Personalized skincare recommendation

Abstract

A system, apparatus and method are provided for generating personalized skincare recommendations. The system collects data reflecting a user's health, skin type, skin conditions, skincare goals, demographics, and other useful information. Based on the collected data, the system assembles a skin profile for the user, which is populated with attributes that reflect the data and that identify skin conditions and / or concerns to be addressed within a recommended skincare routine. Based on the attributes, the system applies a set of rules to identify skincare product ingredients and / or formulations known to help the user's type of skin, and generates the user's personalized routine to include products having those ingredients and / or formulation. The system may also identify ingredients and / or formulation that would aggravate the user's skin and ensures they are omitted from the routine.

Description

PERSONALIZED SKINCARE RECOMMENDATIONBACKGROUND

[0001] This disclosure relates to the fields of dermatology and computer systems. More particularly, a system, apparatus and methods are provided for generating personalized skincare and / or cosmetic recommendations based on characteristics of a user (e.g., skin type, allergies), available dermatological products (e.g., ingredients, effectiveness), and / or other factors.

[0002] Skincare, cosmetic, beauty, personal care and dermatologic products (collectively referred to herein as “products”) are in high demand. Product recommendations are often provided to consumers by retailers, beauty associates, manufacturers, dermatologists, other clinicians, social media influencers, blogs, websites, apps, and other sources. However, the bases for these recommendations are often unclear, and consumers may not be equipped to determine whether a given skincare product recommendation is suitable for them.

[0003] Thus, selecting an optimal product (or even just an effective product, while avoiding a potentially injurious product) can be a time-consuming and complicated process. Stores display myriad products of every type, overwhelming one’s ability to differentiate between them and causing one to rely upon recommendations of others, hasty research, and / or unproven claims included in advertisements. The plethora of available products makes it difficult to perform timely and quality research to identify a useful solution.

[0004] This problem arises from the fact that the sources of many product recommendations lack methods or means for efficiently, accurately, and transparently displaying product performance and risk characteristics to consumers on a personalized basis. Product characteristics are dependent on the ingredient composition of the product, but there are currently over 14,000 ingredients used in products, per the International Nomenclature of Cosmetic Ingredients (INCI).

[0005] A dermatologist may be consulted and can provide a wealth of information regarding the theory behind treating a skin condition (e.g., acne, dry skin) and / or maintaining healthy skin, but is unlikely to be familiar with all products available to all consumers. Abeautician or beauty advisor may be familiar with most or all available products, but is unlikely to know the theory needed to best treat a specific user for a specific condition or help her or him reach a specific goal. Influencers and / or other online personalities have wide reach, and may be versed in available products and / or the theory behind different skincare regimens, but cannot match theories and / or products to individual users having specific needs or conditions. Each of these entities is limited by the capacity of the human memory, and is unable to comprehensively leverage ingredient-specific insights and latest and best dermatological and health evidence on a personalized basis.

[0006] In short, generic advice to consumers regarding myriad products may be of little or no value to some consumers, either because of their type of skin, their allergies or other conditions, their budgets, and / or other factors. What is needed is a system, apparatus and method for helping a person identify beneficial products as well as products that will damage the person’s skin, trigger an allergy, or simply be ineffective.SUMMARY

[0007] In some embodiments, systems and methods are provided for making personalized skincare and / or cosmetic recommendations. In these embodiments, a system is programmed with detailed information regarding some or all available products (and new products as they are released), including their ingredients, as well as positive and negative matches between each ingredient or product and user attributes that correspond to skin or bodily characteristics of a user (e.g., skin traits and conditions, body concerns). Positive matches indicate that a given product or ingredient may ameliorate or otherwise benefit an attribute. Negative matches indicate that a given product or ingredient might endamage or worsen the attribute. Neutral matches may also be identified, to include products / ingredients that are not likely to have a positive or negative effect.

[0008] In some embodiments, for each consumer or other user searching for a skincare or cosmetic product, the system assembles a personalized profile of her skin type, skin conditions, allergies, goals, and so on, each of which is used to configure a corresponding user attribute. Meanwhile, products, ingredients, and interactions between the products / ingredients and human skin are researched to amass one or more comprehensive data repositories foridentifying products and / or product categories that match positively and / or negatively for specific users. Artificial intelligence, machine-learning models, and / or big-data statistical analysis methods may facilitate the matching process as well as the research process.

[0009] In some embodiments, a personalized product recommendation for a given user includes assembly of a routine (e.g., a daily skincare and / or cosmetic routine) based on the user’s skin profile (particularly her skin conditions and concerns), a professional’s prescribed concerns, and / or by linkage to other technologies such as patch tests, other skin tests, image analysis-based modalities, other modalities intended to collect information about a user’s skin, user profiles maintained by retailers (e.g., regarding historical purchase information), etc.

[0010] In other embodiments, methods are provided for: automatically and systematically evaluating published evidence in general skincare and dermatologic skincare, and using the evidence to define dynamic rules that can be used for individual product recommendations and routine building; assigning performance and risk characteristics to ingredients in skincare products, on a user-specific basis; creating and automatically updating personalized recommendation algorithms to a user; maintaining and updating a product database on which personalized recommendations are based; leveraging a community network of users towards alternate product selections and personalized skincare content delivery (e.g., based on their feedback regarding their routines); analyzing a skincare product on a userspecific basis to provide recommendations for use, or providing alternate products that are better matches for the user; tracking user data and applying automatic updates or transient modifications of personalized recommender algorithms and content delivery; collecting mass skincare data pertaining to use characteristics, shopping trends, social media trends, etc.; and identifying new skincare product commercial opportunities by way of methodologies described herein.DESCRIPTION OF THE FIGURES

[0011] Figure 1 is a block diagram of a system for generating personalized recommendations regarding skincare products, according to some embodiments.

[0012] Figure 2 is a flow chart illustrating a method of generating a personalized recommendation regarding skincare products, according to some embodiments.

[0013] Figure 3 illustrates sample rule and ingredient data table, according to some embodiments.

[0014] Figure 4 illustrates sample alerts for evaluating a specific product or a user’s routine (e.g., her current daily routine), according to some embodiments.DETAILED DESCRIPTION

[0015] The following description is presented to enable any person skilled in the art to make and use the disclosed embodiments, and is provided in the context of one or more practical applications and their requirements. Various modifications to the disclosed embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other embodiments and applications without departing from the scope of those that are disclosed. Thus, the present invention or inventions are not intended to be limited to the embodiments shown, but rather are to be accorded the widest scope consistent with the disclosure.

[0016] In some embodiments, systems and methods are disclosed for providing personalized recommendations for products for skincare, cosmetic, beauty, personal body care, and / or other dermatologic uses, including prescription topical products (collective referred to herein as ‘skincare products’ or simply ‘products’). The recommendations are based upon a recipient’s personal skin or body profile and existing research and unbiased evidence regarding the compositions of available products and the effectiveness of the products’ ingredients.

[0017] The term ‘products’ may encompass ingredients when referring to recommendations, interactions with human skin, and / or in other contexts, although in some contexts products and ingredients are addressed separately. For example, when providing a personalized recommendation to a user, the recommendation may identify specific ingredients and / or specific products (i.e., from specific manufacturers); in the former case, the user may use the recommendation to choose from all products containing the ingredients (and the system can identify those products to the user).

[0018] Beyond recommending ingredients and / or products that a specific user should (or should not) use, embodiments described herein also consider the amount and / or concentration of an ingredient. Similarly, the formulation of a product may affect a recommendation if, for example, evidence shows that the efficacy for the entire formulation differs from that of its individual component ingredients. Further, some entire product categories (e.g., cleansers) may be recommended (or omitted from recommendations) based on how they affect humans. For example, facial cleansers are often reported to be beneficial for people with “normal skin,” regardless of the ingredients known to be included in facial cleansers.

[0019] Much-needed assistance is thus provided to consumers facing a marketplace populated with thousands of skincare, cosmetic, and body care products (e.g., over 20,000 products) composed of thousands of different ingredients (e.g., over 14,000 ingredients), while hundreds of medically supported and validated recommendations may exist but be inaccessible or unintelligible to many of the consumers. Embodiments disclosed below may be deployed via mobile app, web app, application programming interfaces (APIs) within third-party websites or apps, in-store physical adjuncts (e.g., virtual reality (VR) headsets, augmented reality displays, shopping displays), and / or in other manners.

[0020] Systems, apparatus and methods are provided to facilitate development and implementation of personalized skincare product matching and routine building algorithms. Illustrative systems include a comprehensive and exhaustive collection of product data, drawn from over 300 sources, that identifies all ingredients used in all commercially available skincare products, and the effects and interactions of those ingredients.

[0021] Illustrative methods commence with collection of user-specific data encompassing factors such as skin type, skin / health concerns, preferences, and environmental / lifestyle factors, in order to assemble personalized skin profiles. Other factors such as skin tests, doctor or beauty advisor information, and other user-specific skin data (e.g., retailer profiles, imaged-based data) may also be collected. The personalized data gathered for a user drives the assignment of corresponding user attributes such as “Dry Skin,” “Acne- prone,” “Pregnant,” etc. Thus, attributes included in a user’s profile correspond to skin, hair, and / or other body characteristics of the user, which may change over time.

[0022] Meanwhile, medical and scientific literature are analyzed to identify proven effects of some or all known ingredients used in products, as well as evidence regarding proven effects of different product categories and the timing with which products are used or applied, and a compendium of rules is created to reflect known interactions between the ingredients, products, and human skin, particularly the effects of the ingredients upon skincare, cosmetic, beauty, and dermatologic factors of interest. For example, the rules can specify the products / ingredients that should or should not be applied for each condition, possibly with consideration of other aspects of users’ skin profiles (e.g., such as their age, locations, allergies, etc.), and may be categorized by product category or type (e.g., skin care, hair care, moisturizer, cleanser, shampoo, etc.).

[0023] Then, based on the user’s personal attributes and rules regarding attribute / product interactions, ingredients are identified that should, or should not, be used for each attribute, which may be stratified by product category. For example, a lipstick product that contains an ingredient that can worsen acne may be irrelevant since the product will not be applied to the acne-prone skin surface. Matching user-specific attributes to evidence pertaining to product use and evidence pertaining to ingredient uses creates a framework for an intelligent network of ingredient-product-user interactions that can be used to define requirements or suggestions for an effective and non-damaging skincare routine. On an individual user basis, the network is updated to specifically correspond to the user’s own attributes. Further, individual product analysis can be performed against this personalized network to enable the user to find products that are best suited to them, and to build a routine that aligns with some or all of their concerns.

[0024] Over time, new evidence (e.g., new medical research) may be automatically incorporated into the product database. Also, user feedback (e.g., regarding a daily skincare routine) may be collected and used to develop and train machine-learning models that dynamically adapt the recommender system to predict skin issues before they arise and to suggest products to mitigate those risks. Feedback may be in the form of skin tracking (e.g., tracking a user’s skin condition over time), metrics obtained through a user’s use of a system application, geolocation data identifying a user’s physical location, and so on, and may be used to continuously update the recommender. Similarly, other users (e.g., some or all system users)may provide insight to the system in terms of the products they use, whether those products are effective, other products they’ve used that were not effective or otherwise unpleasing, etc.

[0025] Further, by tracking products that individuals use or look up in the product / ingredient database, and where they shop (by way of affiliate links, or APIs deployed to retailer sites) their product preferences can be identified, which allows their recommender algorithms to be updated automatically. Yet further, by understanding a user’s location and lifestyle, the recommender algorithm can be dynamically adapted based on predicted issues. Thus, if an acne flare-up is predicted, the recommender algorithm can suggest products or lifestyle changes to counter or suppress the flare-up.

[0026] Moreover, when a user scans and submits a barcode while browsing a particular store or boutique, which may be identified using geolocation or IP address, the recommender may show (or emphasize) products available at that store to enhance opportunities for that retailer, and / or show products at lower price points from other retailers. Similarly, the system may provide a user with insight as to whether a particular product is suitable for use, and may also provide alternatives (i.e., other products that are close in composition or function and that match the user) that address their specific user attributes. In addition, recommendations can further benefit from the network of users of embodiments provided herein, by learning what people with similar skin profiles have used, what they have liked or disliked, and which products have (or have not) been implicitly endorsed by being included in their skincare routines.

[0027] Beyond recommendations, systems provided herein may be used to deliver content to users, such as personalized videos (e.g., specific to certain skin types, conditions, or attributes), social media posts, articles, and advertising. Thus, instead of scrolling through myriad content items on a typical social media site, many or most of which are not personally relevant, every item delivered in this manner is tailored to the recipient’s skin (not just her use habits, though this may be used to further refine the method of delivery of the content). Content provided via a user’s electronic device may be enhanced with advertising that is pertinent to the user’s suggested or recommended skincare products.

[0028] Data assembled regarding skincare products may further be accessed by users to obtain immediate information regarding ingredients, and / or whether a particular product matches the user (i.e., as a positive, negative, or neutral match with any of the user’s attributes). For example, an app or application may be provided for access by the user, and may be used to scan product barcodes or capture other identifiers (e.g., product name and manufacturer) via photograph or manual entry by the user.

[0029] Because skincare needs differ with one’s body parts, some embodiments may be focused upon a particular part of one’s body. For example, many people seek skincare regimens for their faces, and the needs and conditions of their facial skin differ from those of the rest of their skin. Therefore, in an illustrative embodiment, recommendations may be generated only or primarily for facial skincare, which may be referred to as one’s routine (i.e., a suggested routine to be applied or followed daily). However, interactions between skincare products and other products employed by a user may be considered during the recommendation process.

[0030] Figure 1 is a block diagram of a system for generating personalized recommendations regarding skincare products, according to some embodiments.

[0031] In these embodiments, users (e.g., user 112) employ one or more electronic devices (e.g., user device 114), such as smartphones, other handheld devices, laptop / notebook / desktop computers, and so on, to interact with a system described herein. A preliminary interaction is with one or more computer systems (e.g., web servers, data servers, portals) that elicit information from the user to assemble a snapshot of the user (including health, demographics, skin and body characteristics, etc.) to add to skin profdes 110. Thus, for each user of the system, a skin profde 110 is stored, updated as necessary, and used to generate a personalized skincare routine 120.

[0032] Besides any conditions (e.g., acne, eczema, hyperpigmentation), concerns (e.g., dry skin, oily skin), and / or factors (e.g., stress, diet, UV exposure) identified by the user during assembly of his or her profile 110, other conditions and / or concerns specific to the user may be received from other sources (e.g., a healthcare professional, a clinician, a beauty consultant, retailer profiles of the user, image-based modalities, results of physical skin tests(such as patch tests)). Thus, skincare and / or other health concerns 130 for specific users may be initialized and / or updated over time by a user (e.g., over the course of time using a suggested routine), and / or received from external sources. Skin conditions and concerns may be collectively referred to herein as “conditions.”

[0033] The system further comprises rule repository 140, which contains rules that match product ingredients (e.g., botanical substances, chemicals, etc.) with users, based on proven effects of every ingredient recognized by the International Nomenclature of Cosmetic Ingredients (INCI), of which there are over 14,000. More specifically, numerous (e.g., over 300) references, including medical journal articles, research papers, other published literature, test results, and so on, are parsed or analyzed to identify their effect upon human skin (especially facial skin) and skin conditions. The system may also consider ingredients beyond the INCI classification, such as drug ingredients (e.g., Adapalene, Octinoxate), complementary and alternative medicines, natural products / ingredients, and / or medical device-based particles, which may be prescription-based or over the counter.

[0034] Analysis of the literature may be performed automatically (e.g., by a computer system programmed for the purpose) or manually, but regardless of how the analysis is performed, a comprehensive collection of data is assembled that demonstrates the effects (and / or side effects) of each known skincare product ingredient, particularly its effect (if any) upon known skin conditions. From that data, the repository is populated with rules for recommending (or not recommending) a particular ingredient (or a product containing the ingredient) to a user based on the user’s skin profile. Rules in the repository may not only serve to support the recommendation of ingredients, product categories, and / or products, but may also provide timing information (e.g., an order in which to apply an ingredient or product, frequency of use, when to use it, etc.).

[0035] In some implementations, rules may be weighted based on the strength of the evidence of the effects of the associated ingredients and / or the source(s) of the evidence. Thus, a rule that indicates a given ingredient has a negative effect upon a specific attribute and that is derived from tens of references that are in agreement, including well-respected sources, may be weighted more heavily than another rule that indicates that the same ingredient has abeneficial effect upon the same or a different attribute. The rules’ weights may be considered when recommendation system 160 assembles a suggested skincare routine 120 for a user.

[0036] Product database 150 maps each known skincare and cosmetic ingredient to all known products that include the ingredient and / or vice versa. For example, every product may be stored with a plethora of data (e.g., manufacturer, type of product (e.g., skin care, hair care), hyperlinks, price), but particularly a listing of the product’s ingredients. Thus, after an ingredient is found that is a positive match (or a negative match) with a user’s skin profile (using rule repository 140), product database 150 may be consulted to identify corresponding products to recommend for inclusion with (or exclusion from) the user’s routine.

[0037] Finally, recommendation system (or recommender) 160 draws from skin profiles 120, concerns 130 (if any), rule repository 140, and product database 150 to assemble a user’s suggested personalized skincare routine 120, which may be delivered via user device 114.

[0038] The system may further include content generator 170, which generates and dispatches to user 112 (e.g., via user device 114) content that is personalized to the user, especially the user’s attributes. Thus, when a new product is released along with reliable published medical literature indicating that the product is successful in treating a skin condition of the user, information regarding the product may be transmitted to the user. Information regarding other new products, however, that are neutral matches with the user’s attributes (e.g., its ingredients are inert and no data are available to suggest a beneficial or detrimental effect), may or may not be dispatched to the user.

[0039] In some embodiments, rule repository 140 comprises separate tables, databases, tabs, or other partitions for identifying ingredients that should be included in a user’s suggested skincare routine 120 (i.e., positive matches) and for identifying ingredients that should be excluded from the routine (i.e., negative matches). In these embodiments, each ‘include’ and ‘exclude’ rule maps an attribute (i.e., one of the attributes that may be assigned to a user as part of her skin profile) to categories of products known to interact positively or negatively with the attribute (or, more specifically, to a skin or body characteristic represented by the attribute).

[0040] In some implementations, all known products are divided into first-level product “classes,” such as skincare, personal care, hair care, and makeup. Class categories may be divided into second-level product “families,” such as: cleansers, treatments, moisturizers and sun care (for the “skincare” class of products); deodorant, shaving and hair removal, hand soap, and other (for the “personal care” class); shampoos, conditioners, and styling and treatment (for the “hair care” class); and face, eyes, and lips (for the “makeup” class).

[0041] Moreover, second-level classes may be subdivided into third-level product “species.” In an illustrative implementation, the family of skincare cleansers may include species such as facial cleansers, masks, and makeup removers; the skincare moisturizers family may include species such as face moisturizers, eye creams, and mists; the hair care styling and treatment family may include species such as hair oils, hair masks, and scalp treatments. In different embodiments and implementations, known products may be categorized in different ways.

[0042] Thus, for each attribute within users’ skin profiles, for each class / family / species of product that can improve or degrade the attribute or a characteristic associated with the attribute, the rules repository provides a tag that corresponds to ‘include’ and / or ‘exclude’ ingredients known to cause the improvement or degradation. For example, for the “Dry Face Skin” attribute, intersections of the attribute with all species of skincare cleansers may yield a tag such as “dry_friendly_cleanser.” Using that ingredient tag, product database 150 or some other data repository can be searched to identify all cleanser ingredients that were assigned that tag (e.g., glycerin, propylene glycol). Other positive matches for the “Dry Face Skin” attribute may be with skincare / moisturizers / face moisturizers, which may yield a “hydration” ingredient tag that can then be matched to identically tagged ingredients (e.g., glycolic acid, hyaluronic acid, urea).

[0043] Similarly, for third-level species of products that can worsen a user’s skin condition associated with an attribute, each intersection of those products with the attribute yields a tag that identifies ingredients that should be omitted from the user’s skincare routine. Thus, the rules database may contain, for the intersection of all skincare cleanser products and the “acne” attribute, a tag such as “oily acne avoid.” Consulting the product database (or aseparate collection of ingredient tags) reveals that the suggested routine should avoid skincare cleansers that include coconut oil, for example.

[0044] In short, each known positive or negative interaction between a given user attribute (or condition corresponding to the given attribute) and known products (e.g., categorized by class, family, and species) is identified in the rule repository. Each such interaction is represented with a tag that has been assigned to ingredients that are known to ameliorate (for positive matches) or endamage (for negative matches) human skin having the attribute / condition.

[0045] Based on the literature that was analyzed and used to create rule repository 140, user feedback (e.g., regarding their routines and / or products they use), ingredients may be graded, weighted, sorted or otherwise arranged in some order of effectiveness. For example, among the ingredients that match positively for the “Dry Face Skin” attribute, ingredients found to be more effective than others in treating users’ dry facial skin (e.g., as reported by users) may be prioritized over others. Similarly, among ingredients that match negatively, ingredients found to exacerbate the problem may be prioritized to ensure they are not included in the routine of a user having dry facial skin.

[0046] The system may also be configured to understand relative contributions of ingredients within a given product. For example, products containing multiple beneficial ingredients may be prioritized as a recommendation. Also, where supporting evidence exists, certain combinations of product ingredients and formulations, as well as ingredient concentrations, may affect an evaluation. For example, one specific ingredient may match negatively with a user when considered in isolation, but evidence may indicate that, when used with some other ingredient, the conflict is eliminated and the combination of the ingredients need not be avoided.

[0047] In some implementations, attributes that may be included in a user’s skin profile include acne, anti-aging, atopic dermatitis / eczema black skin, black / afro-textured hair, bromhidrosis, cheilitis / chapped lips, climate, combination face skin, cracked heels, curly hair, chemically treated hair, dandruff / seborrheic dermatitis, dry body skin, dry face skin, dry hair, dry hands, dullness, hair loss / alopecia, hyperhidrosis, hyperpigmentation, keratosis pilaris,lifestyle (e.g., diet, exercise, sleep, stress, smoking, UV exposure), melasma, oily face skin, oily hair, perioral dermatitis, pregnant or breastfeeding, psoriasis, razor bum / bumps, redness, rosacea, sun damage (e.g., actinic keratosis), and specific allergies (e.g., acrylates, fragrance, lanolin, tea tree oil).

[0048] Figure 2 is a flow chart illustrating a method of generating a personalized recommendation regarding skincare products, according to some embodiments. In some embodiments, one or more of the illustrated operations may be omitted, repeated, and / or performed in a different order. Accordingly, the specific arrangement of steps shown in Fig. 2 should not be construed as limiting the scope of the embodiments.

[0049] In operation 202, a repository of skin care rules is assembled (rule repository 140 of Figure 1) from a large corpus of medical, scientific, and other trustworthy data. Operation 202 may be continuous, meaning that as new references appear, they will be analyzed and the rules repository will be updated accordingly. As discussed above, each rule serves to identify, for each profile attribute associated with a condition that can be improved (or degraded) by a particular type of skincare product, the specific ingredients that are known to improve (or degrade) the condition.

[0050] In operation 204, a user ’ s input is solicited regarding her skin, for the purpose of assembling a custom skin profile. For example, the user may be asked to provide bodily characteristics such as age, gender, skin color, skin conditions, location, hair type, cosmetic use, pregnancy status, lifestyle, exercise level, skin test results, current skincare routine, medications, goals, etc. Thus, besides providing characteristics of her skin / body, the user may identify concerns such as a desire to resist the normal aging process (e.g., with anti-aging products), or a concern regarding her climate (e.g., if she lives in a cold and dry region).

[0051] In optional operation 206, concerns regarding the user’s skin or health may be received from other sources, such as a healthcare professional, retailers that have accumulated user-specific data, beauticians, cosmetologists, a repository of data regarding the user (e.g., images, test results), etc.

[0052] Based on the obtained information, in operation 208 pertinent attributes are assigned to the user. For example, based on her self-reported age, an “anti-aging” attribute may be assigned if she should begin using such products. Based on her location (e.g., one that has an arid climate), a “Dry Face Skin” and / or “Dry Body Skin” attribute may be assigned; based on her reported conditions (e.g., “psoriasis,” “acne,” “oily hair,” specific allergies), corresponding attributes are assigned; and so on.

[0053] In operation 210, a skin profile is created for the user, to include all attributes configured based on her self-reported profile information and any information pertaining to her received from third parties. Her profile may be stored (e.g., in skin profiles database 110 of Figure 1) and may be updated over time.

[0054] In operation 212, the rule repository is consulted to identify, for each profile attribute, ingredients that positively match the attribute, meaning that they will treat a condition associated with the attribute or otherwise be beneficial for the user. The repository may also be consulted to identify ingredients that match negatively, meaning that they may irritate the user’s skin or aggravate a condition. In short, the data repository is used to identify ingredients that the user should use (positive or ‘include’ ingredients) and / or should not use (negative or ‘exclude’ ingredients), based on her assigned attributes. In some implementations, instead of directly identifying ingredients, the rules repository produces ingredient tags that are used to lookup matching ingredients in some other collection of data (e.g., a product or ingredient database).

[0055] Neutral matching ingredients (or “neutral ingredients”) may also be identified, which will include those that either may be included in or excluded from the user’s routine, without significant effect, since they will neither help or harm the user’s skin or reported conditions.

[0056] Not every skin profile attribute will match positively or negatively with every product species, family, or even class. For example, since cracked heels are not known to be positively or negatively affected by any product that is applied to a user’s face, this attribute may not affect or be addressed by a suggested skincare routine. The attribute may, however, affect a body care routine if one is prepared.

[0057] In operation 214, a recommended skincare routine (e.g., a suggested daily routine) is generated for the user based primarily on the positive and negative ingredient matches with the user’s attributes. Specifically, the routine is constructed to include ingredients that should ameliorate the user’s attributes and / or help her achieve her goals (i.e., some or all positive matches), while excluding ingredients that would or could aggravate a condition or that would thwart a goal (i.e., all negative matches). Neutral matching ingredients may be included or excluded.

[0058] In operation 216, the routine may be enhanced to identify specific products (or classes / families / species of products) that include the identified positive ingredients and that eschew negative ingredients. Thus, the routine may indicate that a facial moisturizer should be used, and may even provide suggested options - such as specific moisturizer products that contain positively matching ingredients and no negatively matching ingredients. Alternatively, the routine may focus upon including and excluding individual ingredients, but links may be provided to help the user search system data (e.g., the product database) to find products that correspond to the ingredients. She may be able to enter other criteria to aid the search (e.g., a price point, a manufacturer or brand, a vendor that carries the products, a preferred local brick- and-mortar store).

[0059] In some implementations, a product database such as database 150 of Figure 1 is consulted for each positively matching ingredient to identify some or all known products that include the ingredient. The output may be sorted or filtered in any desired manner, such as by estimated or expected effectiveness, cost, overlap with other matching products (e.g., a product that includes multiple positive ingredients may be recommended over multiple products having individual positive ingredients), etc.

[0060] As part of the process of generating a suggested skincare routine for a user, the system may consider feedback provided by other users. For example, when a user reports a particular skin condition or concern, the system may search feedback from other users with the same condition, regarding their experience with the ingredients and / or products recommended to them for treating the condition. The more users that report satisfaction, the more likely a particular ingredient (or product) will be included in the suggested skincare routine, thereby prioritizing optimized products and ingredients for a given user.

[0061] In some implementations, the user identifies her current routine as part of the process of assembling her skincare profile. When the system generates her suggested personalized routine, issues regarding her current routine may be highlighted for the sake of comparison. For example, if she currently uses a product that includes an ingredient that is known to aggravate a condition she identified, or that conflicts with one of her assigned attributes (i. e. , a negative ingredient), she will be informed that the ingredient should be omitted (and possibly why). Similarly, if her current routine omits ingredients that would help treat a specified condition (i.e., positive ingredients), she will be informed of a suitable product to include in her suggested routine.

[0062] After the user starts implementing her recommended skincare routine, feedback is requested regarding its effectiveness, not only to help refine her routine, if deemed advisable, but also to help during generation of other users’ routines. Illustratively, if the user reports that the routine is not helping with some aspect of her skin care regimen (e.g., her dry skin is not clearing up), the system may adjust the routine to change ingredients or switch to a product (or products) with different ingredients. For example, assuming an inclusion rule that was triggered during generation of her initial recommended skincare routine output a “dry_friendly_cleanser” ingredient tag, and the routine was configured to include a face cleanser product with one ingredient (e.g., glycerin), the routine may be modified to remove that product and add one with a different ingredient (e.g., sodium laureth sulfate).

[0063] In some embodiments, the system monitors a user’s lifestyle factors to determine whether and when a skin flare-up is impending or likely, and to make recommendations regarding the user’s routine and / or a particular product to mitigate the flare- up. These interventions may illustratively help a user who travels from a warm and humid climate to a cold and dry climate or a user whose acne flares with stress. In short, the system may recognize warning signs and proactively recommend a product-based course of action and / or lifestyle changes.

[0064] Also, when a user (or a third party) provides test results, such as from a skin patch, those results may be used to update the user’s skin profile and / or her personalized skincare routine. For example, if an allergen patch test indicates that the user is allergic to one or more particular ingredients, an attribute identifying the allergy may be added to her profileand / or a personalized ingredient exclusion list may be created for her to ensure those ingredients (and products that include those ingredients) are not recommended to her.

[0065] These actions can also be taken if the user expresses displeasure or disappointment with a product for some reason other than lack of effectiveness (which, if reported, may lead to substitute recommendations. For example, if she indicates that a product is too oily or sticky in her opinion, the system can determine the ingredients of the product that likely caused the dissatisfaction, and help avoid recommending those ingredients (or products including those ingredients or similar ingredient combinations) in the future.

[0066] To aid with gathering feedback, an app or application operated by the user to interact with the system may track the user’s progress, such as inquiring whether the routine has been and is being followed, how her skin health has changed, whether she has noticed her skin improving or getting worse, any other changes to her health or life (e.g., diet changes, increased stress, relocation), etc.

[0067] When a user identifies products used in her current skincare regimen, and / or when she queries the system regarding a particular product, the system may compare the product(s) to a series of alerts. Each alert determines whether a product includes an ingredient that is one of her positive matches (meaning it should be included in her skincare) or one of her negative matches (meaning it should not be used by her). For example, the alerts may operate similarly to rules in a rule repository, in that they involve examining the attributes of her skin profile and advising her if she is using products that are good or bad for her skin.

[0068] Figure 3 illustrates rules and ingredient tags for facilitating generation of a personalized skincare routine, according to some embodiments. In these embodiments, a system for providing suggested routines (such as the system of Figure 1) comprises Include Rules 302 and Exclude Rules 304 within a single rule repository or a bifurcated rule repository, and maintains ingredient tags 306 to map from triggered rules to matching ingredients.

[0069] Include Rules table 302 lists some or all user attributes (i.e., user attributes 310a - 31 On), as well as one or more levels of product categories. In the illustrated embodiments, only two layers of categories are provided, which may be equated with productfamilies 320 and product species 322, 324. Thus, in this illustrative table, include tags or “I- Tags” indicate that positive matches exist between user attribute 310a and product species 322a, 322b within product family 320a via I-Tag 330a, and a positive match exists between attribute 310c and product species 330b via I-Tag 330b. Furthermore, attribute 310c also matches positively with product species 324b, 324c of product family 320b via I-Tag 330b, and attribute 31 On positively matches product species 324a via I-Tag 330c. Reference to Ingredient Tags table 306 reveals that I-Tag 330a corresponds to ingredients 350a-350c, while I-Tag 330b comprises ingredients 350d-350g. I-Tag 330c maps to ingredients 350a, 350e, and 350h.

[0070] Exclude Rules table 304 operates similarly to Include Rules table 302, but instead identifies ingredients that should be avoided for particular attribute-product category interactions. Thus, for attribute 310a, all product species 322, 324 for both product families 320a, 320b map negatively to exclude tag or E-Tag 340a, and attribute 310c also maps negatively to product species 324b with E-Tag 340a. In addition, attribute 310b matches negatively with product species 324b via E-Tag 340b, and attribute 310c has a negative match with product species 322a, 322b, and 324b via E-Tag 340c. Ingredient Tags table 306 reveals that E-Tag 340a corresponds to ingredients 350j -3501, E-Tag 340b maps to ingredients 350a- 350d, and E-Tag 340c comprises ingredients 350m, 350n.

[0071] Figure 4 illustrates sample alerts for evaluating a specific product or a user’s routine (e.g., her current daily routine), according to some embodiments. Although these embodiments principally consider the presence or absence of specific ingredients when applying an alert and making an evaluation, in other embodiments the ingredient is just one consideration, along with the amount and / or concentration of the ingredient, the entire formulation of a product that includes the ingredient, and / or a category of product that includes the ingredient.

[0072] Product include alert 410 is an alert configured to inform a user whether a particular product (e.g., a product identified by the user) matches or interacts positively with one particular attribute (i.e., attribute 310a). Thus, (a) if the product is categorized within any product species of product class 321a of product family 320a, and a rule repository (e.g., rule repository 140 of Figure 1 or include rules table 302 of Figure 3) maps the product to theattribute via ‘include’ ingredient tag (I-Tag) 330a, OR (b) the product is categorized within any product species of product class 321b of product family 320a, and a rule repository maps the product to the attribute via any of I-Tags 330b, 330c, and 330d, OR (c) the product is categorized within any product species of product class 321c of product family 320a, and a rule repository maps the product to the attribute with any of I-Tags 330b, 330c, and 330d, the system will output a message indicating that the product includes one or more ingredients that will help with attribute 310a (e.g., a skin condition such as acne).

[0073] Product exclude alert 420 is an alert configured to inform a user whether a particular product matches or interacts negatively with one particular attribute (i.e., attribute 310a). Thus, if (a) the product is categorized within any class and species of product family 320a, EXCEPT product species 324a, 324b or 324c of product class 323a, and maps to the attribute via ‘exclude’ ingredient tag (E-Tag) 340a, OR (b) if the product is categorized within any class and species of product family 320b and maps to the attribute via E-Tag 340a, the system will output a message indicating that the product may worsen attribute 310a.

[0074] Routine include alert 430 is an alert configured to inform a user whether a routine (e.g., the user’s current daily skincare routine) matches or interacts positively with attribute 310a. As shown in Fig. 3, if (a) the routine includes any product that is categorized within any product species of product class 321a of product family 320a and that maps to attribute 310a via I-Tag 330a, OR (b) the routine includes a product categorized within any product species of product class 321b of product family 320a and that maps to attribute via any of I-Tags 330b, 330c, or 330d, OR (c) the routine includes a product categorized within any product species of product class 321c of product family 320a and that maps to attribute via any of I-Tags 330b, 330c, or 330d, the system will output a message the same or similar to the output message of product include alert 410. However, if the routine does not include a product that satisfies any of these examinations, the system will output a message indicating that the routine is deficient with respect to attribute 330a.

[0075] Routine exclude alert 440 is an alert configured to inform a user whether a routine (e.g., the user’s current daily skincare routine) matches or interacts negatively with attribute 310a. Thus, if (a) the routine includes any product categorized within any species and class of product family 320a, EXCEPT species 324a, 324b, or 324c of product class 323a, andmaps to attribute 310a via E-Tag 340a, OR (b) the routine includes any product categorized within any species and class of product family 320b and maps to the attribute via E-Tag 340a, the system will output a warning that the routine may worsen the attribute.

[0076] Category include alert 450 is configured to notify a user whose skin profile includes attribute 310a that her current routine omits a product category (i.e., species 322b) known to ameliorate or help with the attribute. She is prompted to take action (e.g., click on a provided link) to find one that has ingredients that help with the attribute.

[0077] In some embodiments, when one or more ingredients conflict with a user (i.e., they are negative matches when considered in isolation), but one or more formulations that include the ingredient(s) are acceptable, an alert that would otherwise be triggered may be suppressed. For instance, if a product would normally create an Exclude Alert, but there is specific information about the product itself or the formulation combination as a whole, the alert may be suppressed. Alternatively, the user may be informed that while the product contains ingredients known to individually be problematic, data indicate that the formulation itself is compatible with the user, and a citation may even be provided for the user if / when desired.

[0078] Methods described herein may be aided through the training and application of a machine-learning model that can consider a user’s current and history of product use, effectiveness of those products (as reported by her and / or other users, as described in medical / dermatological literature), and / or other factors. Yet further, or instead, a generative pre-trained transformer (or GPT) model could be developed for users, to assist in formulation of a recommended skincare routine, and could serve as a personal dermatologist educated not only with all applicable medical theory, but also with in-depth familiarity with all skincare products and their ingredients.

[0079] In some implementations, users’ skincare data may be leveraged to drive creation of a new skincare, healthcare, or cosmetic product. For example, based on users’ attributes and known interactions of associated conditions with known product ingredients, a new product could be formulated that contains multiple positively matched ingredients, so thata user can obtain multiple benefits from a single product instead of having to use multiple separate products.

[0080] In some embodiments, one or more components of the system are co-located with a product retailer, healthcare or dermatology professional, or some other entity. In these embodiments, a user can interact with the system via the co-located system component(s) to identify products that match the user (positively and / or negatively) and that are available at that location, thereby guiding the user to or away from products, as appropriate. In an illustrative implementation, the user may physically operate a system console or kiosk, or may operate her smartphone (or other digital device) to connect to a wireless network or access point provided by the entity but associated with an organization that operates or provides the system.

[0081] In some other embodiments, the system operates through one or more APIs executing on a website associated with a product retailer or different entity (e.g., a product manufacturer or distributor) other than the system and the organization that provides the system. In these embodiments, while the user’s main interaction may be with the retailer (e.g., to identify the user’s skin conditions and concerns), the retailer may use the system’s product database (or a version of the database focused on the retailer’s products), and rules applied during construction of a user-personalized routine may focus upon categories of products offered by the retailer.

[0082] By configuring privacy controls or settings as they desire, members of a social network, a professional network, or other user community that may use or interact with embodiments described herein can control or restrict the information collected from them, the information that is provided to them, their interactions with such information and with other members, and / or how such information is used. Implementation of an embodiment described herein is not intended to supersede or interfere with the members’ privacy settings.

[0083] An environment in which one or more embodiments described above are executed may incorporate a general-purpose computer or a special-purpose device such as a hand-held computer or communication device. Some details of such devices (e.g., processor, memory, data storage, display) may be omitted for the sake of clarity. A component such as a processor or memory to which one or more tasks or functions are attributed may be a generalcomponent temporarily configured to perform the specified task or function, or may be a specific component manufactured to perform the task or function. The term “processor” as used herein refers to one or more electronic circuits, devices, chips, processing cores and / or other components configured to process data and / or computer program code.

[0084] Data structures and program code described in this detailed description are typically stored on a non-transitory computer-readable storage medium, which may be any device or medium that can store code and / or data for use by a computer system. Non-transitory computer-readable storage media include, but are not limited to, volatile memory; non-volatile memory; electrical, magnetic, and optical storage devices such as disk drives, magnetic tape, CDs (compact discs) and DVDs (digital versatile discs or digital video discs), solid-state drives, and / or other non-transitory computer-readable media now known or later developed.

[0085] Methods and processes described in the detailed description can be embodied as code and / or data, which may be stored in a non-transitory computer-readable storage medium as described above. When a processor or computer system reads and executes the code and manipulates the data stored on the medium, the processor or computer system performs the methods and processes embodied as code and data structures and stored within the medium.

[0086] Furthermore, the methods and processes may be programmed into hardware modules such as, but not limited to, application-specific integrated circuit (ASIC) chips, field- programmable gate arrays (FPGAs), and other programmable-logic devices now known or hereafter developed. When such a hardware module is activated, it performs the methods and processes included within the module.

[0087] The foregoing embodiments have been presented for purposes of illustration and description only. They are not intended to be exhaustive or to limit this disclosure to the forms disclosed. Accordingly, many modifications and variations will be apparent to practitioners skilled in the art. The scope is defined by the appended claims, not the preceding disclosure.

Claims

What Is Claimed Is:

1. A method of generating a personalized skincare routine, the method comprising: creating a skin profile for a user, wherein the skin profile comprises one or more attributes of the user; for each attribute in the skin profile, for each of multiple categories of skincare products, determining whether any products in the category contain an ingredient or formulation that can have a positive or negative effect on the corresponding user attribute; and forming the personalized skincare routine for the user, wherein: the personalized skincare routine includes one or more ingredients or formulations that can have positive effects on one or more of the user attributes; and the personalized skincare routine omits all ingredients and formulations that can have negative effects on any of the user attributes.

2. The method of claim 1, further comprising, prior to said creating: analyzing a corpus of literature reporting effects of skincare product ingredients and formulations; tagging skincare product ingredients that have identical effects with a descriptive tag; and forming a set of rules that map ingredients that affect user attributes to categories of products that include the ingredients; wherein said determining whether any products in the category contain an ingredient that can have a positive or negative effect on the corresponding user attribute comprises consulting the set of rules.

3. The method of claim 2, wherein creating the set of rules is updated based on one or more of: a user update of his or her corresponding skin profile; feedback and / or preferences received from one or more users regarding their personalized routines; and analysis of multiple users having a common skin condition or concern.

4. The method of claim 1, wherein creating the skin profile for the user comprises:eliciting from the user one or more skin conditions characteristic of the user’s body.

5. The method of claim 1, wherein forming the personalized skincare routine for the user comprises: for each attribute that matches one or more ingredients that can have a positive effect on the user attribute, recommending one or more skincare products having at least one of the one or more ingredients or relevant formulation.

6. The method of claim 1, further comprising: assembling a set of alerts that, when applied to a current routine of the user, informs the user whether the routine includes any ingredients or formulations likely to have a negative effect on a user attribute.

7. The method of claim 6, wherein the set of alerts, when applied to the current routine of the user, also informs the user whether the routine includes any ingredients or formulations likely to have a positive effect on a user attribute.

8. The method of claim 1, further comprising: receiving from the user feedback regarding the routine; and in response to the feedback, adjusting the routine to replace a first ingredient or formulation that can have a positive effect on a first user attribute with a second ingredient or formulation that can have a positive effect on the first user attribute.

9. The method of claim 1, further comprising: receiving from the user information identifying a first skincare product; determining whether the first skincare product contains an ingredient or formulation that can have a positive or negative effect on a user attribute; and informing the user whether the first skincare product is compatible with the user.

10. A non-transitory computer-readable medium storing instructions that, when executed by one or more computer systems, cause the one or more computer systems to perform a method of generating a personalized skincare routine, the method comprising: creating a skin profile for a user, wherein the skin profile comprises one or more attributes of the user;for each atribute in the skin profile, for each of multiple categories of skincare products, determining whether any products in the category contain an ingredient or formulation that can have a positive or negative effect on the corresponding user atribute; and forming the personalized skincare routine for the user, wherein: the personalized skincare routine includes one or more ingredients or formulations that can have positive effects on one or more of the user atributes; and the personalized skincare routine omits all ingredients and formulations that can have negative effects on any of the user atributes.

11. A method of generating a personalized routine for personal care, the method comprising: assembling a body profile for a user, to include a set of atributes of the user’s body; for each atribute: when one or more product ingredients or formulations are known to ameliorate the atribute, identifying at least one of the ameliorative ingredients or formulations; and when one or more product ingredients or formulations are known to endamage the attribute, identifying at least one of the endamaging ingredients or formulations; creating the personalized routine for the user to include at least one product ingredient or formulation known to ameliorate an atribute of the user and to exclude all product ingredients and formulation known to endamage an atribute of the user.

12. The method of claim 11, wherein assembling the body profile for the user comprises: eliciting the set of atributes from the user; and when specified by the user, receiving from a third-party one or more concerns regarding the user’s body.

13. The method of claim 11, wherein identifying one or more product ingredients or formulations that are known to ameliorate an atribute comprises: within a set of body care rules, mapping the atribute to categories of body care products associated with the atribute to identify one or more first ingredient tags; wherein each first ingredient tag corresponds to ingredients known to ameliorate theattribute.

14. The method of claim 11, wherein identifying zero or more product ingredients or formulations that are known to endamage an attribute comprises: within a set of body care rules, mapping the attribute to categories of body care products associated with the attribute to identify zero or more second ingredient tags; wherein each second ingredient tag corresponds to ingredients known to endamage the attribute.

15. The method of claim 11 , further comprising: enhancing the personalized routine to suggest use of at least one product having at least one product ingredient or formulation known to ameliorate an attribute of the user; or enhancing the personalized routine to indicate that no ingredient or formulation will ameliorate the user’s attributes.

16. The method of claim 11 , further comprising: receiving from the user an identifier of a first body care product; identifying ingredients and formulations included in the first product; determining whether any ingredients or formulations in the first product are known to ameliorate any attribute of the user; determining whether any ingredients or formulations in the first product are known to endamage any attribute of the user; and informing the user whether the first product is compatible with the user’s body profile.

17. A system for generating personalized skincare routines, comprising: a rule repository that maps user attributes to categories of products associated with the user attributes, wherein: when a product category includes at least one product having a first set of ingredients or a first formulation known to be helpful for a given user attribute, or a second set of ingredients or a second formulation known to aggravate the given user attribute, the mapping identifies the first set or the second set of ingredients and the first or second formulation, respectively; a product database that includes details of every known product, including ingredientsand formulations of the product; a collection of skin profiles, wherein each skin profile identifies one or more attributes of a corresponding user; and a recommendation apparatus configured to generate for a user a personalized skincare routine that includes at least one ingredient from a first set of ingredients or a first formulation and no ingredients from a second set of ingredients or a second formulation.

18. The system of claim 17, wherein the recommendation apparatus comprises: one or more processors; and memory storing instructions that, when executed by the one or more processors, cause the recommendation apparatus to: input a first skin profile corresponding to a first user; and for each attribute in the first set of attributes, search the rule repository for skincare product categories that, when mapped to the attribute, identify a non-null first set of ingredients or formulation and / or a non-null second set of ingredients or formulation.

19. The system of claim 18, wherein the recommendation apparatus memory further stores instructions that, when executed by the one or more processors, cause the recommendation apparatus to: for each skincare product category that mapped to an attribute to identify a non-null first set of ingredients or formulation, consult the product database to identify at least one product that includes at least one of the non-null first set of ingredients or formulation; and include the at least one product in the personalized skincare routine.

Images