Computer-implemented method for assessing a compatibility of ingredients of a cosmetic composition

Abstract

In a computer-implemented method for assessing a compatibility of at least two ingredients of a cosmetic composition a list (1) of the at least two ingredients is requested. A first database (10) storing a first set (11) of compositions is accessed, wherein for each composition (12.1, 12.2, 12.3, 12.4, 12.5) of the first set (11) of compositions a corresponding compatibility rating (13.1, 13.2, 13.3, 13.4, 13.5), obtained from laboratory tests, is stored. A first score (s1) is obtained from a first comparison of the list (1) with the first set (11) of compositions to identify a first set of matches and from the compatibility rating (13.1, 13.2, 13.3, 13.4, 13.5) of the matches in the first set (11) of matches. A second database (20) is accessed, storing a second set (21) of compositions of cosmetic products available on the market, and a second score (s2) is obtained from a second comparison of the list (1) with the second set (21) of compositions to identify a second set of matches. A measure for compatibility of the at least two ingredients is determined based on the first score (s1) and the second score (s2). The method provides a quick, non-expensive way of preliminary testing of ingredients' compatibility.

Description

[0001] Computer-implemented method for assessing a compatibility of ingredients of a cosmetic composition

[0002] Technical Field

[0003] The invention relates to a computer-implemented method for assessing a compatibility of at least two ingredients of a cosmetic composition. It further relates to a computer program for carrying out the method, a method for assessing a compatibility of at least two ingredients of a cosmetic composition and a method for the manufacture of a cosmetic composition.

[0004] Background Art

[0005] Compositions of cosmetic products such as creams, ointments, tinctures, shampoos, sunscreen, toothpaste, perfumes, deodorants, makeup, etc., are complex and typically include a rather large number of ingredients fulfilling different purposes, such as solvents, active ingredients, emulsifiers, colorants, fragrance, etc. Therefore, the formulation of cosmetic products is a challenging process. The requirements that have to be fulfilled by cosmetic products include inter alia long-term stability and unimpaired effect of the active substances. Often, these requirements cannot be fulfilled by a potential composition because two or more ingredients of the composition are mutually incompatible. Therefore, it is one of the tasks of the cosmetic formulation developers to ensure that the ingredients used in a composition are compatible with each other.

[0006] The compatibility may be assessed based on laboratory tests, including long-term tests. These tests are expensive and time-consuming. Therefore, in an early development ("exploration") phase, they are avoided as possible. Nevertheless, potential incompatibilities should be taken into account as early as possible in order to avoid development work that turns out to having been useless at a later stage. Today, cosmetic formulation developers have some tools available to predict if ingredients can be formulated together without having a destabilizing effect in the formulation. In particular, the Hansen solubility parameters (HSP) approach (C. Hansen, The Three Dimensional Solubility Parameter and Solvent Diffusion Coefficient and Their Importance in Surface Coating Formulation. Danish Technical Press, Copenhagen, 1967) allows assessing if ingredients will be compatible based on different physicochemical parameters of the ingredients. However, this approach does not take into account all potential causes for incompatibilities. Furthermore, millions of combinations of cosmetic ingredients are possible, and using the HSP approach would require significant amounts of time and resources to assess all physicochemical parameters that enable the prediction of ingredient compatibility.

[0007] There is therefore a need for improved methods for assessing the compatibility of ingredients of cosmetic compositions, in particular in an early development stage.

[0008] Summary of the invention

[0009] It is the object of the invention to create a method pertaining to the technical field initially mentioned, that provides an improved assessment of the ingredients' compatibility.

[0010] The solution of the invention is specified by the features of claim 1. According to the invention, the computer-implemented method comprises the steps of: requesting a list of the at least two ingredients; accessing a first database storing a first set of compositions, wherein for each composition of the first set of compositions a corresponding compatibility rating, obtained from laboratory tests, is stored, wherein a first score is obtained from a first comparison of the list with the first set of compositions to identify a first set of matches and from the compatibility rating of the matches in the first set of matches; accessing a second database storing a second set of compositions of cosmetic products available on the market, wherein a second score is obtained from a second comparison of the list with the second set of compositions to identify a second set of matches; determining a measure for compatibility of the at least two ingredients based on the first score and the second score.

[0011] The list of the at least two ingredients may be provided by different means, e. g. entered in a (graphical) user interface, read from a database, provided in a data file or through an application programming interface (API). More than one list and lists from different sources may be provided and processed in parallel and / or in succession.

[0012] The information on the ingredients, in particular the designation of the ingredients, may be provided using different nomenclatures, in particular:

[0013] INCI names (i. e. using the unique identifiers according to the International Nomenclature of Cosmetic Ingredients); common names;

[0014] CAS Registry Numbers; and / or proprietary names of one or several providers of such ingredients.

[0015] The method may include the step of translation between the different nomenclatures, e. g. based on a correspondence table between the different formats / nomenclatures. In principle, all ingredients, irrespective of their source (list, databases) may be translated to a designated format, e. g. INCI, and the comparisons may subsequently be based on this designated format. Alternatively, translation between the formats is included in the comparison steps, if required. Alternatively or in addition, the used data is required to be in one designated format or in one of a group of designated formats.

[0016] The databases may be stored locally or accessed through a network, such as the internet. The first database and second database may include two or more information repositories that may be stored at different locations and be accessed through different channels. The terms "first database" and "second database" have to be understood functionally: Technically, the data constituting the "first database" and the "second database" may be stored in a single database and be distinguished by the stored content, e.g. by a flag indicating the function of a specific record.

[0017] The information in the first database is based on rigorous laboratory tests for stability. In particular, the first database may store compositions where no incompatibility issues have been found. In this case, the compatibility rating of all stored compositions may be equal ("compatible"), and, in principle, does not need to be recorded for each of the stored compositions but storage in the first database means that the this compatibility rating is assigned to each of the stored compositions. More specific compatibility ratings are possible, e. g. "fully compatible" if the substances in a composition are compatible irrespective of their amounts, as well as one or several "partly compatible" ratings, where there are some restrictions with respect to the (relative) amounts of the substances and / or minor compatibility issues. These restrictions may be stored in the first database as well.

[0018] The compatibility rating may relate in particular to a selection from the following properties of a mixture of two or more ingredients: physical stability; breakdown of the mixture; colour stability; development of odors; chemical degradation; inactivation (i. e. detrimental effects to active substances).

[0019] Some of these properties may relate to the development of a mixture under the influence of time, temperature change, UV radiation, humidity change, mechanical impact, etc.

[0020] In preferred embodiments, at least two of these properties are taken into account for the compatibility rating. More preferably, a composition may be considered stable when preferably two or more, more preferably all of the following criteria are fulfilled in combination: formulation stable at 40 °C for three months (no phase separation); no big colour difference (Delta E > 2) from visual examination (at all conditions); no odor issues (at all conditions); no substantial viscosity changes.

[0021] The skilled person will be aware that the term Delta E or AE can be used to describe colour differences in the so-called CIELAB colour space, as defined by the Commission Internationale de L'Eclairage (CIE, or translated to English, International Commission on Illumination, or ICI). Preferably the colour difference Delta E is equal to or less than 2.

[0022] The compatibility rating of a composition may be a single value representing the selected properties, or it may be a vector including several values each representing one of the selected properties (or subsets of the selected properties). For example, from the first database storing a first set of compositions, for each composition of the first set of compositions a corresponding compatibility rating, obtained from laboratory tests, is stored, wherein such compatibility rating can be a single value representing one specific selected property or it can be a vector, including a value for two or more selected properties. The second database contains information on compositions of cosmetic products that are available on the market. Lists of the ingredients of such compositions are available. In principle, most of the information is publicly available, as manufacturers are obliged to provide this information with respect to products that are put on the market. Furthermore, specialized database providers offer access to databases where such information is systematically compiled.

[0023] The first and the second score as well as the measure may be numerical values in a continuous bounded or unbounded range, values from a discrete set of values, assignments to one of several classes, etc. The scores and the measure may be scalar or vectorial quantities.

[0024] Different approaches may be chosen for the determination of the first score, the second score and the measure for compatibility as will be discussed in more detail below.

[0025] The inventive method may be carried out by a computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method. The computer program may be run on a single computer or comprise several modules that interact with each other and are run on two or more computers connected to each other for the exchange of data.

[0026] Therefore, an inventive method for assessing a compatibility of at least two ingredients of a cosmetic composition, comprises the steps of: performing laboratory tests to obtain compatibility ratings for each of a first set of compositions; storing the first set of compositions and the associated compatibility ratings in a first database; requesting a list of the at least two ingredients; accessing the first database and obtaining a first score from a first comparison of the list with the first set of compositions to identify a first set of matches and from the compatibility rating of the matches in the first set of matches; accessing a second database storing a second set of compositions of cosmetic products available on the market, wherein a second score is obtained from a second comparison of the list with the second set of compositions to identify a second set of matches; determining a measure for compatibility of the at least two ingredients based on the first score and the second score.

[0027] As a matter of course, a comprehensively documented laboratory test under known and controlled conditions provides the most reliable information on the compatibility of ingredients in a (candidate) cosmetic composition. Eventually, when developing a product, such laboratory tests will have to be performed. However, performing such laboratory tests in an early phase ("exploration phase") of product development is very costly. Therefore, there is a need for (preliminary) information on the expected compatibility of ingredients of potential cosmetic compositions. Using the inventive method, the available information on compatibility is systematically exploited: Where available, relevant laboratory tests are taken into account; in addition, or as a fallback position, it is checked whether there are existing compositions including the same ingredients (or subsets thereof, see below) on the market, as this provides an indication that the ingredients may be compatible.

[0028] Therefore, the inventive method provides a quick, non-expensive way of preliminary testing of ingredients' compatibility. In addition, the results of the inventive method may provide indications pointing to alternative formulations and / or the use of the right conditions when a specific combination is known to have issues. Due to the fact that the number of cosmetic products available on the market is constantly increasing, the results of the inventive method will constantly improve, provided that the second database is constantly updated. Similarly, the first database may be constantly updated as soon as new results relating to laboratory tests become available.

[0029] A method for the manufacture of a cosmetic composition may comprise the steps of: assessing a compatibility of at least two ingredients of the composition using the computer-implemented method according to the invention, suitably providing a measure for compatibility of the at least two ingredients; manufacturing the composition, if the measure for compatibility exceeds a predetermined threshold.

[0030] Accordingly, the invention includes a method for the manufacture of a cosmetic composition including the steps of performing the laboratory tests, assessing the compatibility and manufacturing the composition based on the assessed compatibility. If the assessed compatibility indicates a high compatibility, the manufacture may be initiated, if the assessed compatibility indicates a questionable compatibility, further assessments may be done (e. g. dedicated laboratory tests for the composition), and if the assessed compatibility indicates lacking compatibility the composition may be changed and another assessment may be initiated. This process may be repeated until a satisfactory composition is found.

[0031] The invention further includes a cosmetic composition that was manufactured using the inventive method for the manufacture of cosmetic compositions.

[0032] In a preferred embodiment, the second score is obtained from a number of matches in the second set of matches. In the simplest case, for determining the second score it is checked whether the number exceeds a certain threshold, e.g. 1'000 existing compositions. The threshold may be set differently for different product groups, taking into account the total number of available compositions in the respective product group. Furthermore, the availability of the individual ingredients may be taken into account: If one of the ingredients is a comparatively new product the number of available compositions including this ingredient might be still rather small even if the compatibility was not critical. Accordingly, in preferred embodiments, the date of introduction of a product is obtained from a database and be taken into account when determining the second score. In principle, the date may be obtained indirectly by consulting the second database, searching for the earliest composition including a certain ingredient, or directly by consulting a dedicated database storing the dates of introduction of individual substances.

[0033] Similarly, the number of matches may also be a valuable criterion when obtaining the first score, in particular when the results of the matching laboratory tests do not relate to the entire list of ingredients but only to a subset thereof.

[0034] In an alternative embodiment, other criteria than the number of matches are used to obtain the second score. As an example, the maximum number (or proportion) of matching ingredients may be a starting point for the second score.

[0035] In preferred embodiments, the inventive method further comprises the steps of: accessing a third database storing a third set of compositions, wherein for each composition of the third set an incompatibility information is stored; obtaining a third score from a third comparison of the list with the third set of compositions to identify a third set of matches; determining the measure for compatibility based in addition on the third score. The incompatibility information may be obtained e.g. from the literature, general experience, commercially or publicly available databases and / or (further) laboratory tests.

[0036] The first and the third database may be combined, wherein the stored compositions have different compatibility ratings ("compatible" or "non compatible", respectively, in the most simple case). Different orders of the consultation of the databases are possible. In particular, in one preferred embodiment, the third database is consulted first, and if an incompatibility is detected the measure for compatibility may be set to "incompatible" without having to access the first and second database.

[0037] Results obtained from the third comparison and in particular information stored in the third database related to the third set of compositions (or a subset thereof) may be provided to the user, providing specific information on the issues to be expected and potentially about ways for avoiding or mitigating these issues.

[0038] Preferably, the method comprises the further step of issuing a warning if a match in the third set of matches is assigned to an incompatibility level exceeding a threshold incompatibility level. Again, in the simplest case, a warning is issued if a single match is found in the third database, indicating that the ingredients of the composition are incompatible.

[0039] Preferably, the measure for compatibility is an assignment to one of at least three compatibility classes. As mentioned, the main purpose of the inventive method is to provide a tool to support the exploration of new cosmetic formulations. In any case, laboratory tests will be performed prior to deciding about commercially using a formulation. Therefore, the user is interested in obtaining an indication about whether the exploration with respect to a certain combination of ingredients shall be continued. There may be clear-cut cases where there are no indications about an incompatibility whatsoever ("compatible") or where an incompatibility between the ingredients is manifest ("incompatible"). However, there are many cases where the question of compatibility depends on further factors (such as concentrations, the presence of further ingredients, etc.) or where there is only a partial incompatibility (e. g. with respect to a change of colour at UV impact or high temperatures) that may be tolerable in certain applications. Finally, there may be cases where the assessment does not provide conclusive information on compatibility or non-compatibility. Therefore, it may be advantageous to have a further class between the "compatible" and "incompatible" class.

[0040] In a preferred embodiment, the at least three compatibility classes include at least the following classes: a) a high compatibility class, wherein the at least two ingredients of compositions in the high compatibility class have been found to be compatible in the laboratory tests, such that the first score exceeds a first threshold; b) a medium compatibility class, wherein compositions in the medium compatibility class do not fulfill requirements for assignment to the high compatibility class and wherein at least two ingredients of compositions in the medium compatibility class have been identified in a certain minimum number of cosmetic products available on the market, such that the second score exceeds a second threshold; c) a low compatibility class if requirements for assignment to the high compatibility class and the medium compatibility class are not fulfilled.

[0041] It may be required for an assignment to the high compatibility class that the respective composition has been found compatible in a certain minimum number of laboratory tests.

[0042] Further classes are possible, depending on the first and second score and optionally the third score. For each of the scores taken into account one or several threshold values may be defined. As an example, the following classification may be done:

[0043] The values si, s?, S3 denote the respective first, second and third score, whereas the values ti, ti. min, tz, t3 and t3, min denote thresholds for the respective score.

[0044] Other classifications with less or more classes are possible. The conditions on the scores (and potentially further properties) may be chosen differently. Preferably, the list of the at least two ingredients includes quantity information relating to the at least two ingredients and the quantity information is used for determining the measure for compatibility.

[0045] In particular, the quantity information is information relating to amounts, concentrations, ratios or similar. This information may be taken into account when obtaining the first score, the second score and / or the third score and / or in the subsequent step of determining the measure for compatibility based on the scores. Ingredients that do not exceed a certain minimum threshold may be disregarded in some or all of the steps. The threshold may be defined individually for each substance or for classes of substances. Alternatively or in addition, the first, second and / or third database may include reference quantity information, and the quantity information relating to the cosmetic composition may be compared with this reference quantity information when obtaining the corresponding score. Preferably, for obtaining the second score an order of substances listed for a composition in the second set of matches is taken into account.

[0046] Usually, in available product information relating to compositions of substances, the substances with the highest concentration are listed first, which means that the existence of a composition on the market including the ingredients of the tested cosmetic composition among the top substances of the existing, available composition is a stronger indication of these ingredients' compatibility than the mere presence of the ingredients among those listed further down the list, with a lower concentration. Therefore, when determining the second score, compositions in the second set of matches featuring the relevant ingredients among their components listed in a first section may be given a larger weight than compositions in the second set of matches featuring the relevant ingredients (or at least some of them) only in a later section of the product description contained in the second database.

[0047] Alternatively, the order of substances listed in the compositions in the second set of matches is not taken into account. This may be reasonable under the assumption that similar ingredients will usually be used in similar amounts in cosmetic compositions, e.g. the amount of solvents will always be substantially higher than that of emulsifiers or even colorants or fragrances.

[0048] In preferred embodiments of the invention, for the first comparison and / or the second comparison subsets of the list of the at least two ingredients with different cardinalities are compared with the first set of compositions and / or the second set of compositions, respectively.

[0049] Accordingly, in different embodiments of the invention, subsets of the list of the at least two ingredients may be employed as follows: i) for the first comparison such subsets with different cardinalities are compared with the first set of compositions; ii) for the second comparison such subsets with different cardinalities are compared with the second set of compositions; or iii) for the first comparison such subsets with different cardinalities are compared with the first set of compositions and for the second comparison such subsets with different cardinalities are compared with the second set of compositions, wherein the same or different subsets may be used for both the comparison with the first set of compositions and the comparison with the second set of compositions.

[0050] The cardinality denotes the number of ingredients in a given (sub)set. The largest subset (highest cardinality) may correspond to the entire list of the at least two ingredients. The smallest subsets (lowest cardinality) may include two ingredients of the list. In principle, the matching process may start with the largest subset (i. e. the entire list) and continue with subsets of decreasing cardinality. This process may be stopped as soon as a sufficient number of matches have been identified or at a predetermined minimum cardinality.

[0051] Preferably, the first set of matches and / or the second set of matches is a selection of compositions of the first set of compositions or the second set of compositions respectively wherein each composition of the selection contains all ingredients of at least one subset of the list of ingredients.

[0052] In particular, the selection of compositions is a list that is smaller or of equal size as the first set of compositions or the second set of compositions respectively.

[0053] A matching with the above method has the advantage of being easily applicable and computationally fast while delivering reliable and transparent results. Alternatively, a match might be chosen to be a composition that contains ingredients that are chemically similar to the ingredients of the subset, e.g. with a correspondingly trained artificial neural network. Preferably, for obtaining the first score and / or for obtaining the second score matches in the first set of matches and / or the second set of matches are weighted according to a cardinality of the respective subset.

[0054] Alternatively or in addition, the cardinality of the subset may be a factor in the matching process. As an example, the matching is concluded if a match or a certain number of matches have been found for a given cardinality, and the subsets with lower cardinality are not compared with the respective set of compositions.

[0055] Preferably, for obtaining the first score and / or for obtaining the second score, a sum total is obtained, which sum total stems from a summation of a set of individual weighting factors corresponding to the matches of the first set of matches or the second set of matches respectively.

[0056] In particular, each individual weighting factor is a product of several factors (see below) corresponding to one match of the first or the second set of matches.

[0057] Alternatively, the first or the second score may be chosen to correspond to e.g. the cardinality of the first set of matches or the second set of matches respectively.

[0058] In a preferred embodiment of the invention, obtaining the first score and / or obtaining the second score includes the steps of: obtaining the set of individual weighting factors by, for each match of the first set of matches or the second set of matches respectively, multiplying a cardinality of a largest subset that matches the respective match with at least one further weighting factor corresponding to the match; obtaining the first score or the second score respectively from normalizing the sum total or each individual weighting factor of the set of individual weighting factors with a normalization factor. For calculating the first score, the further weighting factor may in particular represent a corresponding compatibility rating of the match (which may be chosen to be 1 for all compatible compositions). For calculating the second score, the further weighting factor may in particular respresent the trust level (which may be chosen to be 1 for all compatible compositions if no specific information for the trust level is available). Another additional weighting factor may depend on whether it is known that an order of the constituents in the corresponding match corresponds to the respective amounts of ingredients in the list of at least two ingredients. This weighting factor preferably has a fixed value if no order information is available for the composition. If order information is available, the value preferably is calculated based on the Levenshtein distance (Levenshtein, Vladimir I. (February 1966). "Binary codes capable of correcting deletions, insertions, and reversals". Soviet Physics Doklady. 10 (8): 707-710) between two strings that represent the order of the ingredients in the list of at least two ingredients and the respective match in the second database.

[0059] Alternatively, as stated above, the first score or the second score may be chosen to correspond to e.g. the cardinality of the first set of matches or the second set of matches respectively.

[0060] A trust rating may be assigned to each of the compositions of the first set of compositions and / or to each of the compositions of the second set of compositions, and the trust rating is considered when obtaining the first score and / or the second score.

[0061] The trust levels may be assigned based on different criteria. For example, with respect to the first database, trust levels may relate e.g. to the extent of the laboratory tests that have been performed with respect to a certain composition. With respect to the second database, some brands offering products on the market may be considered more reliable than others and the trust levels may be chosen accordingly. In one embodiment, the methods according to the invention are carried out with the help of artificial intelligence and / or machine learning.

[0062] In particular, artificial intelligence or machine learning may be applied in order to execute one or more of the following tasks: formulate queries in a database query language (such as SQL) for accessing the respective database; classifying the cosmetic composition and / or identifying the first, second or third set of matches, respectively, e. g. by clustering compositions in the first, second or third set and assigning the cosmetic composition to the most similar cluster; obtaining the first, second and / or third score from the data of the respective database; determine the measure for compatibility based on the first, second and potentially third score; assigning the measure for compatibility to a compatibility class.

[0063] Several models may be used, including supervised machine learning (e. g. deep learning), semi-supervised or unsupervised machine learning (e. g. for clustering), predictive queries (for accessing the databases), etc.

[0064] In a preferred embodiment of the computer-implemented method, the first score is obtained by processing data from the first database using an artificial intelligence and / or machine learning step and / or the second score is obtained by processing data from the second database using an artificial intelligence and / or machine learning step.

[0065] Similarly, the third score may be obtained by processing data from the third database using an artificial intelligence and / or machine learning step. In a particularly preferred embodiment, a supervised machine learning model is trained using the data of the first, second or third database, using the first, second or third set of compositions as features and the corresponding compatibility ratings as labels. The trained model will then be fed an input vector representing a composition the compatibility of which is to be assessed and generate a first, second or third compatibility rating for this composition.

[0066] In the training and / or application of an artificial intelligence or machine learning model, data from more than one of the first, second and third database may be combined. As well, different approaches for processing the data of the individual databases may be chosen.

[0067] Other advantageous embodiments and combinations of features come out from the detailed description below and the entirety of the claims.

[0068] Brief description of the drawings

[0069] The drawings used to explain the embodiments show:

[0070] Fig. 1 a schematic representation of an embodiment of the inventive method for assessing the compatibility of ingredients of a cosmetic composition;

[0071] Fig. 2 a flow chart of the inventive method; and

[0072] Fig. 3 a schematic representation of the generation of subsets for the matching steps.

[0073] In the figures, the same components are given the same reference symbols. Preferred embodiments

[0074] The Figure 1 is a schematic representation of an embodiment of the inventive method for assessing the compatibility of ingredients of a cosmetic composition. First, three databases 10, 20, 30 are provided.

[0075] The first database 10 stores a first set 11 of compositions 12.1, 12.2, 12.3, 12.4 etc. Usually, the compositions in the first set 11 feature different numbers of substances, ranging from two to potentially a large number of constituents. A compatibility rating 13.1, 13.2, 13.3, 13.4 etc. is assigned to each of the compositions 12.1, 12.2, 12.3, 12.4 etc. The compositions in the first set 11 and the assigned compatibility ratings 13 represent the results of laboratory tests, wherein the compatibility of the constituents of the respective compositions 12 has been tested by preparing the compositions, storing them for at least four weeks and testing the compositions for a potential phase separation, colour difference, odor issues and viscosity changes. Based on the results, the corresponding compatibility rating 13 is determined and stored in the database together with a list of the constituents and their amounts.

[0076] The second database 20 stores a second set 21 of compositions 22.1, 22.2, 22.3, 22.4 etc. Again, also in the second set 21 the compositions usually feature different numbers of substances. The information in the second database 20 has been obtained from commercially available databases and represents cosmetic formulations (cosmetic products) that are available on the market. The number of constituents of the compositions 22 may be large. For some of the compositions 22.1, 22.2, 22.3, it is known that the order of the constituents in the record corresponds to the respective amounts in the composition and a corresponding flag 24.1, 24.2, 24.3 assigned to the respective composition 22.1, 22.2, 22.3 is set in the database. For other compositions 22.4 this information is not available and therefore the flag is not set. Furthermore, a trust level 23.1, 23.2, 23.3, 23.4 is assigned to each of the compositions, the trust level 23 relating to the manufacturer of the available composition, the provider of the data, the sold volume of the composition, the time that it has been available on the market and / or the official authorizations that have been obtained for the respective product etc.

[0077] The third database 30 stores a third set 31 of compositions 32.1, 32.2, 32.3, 32.4 etc. These compositions are known to have issues with compatibility. Usually, the number of constituents of the compositions 32 will be small, often only two or three. Further information 33.1, 33.2, 33.3, 33.4 on the specific incompatibility is assigned to each of the compositions 32.1, 32.2, 32.3, 32.4.

[0078] For obtaining a measure for compatibility for a given (candidate) composition, a list 1 of the corresponding ingredients is compared to the three databases 10, 20, 30, as described in more detail below. The comparison with the first database 10 yields a first score si, the comparison with the second database 20 yields a second score s?, and the comparison with the third database 30 yields a third score S3. Based on these scores the composition according to the list 1 is assigned to one of several classes 40.1, 40.2, 40.3, 40.4, 40.5. The classes are as follows:

[0079] Furthermore, the comparisons, in particular the one with the third database 30, yields further information 50 that may be provided to the user in text form. The following information may be provided:

[0080]

[0081] In the example shown in Figure 1, the comparison of the list 1 with the first database 10 yields several matches, including compositions 12.1, 12.3, 12.4 (marked by hatching). The comparison of the list 1 with the second database 20 yields several matches, including compositions 22.2, 22.3. The comparison with the list 1 with the third database 30 yields a single match with composition 32.3. Finally, the composition according to the list 1 is assigned to the second class 40.2.

[0082] The Figure 2 is a flow chart of the inventive method. In a first preparatory step 101, laboratory tests are performed and the results thereof are stored in the first database 10. Access to the further two databases 20, 30 is provided.

[0083] The user provides the list of ingredients of the composition to be assessed, e.g. by entering the corresponding information in a user interface of the used computing system (step 102). The list may contain information on the absolute or relative amounts of the ingredients in the composition. Furthermore, the list may be automatically pre- processed, e. g. by removing all ingredients that do not exceed at minimum threshold, wherein the threshold may be fixed to e. g. 1 vol.% or take different values depending on the class of ingredients.

[0084] As well, depending on the nomenclature used when providing the list (common name, proprietary name, CAS, etc.), the entries in the list may be translated to a designated nomenclature, e. g. INCI. Next, the third database 30 is accessed (step 111) and the combination of ingredients defining the composition and subsets thereof are compared with the entries in the third database 30 (step 112). Again, if needed, the entries in the third database may be translated to the designated nomenclature. Based on the comparison, a third score is obtained (step 113). The third score is compared with a threshold (step 114). If a minimum threshold is not reached, a serious incompatibility has been detected and a corresponding notice is displayed, including information on the detected incompatibility (step 142).

[0085] If the minimum threshold is reached, the assessment is continued. The first database 10 is accessed (step 121) and the combination of ingredients defining the composition (and subsets thereof) are compared with the entries in the first database 10 (step 122). Again, if needed, the entries in the first database may be translated to the designated nomenclature. Based on the comparison, a first score is obtained (step 123).

[0086] Next, the second database 20 is accessed (step 131) and the combination of ingredients defining the composition (and subsets thereof) are compared with the entries in the second database 20 (step 132). Again, if needed, the entries in the second database may be translated to the designated nomenclature. Based on the comparison, a second score is obtained (step 133).

[0087] Based on the three scores, the measure for compatibility is calculated (step 141). This measure as well as additional information is displayed to the user (step 142). Based on the displayed information, the user decides whether the compatibility is sufficient (step 150). If so, the product according to the composition may be eventually manufactured (step 160). If not, a further composition may be defined and provided (step 102) and the process is repeated.

[0088] The Figure 3 is a schematic representation of the generation of subsets for the matching steps. In the example, the list 1 of ingredients of a potential cosmetic composition includes five substances A, B, C, D, and E. In the matching procedures with the first, second and third database, the following protocol may be followed:

[0089] 1. In a first stage, the entire list 1, including five substances, is compared to each of the compositions in the respective database. Matches are those compositions that include all the five substances (and potentially further substances).

[0090] 2. In a second stage, subsets 2.1, 2.2, 2.3, 2.4, 2.5 of four out of the five substances are compared to each of the compositions in the respective database. In total, there are five such subsets in the shown example.

[0091] 3. In a third stage, subsets 3.1, 3.2, 3.3, 3.4, 3.5 etc. of three out of the five substances are compared to each of the compositions in the respective database. In total, there are ten such subsets in the shown example.

[0092] 4. In a fourth stage, subsets 4.1, 4.2, 4.3, 4.4, 4.5 etc. of two out of the five substances are compared to each of the compositions in the respective database. Again, there are ten such subsets in the shown example. In an exemplary embodiment, the scores si, s?, S3 are obtained as follows: a) A number of subsets is generated for the list 1 of ingredients. For a list having n members, these subsets include all subsets of ingredients that comprise at least 1 / 3 and in any case at least two ingredients, i. e. the smallest subsets that are taken into account for a list with n members have the following number of ingredients: b) The subsets are compared to all the compositions 12 in the first database 10 (laboratory tests). A match is identified if all ingredients of at least one given subset can be found in a composition. c) Based on the matches with the compositions in the first database, the first score si is calculated as follows: where n denotes the number of ingredients in the list 1, m, denotes the size of the largest subset matching composition i and wtis a weighting factor representing the compatibility rating 13 (which may be chosen to be 1 for all compatible compositions). Therefore, the score increases with an increasing number of matches, wherein matches with a larger number of corresponding ingredients contribute more than the matches with smaller subsets. d) The same subsets are used for the matching with the compositions in the second database 20 (products available on the market). e) Based on the matches with the compositions in the second database, the second score S2 is calculated as follows: where n denotes the number of ingredients in the list 1, m, denotes the size of the largest subset matching composition i and is a weighting factor respresenting the trust level 23 (which may be chosen to be 1 for all compatible compositions). The further weighting factor o depends on whether it is known that the order of the constituents in the corresponding composition corresponds to the respective amounts of ingredients in the composition (indicated by the flag 24). The weighting factor oi has a fixed value of no order information is available for the composition. If order information is available, the value is calculated based on the Levenshtein distance (Levenshtein, Vladimir I. (February 1966). "Binary codes capable of correcting deletions, insertions, and reversals". Soviet Physics Doklady. 10 (8): 707- 710) between two strings that represent the order of the ingredients in the list 1 and the respective composition in the second database 20. Again, the score increases with an increasing number of matches, wherein matches related to trustworthy sources as well as those with a similar order of the ingredients' amounts contribute more than the others. f) For the matching with the compositions in the third database 30 (incompatibility), all size 2 subsets of the list 1 are compared to the combinations in the database having two ingredients. If the database includes compositions with more than two ingredients, the correspondingly sized subsets of the list 1 are compared to those as well. Matches point to incompatibilities between constituents of the list 1. If no match is detected, the value of S3 is set to 1. If a serious match is detected, the score S3 is set to 0. For each potential ("non-serious") incompatibility, the initial value of 1 is multiplied with a factor in the range of 0 to 1 (not including the boundaries), depending on the type of incompatibility (in the most simple case, the factor is fixed at 0.5).

[0093] Access to the inventive method may be provided by a server-based application providing a user interface and / or API to exchange information with the user. The application may provide a functionality that information available in a user-side database (e.g. information relating to compatibility laboratory tests of the user) may be included in the analysis.

[0094] In a first specific example, the inventive method may proceed as follows:

[0095] 1. The user would like to know whether the ingredient PARSOL® SLX (proprietary name of DSM-Firmenich) is compatible with argan oil. Accordingly, the user provides a list including these two ingredients.

[0096] 2. First, the indications of the ingredients are both translated to INCI: PARSOL® SLX corresponds to Polysilicone-15, whereas argan oil corresponds to Argania Spinosa Kernel Oil.

[0097] 3. Next, the third database 30 (incompatibility) is consulted: No indication is found for an incompatibility of this combination of two ingredients.

[0098] 4. Next, the first database 10 (laboratory tests) is consulted. This yields four hits for formulations that both comprise Polysilicone-15 and Argania Spinosa Kernel Oil.

[0099] 5. Next, the second database 20 (products available on the market) is consulted. This yields 47 hits for products available on the market containing both Polysilicone-15 and Argania Spinosa Kernel Oil.

[0100] 6. Based on the number of hits in the first and second database 10, 20, the combination is classified into the medium compatibility I class (probably compatible). Accordingly, the client is informed that the ingredients are "Worth to combine." and the client is informed about the number of hits and provided with additional information on the relevant four products that underwent the laboratory tests, i. e. with a list as follows:

[0101] The combination of ingredients is present in the following DSM-Firmenich formulations: Smooth & Shine Instant Detangler (Hair Care)

[0102] BB cream conditioner (Hair Care)

[0103] - CURL STYLER (Hair Care)

[0104] Swip and See Color Changing lipstick (Sun Care)

[0105] In a second specific example, the inventive method may proceed as follows:

[0106] 1. The user would like to know whether the ingredients ALPHA-ARBUTIN and PARSOL® HS (proprietary name of DSM-Firmenich) are compatible. Accordingly, the user provides a list including these two ingredients.

[0107] 2. First, the indications of the ingredients are both translated to INCI: ALPHAARBUTIN is the INCI name, whereas PARSOL® HS is translated to Phenylbenzimidazole Sulfonic Acid.

[0108] 3. Next, the third database 30 (incompatibility) is consulted: No indication is found for an incompatibility of this combination of two ingredients.

[0109] 4. Next, the first database 10 (laboratory tests) is consulted. This yields no hits.

[0110] 5. Next, the second database 20 (products available on the market) is consulted. This yields only 6 hits for products available on the market containing both Alpha- Arbutin and Phenylbenzimidazole Sulfonic Acid.

[0111] 6. Based on the number of hits in the first and second database 10, 20, the combination is classified into the medium compatibility III class (no conclusive evidence). Accordingly, the client is informed that they should "Go carefully!" and the client is informed about the number of hits and with the information that no lab test results are available for this combination.

[0112] Two further examples illustrate the case where an incompatibility is listed in the third database 30: For the combination of PARSOL® 1789 (DSM-Firmenich proprietery name, corresponding to Butyl Methoxydibenzoylmethane) and Zinc Oxide, a certain incompatibility is known, depending on the final formulat and the respective amounts in the composition. This combination is assigned medium compatibility II (additional measures required to ensure stability) and corresponding information from the third database is displayed.

[0113] For the combination of Erythrulose (INCI: Erhytrulose, Aqua) with Potassium Sorbate it is known and listed in the third database 30 that these ingredients are chemically incompatible, leading to a discoloration. The combination is assigned low compatbility and a corresponding warning is displayed together with the available information on the incompatibility.

[0114] The user may be given the option to order specific laboratory tests for the assessed combination of ingredients. Whether this option is presented to the user may be dependent from the determine measure of compatibility (e. g. compatibility class). In particular, the option may be displayed if corresponding lab tests are not available at all, if the number of tests is lower than a given threshold and / or if a confidence value assigned to the available test(s) is lower than a given threshold.

[0115] The users' requests may be processed to identify the need for additional laboratory tests or for compositions that should be made available: In a first case, compositions being identified as being (probably) compatible may be determined as being candidates for future laboratory tests in order to improve the reliability of the inventive method. In a second case, compositions identified as being (probably) compatible may be formulated and offered to the customers; in a third case, alternatives to compositions that are identified as being (probably) incompatible may be identified. The method may include the step of suggesting potential alternatives based on this identification or other information. The invention is not restricted to the described embodiment. In particular, the details of the generation of subsets, the matching and the calculation of the scores may be chosen differently.

[0116] The determination of the first, second and / or third score may be dependent from each other. As an example, specific subsets may be matched with the first as well as with the second database and scores obtained from both matchings may be combined. This allows for compensating negative with positive results, i.e. if there are contradicting results with respect to compatibility originating from the matchings with different databases it may be concluded that the results with respect to that specific subset are not conclusive.

[0117] In summary, it is to be noted that the invention provides a method that allows for an improved assessment of the compatibility of ingredients of cosmetic compositions.

Claims

Claims1. Computer-implemented method for assessing a compatibility of at least two ingredients of a cosmetic composition, comprising the steps of: requesting a list of the at least two ingredients; accessing a first database storing a first set of compositions, wherein for each composition of the first set of compositions a corresponding compatibility rating, obtained from laboratory tests, is stored, wherein a first score is obtained from a first comparison of the list with the first set of compositions to identify a first set of matches and from the compatibility rating of the matches in the first set of matches; accessing a second database storing a second set of compositions of cosmetic products available on the market, wherein a second score is obtained from a second comparison of the list with the second set of compositions to identify a second set of matches; determining a measure for compatibility of the at least two ingredients based on the first score and the second score.

2. Computer-implemented method as recited in claim 1, wherein the first score is obtained by processing data from the first database using an artificial intelligence and / or machine learning step and / or the second score is obtained by processing data from the second database using an artificial intelligence and / or machine learning step.

3. Computer-implemented method as recited in claim 1 or 2, wherein the second score is obtained from a number of matches in the second set of matches.

4. Computer-implemented method as recited in one of claims 1 to 3, comprising the further steps of accessing a third database storing a third set of compositions, wherein for each composition of the third set an incompatibility information is stored; obtaining a third score from a third comparison of the list with the third set of compositions to identify a third set of matches; determining the measure for compatibility based in addition on the third score.

5. Computer-implemented method as recited in claim 4, comprising the further step of issuing a warning if a match in the third set of matches is assigned to an incompatibility level exceeding a threshold incompatibility level.

6. Computer-implemented method as recited in one of claims 1 to 5, wherein the measure for compatibility is an assignment to one of at least three compatibility classes.

7. Computer-implemented method as recited in claim 6, wherein the at least three compatibility classes include at least the following classes: a) a high compatibility class, wherein the at least two ingredients of compositions in the high compatibility class have been found to be compatible in the laboratory tests, such that the first score exceeds a first threshold; b) a medium compatibility class, wherein compositions in the medium compatibility class do not fulfill requirements for assignment to the high compatibility class and wherein at least two ingredients of compositions in the medium compatibility class have been identified in a certain minimum number of cosmetic products available on the market, such that the second score exceeds a second threshold;c) a low compatibility class if requirements for assignment to the high compatibility class and the medium compatibility class are not fulfilled.

8. Computer-implemented method as recited in one of claims 1 to 7, wherein the list of the at least two ingredients includes quantity information relating to the at least two ingredients and wherein the quantity information is used for determining the measure for compatibility.

9. Computer-implemented method as recited in one of claims 1 to 8, wherein for obtaining the second score an order of substances listed for a composition in the second set of matches is taken into account.

10. Computer implemented method as recited in one of claims 1 to 9, wherein for the first comparison and / or the second comparison subsets of the list of the at least two ingredients with different cardinalities are compared with the first set of compositions and / or the second set of compositions, respectively.

11. Computer implemented method as recited in claim 10, wherein the first set of matches and / or the second set of matches is a selection of compositions of the first set of compositions or the second set of compositions respectively wherein each composition of the selection contains all ingredients of at least one subset of the list of ingredients.

12. Computer implemented method as recited in claim 10 or 11, wherein for obtaining the first score and / or for obtaining the second score matches in the first set of matches and / or the second set of matches are weighted according to a cardinality of the respective subset.

13. Computer implemented method as recited in claim 12, wherein for obtaining the first score or the second score, a sum total is obtained, which sum total stems from asummation of a set of individual weighting factors corresponding to the matches of the first set of matches or the second set of matches respectively.

14. Computer implemented method as recited in claim 13, wherein obtaining the first score or the second score includes the steps of:- obtaining the set of individual weighting factors by, for each match of the first set of matches or the second set of matches respectively, multiplying a cardinality of a largest subset that matches the respective match with at least one further weighting factor corresponding to the match;- obtaining the first score or the second score respectively from normalizing the sum total or each individual weighting factor of the set of individual weighting factors with a normalization factor.

15. Computer implemented method as recited in one of claims 1 to 14, wherein a trust rating is assigned to each of the compositions of the first set of compositions and / or to each of the compositions of the second set of compositions and wherein the trust rating is considered when obtaining the first score and / or the second score.

16. Computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of one of claims 1 to 15.

17. Method for assessing a compatibility of at least two ingredients of a cosmetic composition, comprising the steps of: performing laboratory tests to obtain compatibility ratings for each of a first set of compositions; storing the first set of compositions and the associated compatibility ratings in a first database; requesting a list of the at least two ingredients;accessing the first database and obtaining a first score from a first comparison of the list with the first set of compositions to identify a first set of matches and from the compatibility rating of the matches in the first set of matches; accessing a second database storing a second set of compositions of cosmetic products available on the market, wherein a second score is obtained from a second comparison of the list with the second set of compositions to identify a second set of matches; determining a measure for compatibility of the at least two ingredients based on the first score and the second score.

18. Method for the manufacture of a cosmetic composition, comprising the steps of: assessing a compatibility of at least two ingredients of the composition using the computer-implemented method of one of claims 1 to 14 or the method of claim 17; manufacturing the composition, if the measure for compatibility exceeds a predetermined threshold.

19. Cosmetic composition, manufactured using a method for the manufacture of a cosmetic composition according to claim 18.

Images