AI intelligent skin health management method based on multi-modal data fusion

Abstract

The invention relates to an AI intelligent skin health management method based on multi-modal data fusion, and belongs to the technical field of skin health management. Comprising the following steps: S1, multi-dimensional real-time detection; s2, AI skin health assessment; s3, scientific step generation; s4, skincare product intelligent matching strategy; s5, AI skin care teaching; s6, traditional Chinese medicine constitution identification; s7, outputting energy; s8, establishing a cross-domain feature fusion model; and S9, performing a dynamic priority algorithm. The system is rich in function, and can dynamically monitor the skin state and generate a personalized nursing scheme; skincare guidance vacancy can be carried out, skincare steps are recommended or skincare products are matched based on real-time skin data, and misoperation or product mismatching is avoided; the problem of data islands is effectively solved, a skin care product component database is opened on the basis of providing use records, and adaptive products are scientifically recommended; the traditional Chinese medicine constitution identification technology and the western medicine skin science are effectively combined, so that the intervention scheme is prevented from being in a one-piece mode.

Description

Technical Field

[0001] The present invention relates to an AI intelligent skin health management method based on multimodal data fusion, belonging to the technical field of skin health management. Background Art

[0002] In recent years, the market for home beauty devices and health monitoring equipment has grown rapidly. However, compared with existing products on the market, most of the existing products focus on a single function. Mainstream products such as the Feichi three-sided beauty device (three-sided guide head switching function) and Yamen ACE MAX 2025 (multi-frequency radio frequency anti-aging) focus on improving technical parameters (such as temperature control accuracy and energy intensity), but lack dynamic monitoring of skin condition and generation of personalized care plans; lack of skin care guidance: products such as Tripollar VX3 rely on users to select modes independently, and do not recommend skin care steps or match skin care products based on real-time skin data, which can easily lead to improper operation or product mismatch; data island problem: although entry-level models such as Amiro ARF001 support APP linkage, they only provide basic usage records, and have not opened up the skin care product ingredient database, making it impossible to scientifically recommend suitable products; at the same time, traditional Chinese medicine constitution identification technology (such as pulse diagnosis and tongue diagnosis) has not been effectively combined with Western medicine dermatology, resulting in one-sided intervention plans.

[0003] Based on this, the present invention provides an AI intelligent skin health management method based on multimodal data fusion Summary of the Invention

[0004] In view of this, the present invention provides an AI intelligent skin health management method based on multimodal data fusion, which is rich in functions; can provide skin care guidance where it is lacking; effectively solve the data island problem; and effectively combine traditional Chinese medicine constitution identification technology with Western medicine dermatology to avoid one-sided intervention plans.

[0005] The present invention provides an AI intelligent skin health management method based on multimodal data fusion. The proposed technical solution includes the following steps:

[0006] S1: Multi-dimensional real-time detection: Integrating multi-spectral imaging, bioimpedance sensors, and environmental sensor multimodal data to generate a dynamic skin digital twin that reflects the user's skin status in real time, providing a basis for subsequent evaluation and care;

[0007] S2: AI Skin Health Assessment: This uses machine learning models to train data covering sensitive skin and acne-prone skin segments, outputs a prioritized ranking of skin issues, and associates potential causes.

[0008] S3: Scientific Step Generation: Automatically generates a phased, targeted care plan based on the test results, detailing the sequence of use, duration, and technique to ensure users can execute it correctly.

[0009] S4: Intelligent Skincare Matching: Accesses a skincare ingredient database, covering detailed information on various brands and products. It intelligently recommends suitable skincare products based on the user's skin condition and ingredient preferences, while avoiding allergenic ingredients. Recommended products are labeled with a price-performance ratio, helping users make more informed purchasing decisions.

[0010] S5: AI skincare tutorial: uses a 3D macro camera to identify user actions, capture the user's skincare process, correct incorrect techniques in real time, and mark facial care zones to help users understand skincare details more comprehensively;

[0011] S6: Traditional Chinese Medicine Constitution Identification: Utilizes a 128-dot matrix flexible piezoelectric sensor with an integrated dynamic pressure compensation unit for pulse acquisition. It also incorporates a multi-spectral ring light source and polarization filter, and uses the U-Net algorithm to segment the tongue for tongue image recognition.

[0012] S7: Energy output: The magnetic guide head supports adaptive switching of radio frequency, microcurrent, red and blue light modes;

[0013] S8: Cross-domain feature fusion model: Based on the Transformer architecture, it realizes the semantic alignment of pulse time series data, tongue features and skin parameters;

[0014] S9: Dynamic priority algorithm: Intelligently judge the user's health status based on the results of pulse, tongue and skin tests.

[0015] Furthermore, the multispectral imaging uses light of different wavelengths to image the skin, which can reveal the structure, blood vessel distribution, and pigment deposition information of the deep layer of the skin, and is used to identify spots and erythema.

[0016] Furthermore, the bioimpedance sensor measures the electrical conductivity of the skin to evaluate the skin's moisture content and physiological indicators of oil secretion, which is used to identify water-oil balance and barrier health.

[0017] Furthermore, the environmental sensor includes a temperature and humidity sensor and a pollutant sensor to monitor the temperature and humidity of the user's environment.

[0018] Furthermore, the skin problem priority ranking provides the user with a skin problem priority list based on the severity and potential impact of the skin problem.

[0019] Furthermore, the pulse condition acquisition is used to accurately capture subtle changes in the pulse condition, and the accuracy of the 128-dot matrix flexible piezoelectric sensor reaches 0.05N.

[0020] Furthermore, the multi-spectral ring light source includes different wavelengths of 450nm, 530nm, and 660nm.

[0021] Furthermore, the magnetic guide head supports fast function switching. The radio frequency mode is used for deep heating to promote collagen regeneration, the microcurrent mode is used for lifting and firming, and the red and blue light modes are used for anti-inflammatory and acne removal.

[0022] Beneficial effects of the present invention:

[0023] The present invention provides an AI intelligent skin health management method based on multimodal data fusion, which is rich in functions and can dynamically monitor skin conditions and generate personalized care plans; it can provide skin care guidance where it is lacking, and recommend skin care steps or match skin care products based on real-time skin data to avoid improper operation or product mismatch; it can effectively solve the problem of data islands, connect the skin care product ingredient database based on the provision of usage records, and scientifically recommend suitable products; it can effectively combine traditional Chinese medicine constitution identification technology with Western medicine dermatology to avoid one-sided intervention plans. DETAILED DESCRIPTION

[0024] The preferred embodiments of the present invention will be described in detail below.

[0025] The present invention provides an AI intelligent skin health management method based on multimodal data fusion, which includes the following steps:

[0026] The following steps are involved:

[0027] S1: Multi-dimensional real-time detection: Integrating multi-modal data such as multispectral imaging, bioimpedance sensors, and environmental sensors, and using light of different wavelengths to image the skin, it can reveal the deep structure of the skin, blood vessel distribution, and pigmentation information, which can be used to identify spots and erythema. Bioimpedance sensors measure the skin's electrical conductivity to assess the skin's moisture content and physiological indicators of oil secretion, which can be used to identify water-oil balance and barrier health. Environmental sensors include temperature and humidity sensors and pollutant sensors, which monitor the temperature and humidity of the user's environment and generate dynamic skin digital twins that reflect the user's skin status in real time, providing a basis for subsequent evaluation and care;

[0028] S2: AI Skin Health Assessment: Through machine learning model training data covering sensitive skin and acne-prone skin segments, it outputs skin problem priority rankings. Based on the severity and potential impact of the skin problem, it provides users with a skin problem priority list, such as repairing the skin barrier first, followed by oil control, etc., and associates potential triggers, such as staying up late, pollution, etc., to help them understand the skin problems that need to be solved first.

[0029] S3: Scientific Step Generation: Automatically generates a phased, targeted care plan based on test results, such as morning: antioxidant essence + physical sunscreen; night: barrier repair cream + blue light anti-inflammatory, with detailed instructions on the order, duration, and technique of use to ensure users can follow it correctly.

[0030] S4: Smart Skincare Matching: Accesses skincare ingredient databases such as COSDNA and EWG, covering detailed information on various brands and products. It intelligently recommends suitable skincare products based on the user's skin condition and ingredient preferences, while avoiding allergenic ingredients such as alcohol and fragrance. Recommended products are labeled with a price-performance ratio, including a SkinCeuticals B5 alternative, to help users make more informed purchasing decisions.

[0031] S5: AI skincare tutorial: This tool uses a 3D macro camera to identify user actions, capture the user's skincare process, and correct incorrect techniques in real time, such as using an RF device to circle too quickly and cause burns. It also identifies facial care zones, such as the T-zone for focused cleansing, to help users gain a more comprehensive understanding of skincare details.

[0032] S6: Traditional Chinese Medicine Constitution Identification: Utilizes a 128-dot matrix flexible piezoelectric sensor with an accuracy of 0.05N and an integrated dynamic pressure compensation unit to accurately capture subtle changes in the user's pulse. Pulse acquisition and calibration are performed using a standard pulse simulator (60-120bpm). Kalman filtering eliminates noise, improving the signal-to-noise ratio to 35dB.

[0033] A multispectral ring light source with wavelengths of 450nm, 530nm, and 660nm is configured with a polarization filter. The U-Net algorithm is used to segment the tongue area. Tongue images are captured under a D65 light source, and an improved ResNet-50 network is used to extract the HSV spatial histogram peak for user tongue image recognition.

[0034] S7: Energy output: Magnetic guide head supports radio frequency (0.5-2W / cm 2 ), microcurrent (50-200μA), adaptive switching of red and blue light modes, when skin impedance is greater than 200kΩ, limit RF power to less than 0.5W / cm 2 The intensity of the microcurrent is dynamically adjusted according to the user's pain threshold. The magnetic guide head supports fast function switching. The radio frequency mode is used for deep heating to promote collagen regeneration. The microcurrent mode is used for lifting and firming. The red and blue light modes are used for anti-inflammatory and acne treatment.

[0035] S8: Cross-domain feature fusion model: Based on the Transformer architecture, it realizes the semantic alignment of pulse time series data, tongue features and skin parameters;

[0036] S9: Dynamic priority algorithm: Based on the results of pulse, tongue and skin tests, the algorithm can intelligently judge the user's health status. The algorithm can dynamically adjust the priority of the skin care plan according to the user's actual situation to ensure that the user receives the most effective care. For example, when the pulse indicates "Qi and blood deficiency" and the skin test shows "collagen loss", internal adjustment plans are recommended first, such as adjusting diet, work and rest, etc.

[0037] The present invention is rich in functions and can dynamically monitor skin conditions and generate personalized care plans; it can fill in the gaps in skin care guidance and recommend skin care steps or match skin care products based on real-time skin data to avoid improper operation or product mismatch; it can effectively solve the problem of data silos, connect the skin care product ingredient database based on usage records, and scientifically recommend suitable products; it can effectively combine traditional Chinese medicine constitution identification technology with Western medicine dermatology to avoid one-sided intervention plans.

[0038] The above description of the present invention and its embodiments is non-limiting, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by the above description and, without departing from the purpose of the present invention, designs a structure and embodiment similar to the technical solution without creatively designing, they shall fall within the scope of protection of the present invention.

Claims

1. An AI-powered skin health management method based on multimodal data fusion, characterized by: The following steps are involved: S1: Multi-dimensional real-time detection: Integrating multi-spectral imaging, bioimpedance sensors, and environmental sensor multimodal data to generate a dynamic skin digital twin that reflects the user's skin status in real time, providing a basis for subsequent evaluation and care; S2: AI Skin Health Assessment: This uses machine learning models to train data covering sensitive skin and acne-prone skin segments, outputs a prioritized ranking of skin issues, and associates potential causes. S3: Scientific Step Generation: Automatically generates a phased, targeted care plan based on the test results, detailing the sequence of use, duration, and technique to ensure users can execute it correctly. S4: Intelligent Skincare Matching: Accesses a skincare ingredient database, covering detailed information on various brands and products. It intelligently recommends suitable skincare products based on the user's skin condition and ingredient preferences, while avoiding allergenic ingredients. Recommended products are labeled with a price-performance ratio, helping users make more informed purchasing decisions. S5: AI skincare tutorial: uses a 3D macro camera to identify user actions, capture the user's skincare process, correct incorrect techniques in real time, and mark facial care zones to help users understand skincare details more comprehensively; S6: Traditional Chinese Medicine Constitution Identification: Utilizes a 128-dot matrix flexible piezoelectric sensor with an integrated dynamic pressure compensation unit for pulse acquisition. It also incorporates a multi-spectral ring light source and polarization filter, and uses the U-Net algorithm to segment the tongue for tongue image recognition. S7: Energy output: The magnetic guide head supports adaptive switching of radio frequency, microcurrent, red and blue light modes; S8: Cross-domain feature fusion model: Based on the Transformer architecture, it realizes the semantic alignment of pulse time series data, tongue features and skin parameters; S9: Dynamic priority algorithm: Intelligently judge the user's health status based on the results of pulse, tongue and skin tests.

2. The AI intelligent skin health management method based on multimodal data fusion according to claim 1, characterized in that: The multispectral imaging uses light of different wavelengths to image the skin, which can reveal the structure, blood vessel distribution, and pigment deposition information of the deep layer of the skin and is used to identify spots and erythema.

3. The AI intelligent skin health management method based on multimodal data fusion according to claim 1, characterized in that: The bioimpedance sensor measures the electrical conductivity of the skin to assess the skin's moisture content and physiological indicators of oil secretion, and is used to identify water-oil balance and barrier health.

4. The AI intelligent skin health management method based on multimodal data fusion according to claim 1, characterized in that: The environmental sensors include temperature and humidity sensors and pollutant sensors, which monitor the temperature and humidity of the user's environment.

5. The AI intelligent skin health management method based on multimodal data fusion according to claim 1, characterized in that: The skin problem prioritization provides the user with a prioritized list of skin problems based on the severity and potential impact of the skin problems.

6. The AI intelligent skin health management method based on multimodal data fusion according to claim 1, characterized in that: The pulse condition acquisition is used to accurately capture subtle changes in the pulse condition, and the accuracy of the 128-dot matrix flexible piezoelectric sensor reaches 0.05N.

7. The AI intelligent skin health management method based on multimodal data fusion according to claim 1, characterized in that: The multi-spectral annular light source includes different wavelengths of 450nm, 530nm and 660nm.

8. The AI intelligent skin health management method based on multimodal data fusion according to claim 1, characterized in that: The magnetic guide head supports fast function switching. The radio frequency mode is used for deep heating to promote collagen regeneration, the microcurrent mode is used for lifting and firming, and the red and blue light modes are used for anti-inflammatory and acne removal.