spreadability (%)| fast / medium / slow 3 36 density of mixture density 1- polarity high / med / low 3- chemistry ester / ether / silicon 3 3filter7 total weight(%) f_w 1(524, 517)8 liquid filters in oil wol 1phase(%) 9solid filters in oil wos 1phase(%) 10 filters in water ww 1phase(%)| 11 pigments in oil wOp 1phase(%) 12 pigments in water wwp 1phase(%) 13 viscosity: liquid filters wol_v 1oil thickener14 cetearyl alcohol (145) ot cta 1(421) 15 glyceryl myristate (118) ot_gm 116 cetyl alcohol (115) ot_ca 117 stearyl alcohol (93) ot_sa 118 oil thickener others ot_o 1solvents19 alcohol (93) etoh 1(104) humectants20 Glycerin (327) gly 1(458) 21 Humectant others h_o 1It should be noted that the features listed herein differ, for the most part, from the prior art. The present method can use the following features that are new:- calculated UV spectrum,- concentration of individual UV filters,- physico-chemical properties of ingredients or ingredient compositions,- concentrations of individual emollients,- ratio of erythemal absorbance and oil phase,- concentrations of ingredients other than UV filters and emollients.It should be noted that individual ingredients may be replaced by their physico-chemical properties for better generalization. The prediction method presented in this document constitutes an indirectmeasurement method for the SPF of sunscreen formulations. Specifically, physico-chemical properties of the formulation or of the constituting ingredients are directlymeasured by specific sensors, or experimental procedures. This data is then usedby the prediction method to obtain indirect (in-silico) measurements of the SPF.The method presented also addresses intrinsic limitations of machinelearning algorithms. Specifically, learning algorithms achieve sub-optimal performance when data is scare (very common for SPF), and when features are poorly correlated with the output. Therefore, even if all information about the input compound is implicitly or indirectly contained in the input features, a model may achieve suboptimal performance, because limited data availability and the finite complexity of the model do not allow to fully model such implicit information. The method presented in this work addresses this by identifying features which haveboth an high information content and are, at the same time, highly correlated withthe output. This is clearly demonstrated in the examples above. Modelperformance when using ingredients and REAE only, which indirectly includes allinformation of the formulation, is relatively modest, while the inclusion of the UVspectra significantly boosts performance.Importantly, the SPF prediction method can be used for the development ofnew sunscreen formulations. Specifically, the formulators may develop tens orhundreds of formulations, while requiring only specific SPF target values. Themethod can be used to select only a subset of formulations that meet the requiredSPF values. The selected formulations can be then further optimized by theformulator and characterized with in-vivo experiments. Therefore, the predictionmethod may be directly used as integral part of the development process, with theeffect of reducing unnecessary and expensive screening rounds and improving thedevelopment efficiency.During the fifth optional phase, the step of sending 160 a digital command representative of an instruction of materialising at least one UV filter composition provided is generated and send to a device capable of materializing the at least one UV filter composition corresponding to at least one UV filter composition digital identifier defined. The step of materializing 165 may be performed manually, via a user interface for example, or automatically, via the device capable of materializing the at least one UV filter composition corresponding to at least one UV filter composition digital identifier defined based on the command sent. Figure 2 represents a block diagram that illustrates an example computer system 200 with which an embodiment of the present invention may be implemented. In the example of figure 8, a computer system 205 and instructions for implementing the disclosed technologies in hardware, software, or a combination of hardware and software, are represented schematically, for example as boxes and circles, at the same level of detail that is commonly used by persons of ordinary skill in the art to which this disclosure pertains for communicating about computer architecture and computer systems implementations. The computer system 205 includes an input / output (IO) subsystem 220 which may include a bus and / or other communication mechanism(s) for communicating information and / or instructions between the components of the computer system 205 over electronic signal paths. The I / O subsystem 220 may include an I / O controller, a memory controller and at least one I / O port. The electronic signal paths are represented schematically in the drawings, for example as lines, unidirectional arrows, or bidirectional arrows. At least one hardware processor 210 is coupled to the I / O subsystem 220 for processing information and instructions. Hardware processor 210 may include, for example, a general-purpose microprocessor or microcontroller and / or a special-purpose microprocessor such as an embedded system or a graphics processing unit (GPU) or a digital signal processor or ARM processor. Processor 210 may comprise an integrated arithmetic logic unit (ALU) or may be coupled to a separate ALU. Computer system 205 includes one or more units of memory 225, such as a main memory, which is coupled to I / O subsystem 220 for electronically digitally storing data and instructions to be executed by processor 210. Memory 225 may include volatile memory such as various forms of random-access memory (RAM) or other dynamic storage device. Memory 225 also may be used for storing temporary variables or other intermediate information during execution of instructions to be executed by processor 210. Such instructions, when stored in non-transitory computer-readable storage media accessible to processor 210, can render computer system 205 into a special-purpose machine that is customized to perform the operations specified in the instructions. Computer system 205 further includes non-volatile memory such as read only memory (ROM) 230 or other static storage device coupled to the I / O subsystem 220 for storing information and instructions for processor 210. TheROM 230 may include various forms of programmable ROM (PROM) such aserasable PROM (EPROM) or electrically erasable PROM (EEPROM). A unit of persistent storage 215 may include various forms of non-volatile RAM (NVRAM), such as FLASH memory, or solid-state storage, magnetic disk, or optical disk such as CD-ROM or DVD-ROM and may be coupled to I / O subsystem 220 for storing information and instructions. Storage 215 is an example of a non-transitory computer-readable medium that may be used to store instructions and data which when executed by the processor 210 cause performing computer-implemented methods to execute the techniques herein. The instructions in memory 225, ROM 230 or storage 215 may comprise one or more sets of instructions that are organized as modules, methods, objects, functions, routines, or calls. The instructions may be organized as one or more computer programs, operating system services, or application programs including mobile apps. The instructions may comprise an operating system and / or system software; one or more libraries to support multimedia, programming or other functions; data protocol instructions or stacks to implement TCP / IP, HTTP or other communication protocols; file format processing instructions to parse or render files coded using HTML, XML, JPEG, MPEG or PNG; user interface instructions to render or interpret commands for a graphical user interface (GUI), command-line interface or text user interface; application software such as an office suite, Internet access applications, design and manufacturing applications, graphics applications, audio applications, software engineering applications, educational applications, games or miscellaneous applications. The instructions may implement a web server, web application server or web client. The instructions may be organized as a presentation layer, application layer and data storage layer such as a relationaldatabase system using structured query language (SQL) or no SQL, an objectstore, a graph database, a flat file system or other data storage. Computer system 205 may be coupled via I / O subsystem 220 to at least one output device 235. In one embodiment, output device 235 is a digital computer display or Human Machine Interface. Examples of a display that may be used invarious embodiments include a touchscreen display or a light-emitting diode (LED)display or a liquid crystal display (LCD) or an e-paper display. Computer system 205 may include other type(s) of output devices 235, alternatively or in addition to a display device. Examples of other output devices 235 include printers, ticket printers, plotters, projectors, sound cards or video cards, speakers, buzzers or piezoelectric devices or other audible devices, lamps or LED or LCD indicators,haptic devices, actuators, or servos.At least one input device 240 is coupled to I / O subsystem 220 for communicating signals, data, command selections or gestures to processor 210. Examples of input devices 240 include touchscreens, microphones, still and video digital cameras, alphanumeric and other keys, keypads, keyboards, graphics tablets, image scanners, joysticks, clocks, switches, buttons, dials, slides. Another type of input device is a control device 245, which may perform cursor control or other automated control functions such as navigation in a graphical interface on a display screen, alternatively or in addition to input functions. Control device 245 may be a touchpad, a mouse, a trackball, or cursor direction keys for communicating direction information and command selections to processor 210 and for controlling cursor movement on display 235. The input device may have at least two degrees of freedom in two axes, a first axis (e.g., x) and a second axis (e.g., y), that allows the device to specify positions in a plane. Another type of input device is a wired, wireless, or optical control device such as a joystick, wand, console, steering wheel, pedal, gearshift mechanism or other type of control device. An input device 240 may include a combination of multiple different input devices, such as a video camera and a depth sensor. In another embodiment, computer system 205 may comprise an Internet of things (IoT) device in which one or more of the output device 235, input device 240, and control device 245 are omitted. Or, in such an embodiment, the inputdevice 240 may comprise one or more cameras, motion detectors, thermometers,microphones, seismic detectors, other sensors or detectors, measurement devices or encoders and the output device 235 may comprise a special-purpose display such as a single-line LED or LCD display, one or more indicators, a display panel, a meter, a valve, a solenoid, an actuator or a servo. Computer system 205 may implement the techniques described herein using customized hard-wired logic, at least one ASIC or FPGA, firmware and / or program instructions or logic which when loaded and used or executed in combination with the computer system causes or programs the computer system to operate as a special-purpose machine. According to one embodiment, thetechniques herein are performed by computer system 205 in response toprocessor 210 executing at least one sequence of at least one instruction contained in main memory 225. Such instructions may be read into main memory 225 from another storage medium, such as storage 215. Execution of the sequences of instructions contained in main memory 225 causes processor 210 to perform the process steps described herein. In alternative embodiments, hard- wired circuitry may be used in place of or in combination with software instructions. The term “storage media” as used herein refers to any non-transitory media that store data and / or instructions that cause a machine to operate in a specific fashion. Such storage media may comprise non-volatile media and / or volatile media. Non-volatile media includes, for example, optical or magnetic disks, such as storage 215. Volatile media includes dynamic memory, such as memory 225. Common forms of storage media include, for example, a hard disk, solid state drive, flash drive, magnetic data storage medium, any optical or physical data storage medium, memory chip, or the like. Storage media is distinct from but may be used in conjunction with transmission media. Transmission media participates in transferring information between storage media. For example, transmission media includes coaxial cables, copper wire and fiber optics, including the wires that comprise a bus of I / O subsystem 220. Transmission media can also take the form of acoustic or light waves, such as those generated during radio-wave and infra-red data communications. Various forms of media may be involved in carrying at least one sequence of at least one instruction to processor 210 for execution. For example, the instructions may initially be carried on a magnetic disk or solid-state drive of a remote computer. The remote computer can load the instructions into its dynamic memory and send the instructions over a communication link such as a fiber optic or coaxial cable or telephone line using a modem. A modem or router local to computer system 205 can receive the data on the communication link and convert the data to a format that can be read by computer system 205. For instance, a receiver such as a radio frequency antenna or an infrared detector can receive the data carried in a wireless or optical signal and appropriate circuitry can provide the data to I / O subsystem 220 such as place the data on a bus. I / O subsystem 220 carries the data to memory 225, from which processor 210 retrieves and executes the instructions. The instructions received by memory 225 may optionally be stored on storage 215 either before or after execution by processor 210. Computer system 205 also includes a communication interface 260 coupled to bus 220. Communication interface 260 provides a two-way data communication coupling to network link(s) 265 that are directly or indirectly connected to at least one communication network, such as a network 270 or a public or private cloud on the Internet. For example, communication interface 260 may be an Ethernet networking interface, integrated-services digital network (ISDN) card, cable modem, satellite modem, or a modem to provide a data communication connection to a corresponding type of communications line, for example an Ethernet cable or a metal cable of any kind or a fiber-optic line or a telephone line. Network 270 broadly represents a local area network (LAN), wide-area network (WAN), campus network, internetwork, or any combination thereof. Communication interface 260 may comprise a LAN card to provide a data communication connection to a compatible LAN, or a cellular radiotelephone interface that is wired to send or receive cellular data according to cellular radiotelephone wireless networking standards, or a satellite radio interface that is wired to send or receive digital dataaccording to satellite wireless networking standards. In any such implementation,communication interface 260 sends and receives electrical, electromagnetic, or optical signals over signal paths that carry digital data streams representing various types of information. Network link 265 typically provides electrical, electromagnetic, or optical data communication directly or through at least one network to other data devices, using, for example, satellite, cellular, Wi-Fi, or BLUETOOTH technology. For example, network link 265 may provide a connection through a network 270 to a host computer 250. Furthermore, network link 265 may provide a connection through network 270 or to other computing devices via internetworking devices and / or computers that are operated by an Internet Service Provider (ISP) 275. ISP 275 provides data communication services through a world-wide packet data communication network represented as Internet 280. A server computer 255 may be coupled to Internet 280. Server 255 broadly represents any computer, data center, virtual machine, or virtual computing instance with or without a hypervisor, or computer executing a containerized program system such as DOCKER or KUBERNETES. Server 255 may represent an electronic digital service that is implemented using more than one computer or instance and that is accessed and used by transmitting web services requests, uniform resource locator (URL) strings with parameters in HTTP payloads, API calls, app services calls, or other service calls. Computer system 205 and server 255 may form elements of a distributed computing system that includes other computers, a processing cluster, server farm or other organization of computers that cooperate to perform tasks or execute applications or services. Server 255 may comprise one or more sets of instructions that are organized as modules, methods, objects, functions, routines, or calls. The instructions may be organized as one or more computer programs, operating system services, or application programs including mobile apps. The instructions may comprise an operating system and / or system software; one or more libraries to support multimedia, programming or other functions; data protocol instructions or stacks to implement TCP / IP, HTTP or other communication protocols; file format processing instructions to parse or render files coded using HTML, XML, JPEG, MPEG or PNG; user interface instructions to render or interpret commands for a graphical user interface (GUI), command-line interface or text user interface; application software such as an office suite, Internet access applications, design and manufacturing applications, graphics applications, audio applications, software engineering applications, educational applications, games or miscellaneous applications. Server 255 may comprise a web application server that hosts a presentation layer, application layer and data storage layer such as a relational database system using structured query language (SQL) or no SQL, an object store, a graph database, a flat file system or other data storage. Computer system 205 can send messages and receive data and instructions, including program code, through the network(s), network link 265 and communication interface 260. In the Internet example, a server 255 might transmit a requested code for an application program through Internet 280, ISP 275, local network 270 and communication interface 260. The received code may be executed by processor 210 as it is received, and / or stored in storage 215, or other non-volatile storage for later execution. The execution of instructions as described in this section may implement a process in the form of an instance of a computer program that is being executed and consisting of program code and its current activity. Depending on the operating system (OS), a process may be made up of multiple threads of execution that execute instructions concurrently. In this context, a computer program is a passive collection of instructions, while a process may be the actual execution of those instructions. Several processes may be associated with the same program; for example, opening up several instances of the same program often means more than one process is being executed. Multitasking may be implemented to allow multiple processes to share processor 210. While each processor 210 or core of the processor executes a single task at a time, computer system 205 may be programmed to implement multitasking to allow each processor to switch between tasks that are being executed without having to wait for each task to finish. In an embodiment, switches may be performed when tasks perform input / output operations, when a task indicates that it can be switched, or on hardware interrupts. Time-sharing may be implemented to allow fast response for interactive user applications by rapidly performing context switches to provide the appearance of concurrent execution of multiple processes simultaneously. In an embodiment, for security and reliability, an operating system may prevent direct communication between independent processes, providing strictly mediated and controlled inter- process communication functionality.

Claims

CLAIMS 1. Computer-implemented method (100) for providing at least one predicted Sun Protection Factor (SPF) value associated to an Ultraviolet (UV) filter composition digital identifier, representing a materializable UV filter composition, associated with at least one ingredient digital identifier representative of an existing ingredient of the materializable UV filter composition, characterized in that it comprises: -a step of providing (135) an original set of exemplar UV filter compositiondigital identifiers, said set of exemplar UV filter composition digital identifiers being representative of materialized UV filter compositions, comprising: -at least one value representative of an existing ingredient or at leastone value representative of a combination of existing ingredients, each existing ingredient being associated with an ingredient digital identifier, -at least one value representative of at least one calculated UVspectrum associated with said UV filter composition digital identifiers, and -at least one value representative of an empirically measured SPFvalue associated with said UV filter composition digital identifiers, to form a training set, -a step of training (140) a predictive machine-learning model using thetraining set, wherein the predictive machine-learning model is trained to predict at least one SPF value, -a step of defining (145) at least one UV filter composition digital identifier,- a step of predicting (150) at least one SPF value using the trained predictivemachine-learning model and -a step of providing (155) the at least one predicted SPF value.

2. Computer-implemented method (100) according to claim 1, wherein, in the training set, at least one value representative of an existing ingredient is an ingredient quantity.

3. Computer-implemented method (100) according to claims 1 or 2, wherein, in the training set, at least one value representative of an existing ingredient is a value representative of a function of the existing ingredient.

4. Computer-implemented method (100) according to one of claims 1 to 3, wherein, in the training set, at least one value representative of an existing ingredient or of a combination of existing ingredients is a value representative of a physico- chemical property of the existing ingredient or of the existing combination of ingredients.

5. Computer-implemented method (100) according to claims 2, 3 and 4, whereinthe training set comprises more than 20 features selected at least from one of eachof value representatives of an existing ingredient quantity, existing ingredientfunction and existing ingredient physico-chemical property or of a combination of ingredients.

6. Computer-implemented method (100) according to one of claims 1 to 5, wherein the training set comprises at least one value representative of the proportion of erythemal energy absorbed in oil phase associated with at least one UV filter composition digital identifier and / or at least one value representative of the total amount of UV filters in the composition associated with at least one UV filter composition digital identifier.

7. Computer-implemented method (100) according to one of claims 1 to 6, which comprises: -a step of assembling (105) a UV filter composition with at least oneingredient, -a step of calculating (110) at least one UV spectrum of the assembled UVfilter composition or at least one UV spectrum of a combination of one or more ingredients,- a step of empirically measuring (115) an SPF value of the assembled UVfilter composition, and -a step of storing (120), in a database, at least one empirically measuredSPF value, at least one calculated UV Spectrum in association with at least one UV filter composition digital identifier representative of the assembled UV filter composition, wherein said database is used during the step of providing (135) a training set.

8. Computer-implemented method (100) according to claim 7, further comprising a step of pre-processing (125) the training set, and wherein the pre-processed training set is used during the step of providing (135) a training set.

9. Computer-implemented method (100) according to one of claims 1 to 8, wherein the trained predictive machine-learning model is a Random Forest Regressor.

10. Computer-implemented method (100) according to one of claims 1 to 8, wherein the trained predictive machine-learning model is a Gradient Boosting Regressor.

11. Computer-implemented method (100) according to one of claims 1 to 8, wherein the trained predictive machine-learning model is a Neural Network.

12. Computer-implemented method (100) according to one of claims 1 to 11, which further comprises a step of sending (160) a digital command representative of an instruction of materialising at least one UV filter composition corresponding to at least one UV filter composition digital identifier defined.

13. Computer-implemented method (100) according to claim 12, which further comprises a step of materialising (165) at least one UV filter composition corresponding to at least one UV filter composition digital identifier defined.

14. Computer-implemented method (130) to train a predictive machine learning model to provide at least one value associated to an Ultraviolet (UV) filter composition digital identifier, representing a materializable UV filter composition, characterized in that it comprises: -a step of providing (135) an original set of exemplar UV filter compositiondigital identifiers, said set of exemplar UV filter composition digital identifiers being representative of materialized UV filter compositions, comprising: -at least one value representative of an existing ingredient or at leastone value representative of a combination of existing ingredients, each existing ingredient being associated with an ingredient digital identifier, -at least one value representative of at least one calculated UVspectrum associated with said UV filter composition digital identifiers, and -at least one value representative of an empirically measured SPFvalue associated with said UV filter composition digital identifiers, to form a training set and -a step of training (140) a predictive machine-learning model using thetraining set, wherein the predictive machine-learning model is trained to predict at least one SPF value.

15. Computer program product (200) characterized in that it comprises instructions which upon execution by a computer cause the computer to execute the method(100) according to any one of claims 1 to 13.

16. Computer-readable storage medium (215) storing programming instructions which upon execution by a computer cause the computer to execute the method(100) according to any one of claims 1 to 14.

17. Device (200) for providing at least one predicted Sun Protection Factor (SPF) value associated to an Ultraviolet (UV) filter composition digital identifier,representing a materializable UV filter composition, associated with at least one ingredient digital identifier representative of an existing ingredient of the materializable UV filter composition, characterized in that it comprises: -a means of providing an original set of exemplar UV filter composition digitalidentifiers, said set of exemplar UV filter composition digital identifiers being representative of materialized UV filter compositions, comprising: -at least one value representative of an existing ingredient or at leastone value representative of a combination of existing ingredients, each existing ingredient being associated with an ingredient digital identifier, -at least one value representative of at least one calculated UVspectrum associated with said UV filter composition digital identifiers, and -at least one value representative of an empirically measured SPFvalue associated with said UV filter composition digital identifiers, to form a training set and -a means of training a predictive machine-learning model using the trainingset, wherein the predictive machine-learning model is trained to predict at least one SPF value, -a means of defining at least one UV filter composition digital identifier,- a means of predicting at least one SPF value using the trained predictivemachine-learning model and -a means of providing the at least one predicted SPF value.

Images