A radio frequency energy intelligent adjusting system based on rapid detection of skin multi-parameters

Abstract

This invention belongs to the field of radiofrequency electrotherapy technology and discloses a radiofrequency energy intelligent adjustment system based on rapid detection of multiple skin parameters. The system adopts an integrated electrode design, ensuring complete overlap between the radiofrequency action area and the parameter detection area, eliminating spatial detection errors. Combined with a flexible conductive hydrogel substrate, it can closely adhere to skin surfaces of different shapes and curvatures. Pressure adaptive adjustment ensures the stability and consistency of the detection contact. Through an intelligent time-sharing control mechanism, it completes rapid acquisition of all parameters during the radiofrequency output interval. Simultaneously, hardware-level shielding and signal optimization processing suppress the influence of radiofrequency interference on the detection signal. Based on multiple core skin physiological parameters, the system establishes a comprehensive skin radiofrequency compatibility assessment system, which can accurately determine the physiological state, tolerance, and treatment suitability range of different skin types, distinguish the functional characteristics of the superficial and deep layers of the skin, and match the corresponding energy action requirements.

Description

Technical Field

[0001] This invention belongs to the field of radiofrequency electrotherapy technology, specifically a radiofrequency energy intelligent adjustment system based on rapid detection of multiple skin parameters. Background Technology

[0002] Radiofrequency ablation uses radiofrequency energy to generate an endogenous thermal effect in skin tissue, achieving therapeutic effects such as collagen contraction and regeneration, scar repair, and inflammation reduction. The core issue is the precise control of energy output; insufficient energy renders the treatment ineffective, while excessive energy can easily cause epidermal burns, pigmentation, or even irreversible dermal damage. Existing radiofrequency energy regulation technologies suffer from the following technical problems: Many of these methods use separate detection and radiofrequency output electrodes, which do not coincide with the detection and application points. Furthermore, the output must be stopped during treatment to collect data. The detection data lags behind the real-time state of the skin, making dynamic and precise adjustment impossible and easily leading to uneven effects or thermal damage.

[0003] Most studies only collect single parameters such as skin impedance or epidermal temperature, which cannot fully reflect core physiological indicators such as epidermal barrier, collagen density, and water content. Energy regulation lacks scientific basis and has extremely poor adaptability.

[0004] Adjusting only the radio frequency power cannot coordinate the adjustment of multiple core parameters such as frequency and pulse width. Furthermore, the adjustment response cycle is long, making it unable to cope with real-time changes in the skin and easily leading to heat accumulation burns or ineffective treatment.

[0005] Setting only a single temperature threshold for power-off protection cannot predict potential risks such as excessively rapid temperature rise or sudden impedance changes. Often, by the time the protection is triggered, skin damage has already occurred, resulting in insufficient safety redundancy. Summary of the Invention

[0006] The purpose of this invention is to provide a radio frequency energy intelligent adjustment system based on rapid detection of multiple skin parameters, so as to solve one or more problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a radiofrequency energy intelligent adjustment system based on rapid detection of multiple skin parameters, comprising: Furthermore, the multi-parameter acquisition module adopts a coaxial nested flexible integrated electrode structure, with a circular radio frequency output main electrode at the center and a ring-shaped nested double-layer sensing unit around it. The inner layer is a bipolar complex impedance detection electrode, and the outer layer is an NTC temperature sensing unit and a thin film pressure sensing unit. The skin microcurrent sensing unit is integrated into the gap between the impedance detection electrodes, so that the radio frequency action point and the detection point are coaxially coincident. The electrodes use a flexible conductive hydrogel substrate, which can adaptively fit the skin surface with different curvatures such as the face, neck, and joints. Through real-time feedback from the pressure sensing unit, the electrode application pressure is automatically adjusted. The acquisition rate is 3ms / time, covering five major categories of core parameters: skin complex impedance, real-time epidermal temperature, electrode-skin contact pressure, skin equivalent conductivity, and skin microcurrent. It covers the core physiological indicators that affect the efficacy and safety of radiofrequency treatment and can identify epidermal barrier damage and skin inflammation.

[0008] Furthermore, the time-division control module, based on the raw skin data acquired by the multi-parameter acquisition module, adopts an adaptive time slot dynamic allocation algorithm, which can dynamically adjust the ratio of output time slot to detection time slot according to the radio frequency output pulse width and the rate of change of skin parameters; During the inter-pulse gap of the RF output, the hardware disconnects the RF output link and activates the electromagnetic shielding circuit, while simultaneously connecting the multi-parameter acquisition link to complete the rapid acquisition of all parameters; during the output time slot, the hardware disconnects the acquisition link and simultaneously connects the RF output link; the acquired skin parameters are transmitted to the signal preprocessing module in real time.

[0009] Adaptive time slot dynamic allocation formula: The duration of the radio frequency output time slot refers to the duration during which the system, within a complete working cycle, connects the radio frequency output link at the hardware level and releases radio frequency energy directionally to the skin. The duration of the skin parameter detection time slot refers to the time it takes for the system to quickly acquire skin physiological parameters after the system disconnects the RF link and opens the electromagnetic shield during the RF output pulse interval and connects the multi-parameter acquisition module. This represents the basic output time slot proportion coefficient, which is a fixed reference coefficient preset by the system. Its value range is strictly limited to 0.7~0.9, corresponding to the radio frequency output time slot accounting for 70%~90% of the total working cycle. This represents the RF output pulse width adjustment coefficient, which is a fixed coefficient calibrated in hardware. It is used to quantify the impact of RF output pulse width changes on time slot allocation. The wider the pulse width, the more significant the positive effect of this coefficient on the proportion of output time slots. It represents the rate of change of the RF output pulse width, which refers to the fluctuation amplitude of the RF output pulse width per unit time. It reflects the rhythmic change of RF energy output and is a key input variable for dynamically adjusting the time slot ratio. This represents the skin parameter change rate adjustment coefficient, which is a compensation coefficient calibrated by the system algorithm. It is used to quantify the impact of fluctuations in skin physiological parameters on the detection time slot requirements. The faster the skin condition changes, the stronger the effect of this coefficient on increasing the proportion of detection time slots. It represents the comprehensive rate of change of five core physiological parameters of the skin, which refers to the comprehensive fluctuation amplitude of skin complex impedance, epidermal temperature, contact pressure, equivalent conductivity and skin microcurrent per unit time, and directly reflects the speed of change of the skin's real-time state.

[0010] Furthermore, the signal preprocessing module receives multi-source skin detection signals synchronously transmitted by the time-division control module, adopts a parallel processing architecture, performs skin inflammation signal feature extraction, and can extract inflammatory factor-related features from changes in skin microcurrent and imaginary part of complex impedance, thereby realizing the identification of skin inflammation status; The core processing flow includes adaptive co-frequency interference suppression, pressure compensation normalization, rapid outlier removal, real-time temperature and time drift calibration, and inflammatory signal feature extraction. The adaptive co-frequency interference suppression uses an adaptive notch filter algorithm to eliminate residual electromagnetic interference from the radio frequency output. The pressure compensation normalization is based on the contact pressure data to establish a dynamic compensation model and eliminate signal deviations caused by different contact pressures. The rapid outlier removal uses a sliding window combined with a threshold adaptive algorithm to remove abnormal signals caused by motion artifacts and contact jitter. The real-time temperature and time drift calibration is based on a built-in standard calibration library to dynamically calibrate the temperature and time drift of the sensing unit and output calibration parameters to the skin grading module.

[0011] The real-time temperature drift calibration relies on the built-in standard calibration library. The system retrieves the standard reference value every 100ms, compares the real-time acquired sensor signal with the reference value, calculates the current drift deviation, and then uses the deviation to correct the acquired data in real time. The calibration process is performed synchronously with the acquisition of skin parameters, without taking up extra processing time, and can eliminate signal drift caused by changes in ambient temperature and long-term operation.

[0012] Furthermore, the skin grading module constructs a multi-dimensional dynamic iterative quantitative grading system for skin radiofrequency adaptability based on the calibration parameters output by the signal preprocessing module. Based on the five core parameters after preprocessing, it is divided into six evaluation dimensions: epidermal barrier state, dermal collagen state, skin moisture content, skin sensitivity, heat tolerance, and inflammation level. Each dimension is divided into four levels, ultimately forming a 24-level skin radiofrequency adaptability grading system. Employing a lightweight fuzzy logic reasoning and incremental machine learning fusion algorithm, it completes a full-dimensional hierarchical assessment, distinguishing the physiological state of the epidermis and dermis, matching the corresponding radiofrequency treatment depth and energy threshold, and identifying the radiofrequency tolerance limits and optimal treatment windows of different skin types such as sensitive skin, inflamed skin, dehydrated skin, aging skin, and skin with damaged barrier. At the same time, it adjusts the hierarchical results in real time based on the dynamic changes in skin condition during the same treatment process.

[0013] Formula for comprehensive scoring of skin radiofrequency compatibility: The comprehensive skin radiofrequency compatibility score is a quantitative assessment of skin condition by the system. The score directly corresponds to the 24-level skin radiofrequency compatibility grading results in the patent, which is used to match radiofrequency energy threshold, depth of action and treatment window. Let represent the weight coefficient of the i-th evaluation dimension, which is a weighted coefficient based on clinical data and machine learning iterations. The sum of all dimension weight coefficients satisfies =1, the weight reflects the priority of this dimension's impact on radiofrequency treatment; This represents the score of the i-th dimension. Each evaluation dimension is divided into 4 levels, with scores ranging from 1 to 4. The higher the level, the more suitable the skin condition is for radiofrequency treatment in that dimension. This is the basic quantitative value for graded evaluation. =1 indicates the state of the epidermal barrier, assesses the integrity of the skin's epidermis and the strength of its barrier function, and reflects the skin's basic ability to resist stimulation and tolerate radiofrequency energy. =2 indicates the dermal collagen status, assesses the collagen density, elasticity and regeneration capacity of the dermal layer, and determines the core effects and energy requirements of radiofrequency anti-aging and firming treatments; =3 indicates skin moisture content, which assesses the moisture content of the epidermis and dermis. The moisture value directly affects skin impedance and radiofrequency energy conduction efficiency. =4 indicates the skin sensitivity level, which assesses the intensity of the skin's response to external stimuli. The higher the sensitivity level, the lower the radiofrequency energy threshold needs to be. =5 indicates heat tolerance, which assesses the upper limit of the skin's tolerance to radiofrequency thermal effects and is a core indicator for avoiding thermal damage and setting a safe temperature threshold. =6 indicates the level of inflammation, assessing the degree of skin inflammation. In an inflammatory state, radiofrequency energy should be significantly reduced to avoid irritation and aggravation of inflammation.

[0014] Furthermore, the radio frequency adjustment module receives the skin adaptation and grading results output by the skin grading module, and combines them with the real-time skin parameters synchronously transmitted by the time-division control module. It adopts a dual closed-loop multi-dimensional adjustment architecture of feedforward prediction and feedback fine-tuning to achieve the linkage and coupling adjustment of six core parameters: radio frequency, peak power, pulse width, duty cycle, duration of action, and depth of action. The feedforward closed loop is based on the skin radiofrequency adaptability classification results and historical treatment data, matched with the built-in clinically validated radiofrequency treatment parameter library, preset the initial radiofrequency output parameters, and matched differentiated parameter combinations for different skin grades and treatment goals to lock in the safe treatment window; the feedback closed loop is based on real-time collected changes in skin parameters and performs multi-dimensional linkage fine-tuning of radiofrequency parameters. When the temperature rise rate exceeds the threshold, the duty cycle and peak power are simultaneously reduced, and the pulse width is shortened. When the skin impedance decreases, the radio frequency frequency is increased, the depth of action is increased, and the duty cycle is simultaneously optimized. For inflamed skin, the peak power is further reduced and the pulse width is extended.

[0015] Dual-loop radio frequency energy coupling regulation correction formula: This represents the real-time corrected radio frequency output energy, which is the actual radio frequency energy that the system ultimately releases onto the skin. It is the final output value after the coupling of feedforward prediction and feedback fine-tuning, and dynamically matches the real-time state of the skin. This indicates that the feedforward closed-loop preset initial radiofrequency energy is based on the skin radiofrequency compatibility grading results, the clinical validation parameter library, and historical treatment data, and locks in the basic safe treatment window; This represents the temperature rise rate correction coefficient, which is a hardware-calibrated safety compensation coefficient used to quantify the correction magnitude of the skin temperature rise rate on energy output. The faster the temperature rise, the greater the energy reduction. It represents the real-time rate of temperature rise of the epidermis, which is the increase in skin temperature per unit time. It is a core monitoring indicator for predicting heat accumulation and avoiding burns. This indicates the safe threshold for the temperature rise rate, which is a clinically calibrated safety threshold. When the real-time temperature rise rate exceeds this threshold, the system immediately reduces the radio frequency energy and triggers an early warning protection. This represents the skin impedance correction coefficient, used to quantify the impact of changes in skin complex impedance on energy conduction. The greater the impedance fluctuation, the more significant the energy correction. The reference skin complex impedance is the standard complex impedance value under healthy skin conditions, serving as a benchmark for real-time impedance comparison. This indicates real-time acquisition of skin impedance. The multi-parameter acquisition module detects skin impedance values ​​in real time, reflecting the comprehensive changes in skin moisture, collagen, and barrier status. This represents the skin inflammation correction coefficient, a compensation coefficient specific to the inflammatory state. The higher the degree of inflammation, the stronger the downregulation effect of this coefficient on energy. This represents a characteristic factor of skin inflammation, a quantitative value extracted from changes in skin microcurrent and the imaginary part of complex impedance. The value ranges from 0 to 1. The larger the value, the more severe the skin inflammation, and the more necessary it is to further reduce the radiofrequency energy.

[0016] Furthermore, the safety interlock module collects relevant parameters from the multi-parameter acquisition module, the time-sharing control module, and the radio frequency adjustment module in real time to construct a three-level multi-parameter linkage predictive safety interlock system. The hardware level is executed first, and its response priority is higher than that of other modules. Based on five parameters, namely temperature, impedance, temperature rise rate, contact pressure, and skin microcurrent, it achieves a combination of predictive protection and graded protection. The warning level is based on multi-parameter trend analysis to predict risk trends. When any parameter approaches the safety threshold, or two or more parameters deviate from the normal range, an early warning is triggered, and the radio frequency output power is automatically reduced. At the same time, the risk is indicated through the human-machine interface. The pause level is when any parameter reaches the safety threshold, or three or more parameters trigger an early warning at the same time. The hardware pauses the radio frequency output. After the skin parameters return to the safe range, the output is automatically verified and resumed. In emergency power failure mode, when there are sudden parameter changes, poor electrode contact, or risk of electric arc, the main RF power supply and auxiliary power supply are cut off, the output is locked, and manual reset and skin condition detection are required before restarting. The protection data is synchronized to the self-learning module for parameter iteration.

[0017] Furthermore, the self-learning module integrates the full-process data from the multi-parameter acquisition module, time-sharing control module, signal preprocessing module, skin grading module, radiofrequency adjustment module, and safety interlock module to construct a site-specific and effect-oriented skin radiofrequency response model. It adopts an incremental machine learning and reinforcement learning fusion algorithm to automatically collect the full-process data of each treatment after each treatment, including dynamic changes in skin parameters, full record of energy output, treatment effect feedback, and safety event record. The system incrementally iterates and updates the skin tolerance threshold, optimal treatment parameter combination, and skin response pattern. It can distinguish the skin characteristics of different parts such as the face, neck, and body, establish a skin radiofrequency response sub-model for each part, and optimize the corresponding radiofrequency parameter combination for different parts of the skin thickness, collagen density, and heat tolerance. It includes a closed-loop feedback mechanism for treatment effects, which links postoperative skin condition with treatment parameters, iteratively optimizes the parameter library, and feeds back the optimized parameters to the skin grading module and the radiofrequency adjustment module.

[0018] The beneficial effects of this invention are as follows: 1. The system of this invention adopts an integrated electrode design, which makes the radio frequency action area and the parameter detection area completely overlap, eliminating spatial detection errors. Combined with a flexible conductive hydrogel substrate, it can closely fit the skin surface with different shapes and curvatures. The pressure adaptive adjustment ensures the stability and consistency of the detection contact. Through an intelligent time-division control mechanism, all parameters are rapidly acquired during the radio frequency output gap. At the same time, through hardware-level shielding and signal optimization processing, the influence of radio frequency interference on the detection signal is suppressed, ensuring that the acquired skin physiological data truly reflects the real-time state.

[0019] 2. Based on multiple core skin physiological parameters, the system of this invention establishes a comprehensive skin radiofrequency compatibility assessment system, which can accurately identify the physiological state, tolerance and treatment compatibility range of different skin types, distinguish the functional characteristics of the skin surface and deep layers, and match the corresponding energy action requirements. It adopts a dual closed-loop intelligent adjustment architecture to realize the linkage and coupling control of the core radiofrequency working parameters, and dynamically optimize the energy output scheme according to the real-time skin status and treatment goals, so as to avoid poor treatment effect caused by insufficient energy and prevent skin damage caused by excessive energy.

[0020] 3. The present invention establishes a multi-level linkage prediction and safety protection system. With a hardware-level priority response mechanism, it realizes risk prediction and graded protection based on the trend of multi-parameter changes. It can avoid various safety risks such as thermal damage and abnormal contact in advance, and maximize the safety of the treatment process. At the same time, it integrates the treatment data of the whole process and builds a special skin response model for different body parts. Through continuous learning and iteration, it optimizes the skin tolerance threshold and the optimal combination of treatment parameters, so that the system can continuously improve the adaptation accuracy and treatment effect with repeated use. Attached Figure Description

[0021] Figure 1 This is the main flowchart of the system operation of the present invention; Figure 2 This is a flowchart of the multi-parameter skin acquisition and signal preprocessing sub-process of the present invention; Figure 3 This is a flowchart of the intelligent regulation and safety interlocking of radio frequency energy in this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] like Figures 1 to 3As shown, this embodiment of the invention provides a radiofrequency energy intelligent adjustment system based on rapid detection of multiple skin parameters, comprising: In this embodiment of the invention, the multi-parameter acquisition module adopts a coaxial nested flexible integrated electrode structure, with a circular radio frequency output main electrode at the center and a ring-shaped nested double-layer sensing unit around it. The inner layer is a bipolar complex impedance detection electrode, and the outer layer is an NTC temperature sensing unit and a thin film pressure sensing unit. The skin microcurrent sensing unit is integrated into the gap between the impedance detection electrodes, so that the radio frequency action point and the detection point are coaxially coincident, with no spatial displacement error. The coaxial nested flexible integrated electrode is fabricated using a flexible PCB substrate. The central RF output main electrode has a diameter of 8mm. The inner bipolar complex impedance detection electrode ring is 1mm wide with a 0.5mm gap between its inner and outer diameters. The outer NTC temperature sensing unit and the thin-film pressure sensing unit are uniformly distributed. Each sensing unit is 1mm × 1mm in size. The electrode layers are isolated by a polyimide insulating film with a thickness of 0.1mm, and the overall electrode thickness is 0.5mm. The skin microcurrent sensing unit is uniformly embedded in the gap between the impedance detection electrodes with a gap width of 0.3mm. All sensing units are coaxially and concentrically arranged with the RF electrode.

[0024] The coaxial nested electrode adopts a layered independent signal routing structure. The radio frequency high-voltage energy transmission line and the weak skin physiological parameter detection line are arranged in different insulating dielectric layers. A continuous and complete signal isolation grounding layer is set between the two layers of lines. The grounding layer is fully covered by high conductivity copper material, which can block the penetration interference of radio frequency electric field and magnetic field, and block the crosstalk of radio frequency high voltage signal to microampere and milliohm weak detection signals from the physical structure level.

[0025] The electrodes are made of flexible conductive hydrogel substrate, which can adaptively fit the skin surface with different curvatures such as face, neck, and joints. Through real-time feedback from the pressure sensing unit, the electrode bonding pressure is automatically adjusted to ensure the consistency and stability of electrode contact with the skin and avoid detection errors caused by poor contact. The flexible conductive hydrogel substrate is made of sodium polyacrylate and polyvinyl alcohol composite hydrogel material, with 5% by mass of conductive silver nanoparticles added to improve conductivity. The hydrogel is 2mm thick, has a Shore hardness of A10, and an elongation of ≥200%. It can adaptively conform to skin surfaces with curvature radii of 5mm-50mm, such as the face, neck, and joints. The adaptive pressure adjustment range is 5gf-20gf. The pressure sensing unit collects the pressure value in real time. When the pressure is below 5gf, the electrode bonding force is automatically increased, and when it is above 20gf, the bonding force is automatically reduced to eliminate detection deviation caused by pressure fluctuations.

[0026] The adaptive pressure adjustment is achieved through the built-in elastic micro-support bracket and micro-tension adjustment component in the electrode base. The bracket is made of medical-grade silicone elastic material, which can automatically adjust the support force according to the real-time value of the pressure sensing unit. The adjustment process is smooth and shock-free, and will not cause pressure discomfort to the skin. At the same time, it ensures that the electrode is in close contact with the skin throughout the process, and there will be no local lifting or contact suspension.

[0027] The acquisition rate is 3ms / time, covering five core parameters: skin complex impedance, real-time epidermal temperature, electrode-skin contact pressure, skin equivalent conductivity, and skin microcurrent. It covers the core physiological indicators that affect the efficacy and safety of radiofrequency treatment, and can identify epidermal barrier damage and skin inflammation. The real part of the skin complex impedance corresponds to dermal collagen density, and the imaginary part corresponds to epidermal or dermal water content. The skin microcurrent reflects the integrity of the epidermal barrier.

[0028] The five core parameters and their acquisition ranges and sensor selections are all clinically specific specifications. The skin complex impedance acquisition range is 100Ω-1MΩ, using AC bipolar impedance detection; the epidermal temperature acquisition range is 25℃-45℃, with an NTC temperature sensing unit accuracy of ±0.1℃; the contact pressure acquisition range is 0gf-50gf, with a thin-film pressure sensing unit resolution of 0.1gf; the skin equivalent conductivity acquisition range is 0.01S / m-10S / m; and the skin microcurrent acquisition range is 1nA-1μA.

[0029] The multi-parameter acquisition module is equipped with a weak signal conditioning circuit, which performs low-noise pre-amplification on weak electrical signals such as skin microcurrent and complex impedance. The amplification factor is adjustable from 100 to 1000 times. The circuit uses a low-temperature drift operational amplifier with an input bias current of less than 1 nA, which can convert weak skin signals at the nanoampere and milliohm levels into voltage signals that are compatible with the acquisition chip, while preserving the original fluctuation characteristics and trend of the signal.

[0030] In this embodiment of the invention, the time-division control module, based on the raw skin data obtained by the multi-parameter acquisition module, adopts an adaptive time slot dynamic allocation algorithm, which can dynamically adjust the ratio of output time slot to detection time slot according to the radio frequency output pulse width and the rate of change of skin parameters. The output time slot ratio is adjustable from 70% to 90%, and the detection time slot ratio is adjustable from 10% to 30%. The adaptive time slot dynamic allocation algorithm is an intelligent scheduling algorithm deeply optimized for radiofrequency treatment scenarios. The algorithm inputs are two core control variables: radiofrequency output pulse width and skin parameter change rate. The outputs are two execution variables: radiofrequency output time slot ratio and detection time slot ratio. The output time slot ratio can be adjusted from 70% to 90%, the detection time slot ratio can be adjusted from 10% to 30%, the time slot adjustment step size is fixed at 5%, and the overall response time of the algorithm is ≤2ms. The algorithm incorporates a three-level dynamic adaptation rule, with skin parameter change rates categorized into fast, medium, and slow levels. The fast level refers to a change rate ≥ 5% / ms, the medium level to a change rate of 1%-5% / ms, and the slow level to a change rate ≤ 1% / ms, corresponding to detection time slot percentages of 30%, 20%, and 10%, respectively. The RF output pulse width is divided into three levels: long, medium, and short, with longer pulse widths resulting in lower detection time slot percentages. Hardware-level link switching and time slot allocation are executed synchronously.

[0031] During the inter-pulse interval of the RF output, the hardware-level RF output link is disconnected and the electromagnetic shielding circuit is activated, while the multi-parameter acquisition link is simultaneously connected to complete the rapid acquisition of all parameters. During the output time slot, the hardware-level acquisition link is disconnected and the RF output link is simultaneously connected to avoid the influence of the fundamental and harmonic electromagnetic interference of the RF output on the detection signal. The dynamic update of the skin condition can be completed without stopping the RF output, while reducing RF energy loss and improving treatment efficiency. The acquired skin parameters are transmitted to the signal preprocessing module in real time.

[0032] The hardware-level electromagnetic shielding circuit uses a copper foil shielding layer to wrap the acquisition link and sensing unit. The grounding impedance of the shielding layer is ≤0.1Ω, which can shield the radio frequency electromagnetic interference in the 1MHz-100MHz frequency band. The radio frequency output link and the parameter acquisition link are physically isolated with a spacing of ≥5mm and no cross wiring. The link switching uses a high-speed analog switch with a switching response time of ≤0.1ms.

[0033] The hardware-level link switching uses medical-grade high-speed, low-crosstalk analog switching devices. The switching action is directly triggered by the system's main control chip. The switching sequence strictly follows the principle of disconnecting before connecting, completely disconnecting the current working link first, and then connecting the target working link. Sufficient isolation delay is set between the two links to avoid crosstalk caused by the simultaneous conduction of two signals.

[0034] The hardware-level link switching process is equipped with blank protection time slots. Between the RF output link being turned off and the detection link being turned on, and between the detection link being turned off and the RF output link being turned on, there are reserved blank protection periods with no signal transmission. The duration of the blank period is synchronized with the system clock, which can eliminate signal glitches, energy residues and electromagnetic transient interference generated at the moment of link switching.

[0035] The adaptive time slot allocation result is synchronously sent from the main control chip to the RF drive unit and the signal acquisition unit. The time slot switching command and the hardware link switching command are triggered synchronously. The three actions of RF output start / stop, acquisition link on / off, and blank protection period start follow a unified timing sequence, with no time deviation between the three.

[0036] In this embodiment of the invention, the signal preprocessing module receives multi-source skin detection signals synchronously transmitted by the time-division control module, adopts an FPGA parallel processing architecture, performs skin inflammation signal feature extraction, and can extract inflammatory factor-related features from changes in skin microcurrent and imaginary part of complex impedance, such as the amplitude of microcurrent fluctuations and the rate of impedance change caused by inflammation, so as to realize the identification of skin inflammation status. The FPGA parallel processing architecture uses a medical-grade low-power FPGA chip with 8 independent parallel processing channels, corresponding to five types of skin parameters and interference suppression, calibration, and feature extraction processes. The data processing bit width is 16 bits, and the single-channel processing latency is ≤0.5ms. The parallel channels are triggered by a data synchronization clock with a clock frequency of 50MHz to ensure that the synchronous processing of multi-source signals is free from timing deviations. It can simultaneously complete the entire process of interference suppression, parameter calibration, anomaly removal, and inflammatory feature extraction.

[0037] The inflammatory signal feature extraction adopts a two-parameter coupling correlation judgment method. The system simultaneously collects the dynamic change features of two types of signals: skin microcurrent and imaginary part of complex impedance. When the fluctuation amplitude of skin microcurrent continuously exceeds the normal stable range, and the imaginary part of complex impedance shows a continuous downward trend, the two types of features corroborate each other and the system determines that the skin is in an inflammatory state. This method can avoid misjudgment caused by interference factors such as motion artifacts and contact fluctuations, and can distinguish the signal feature differences between inflamed skin and normal skin.

[0038] The core processing flow includes adaptive co-frequency interference suppression, pressure compensation normalization, rapid outlier removal, real-time temperature and time drift calibration, and inflammatory signal feature extraction. The adaptive co-frequency interference suppression uses an adaptive notch filter algorithm to eliminate residual electromagnetic interference from the radio frequency output. The pressure compensation normalization is based on the contact pressure data to establish a dynamic compensation model and eliminate signal deviations caused by different contact pressures. The rapid outlier removal uses a sliding window combined with a threshold adaptive algorithm to remove abnormal signals caused by motion artifacts and contact jitter. The real-time temperature and time drift calibration is based on a built-in standard calibration library to dynamically calibrate the temperature and time drift of the sensing unit and output calibration parameters to the skin grading module.

[0039] The sliding window length is fixed at 5 sets of continuously acquired data, and the window step size is 1 set of acquired data, realizing real-time detection of continuous data sliding; the threshold adaptive algorithm constructs a dynamic threshold model based on the 3σ principle, using the arithmetic mean and standard deviation of the data in the window as the calculation basis, and updates the upper and lower limit judgment thresholds in real time. The threshold update frequency is synchronized with the skin parameter acquisition rate at 3ms / time. The algorithm establishes judgment rules for three typical types of interference: skin motion artifacts, electrode contact jitter, and radio frequency electromagnetic interference, respectively, for spike anomalies, drift anomalies, and fluctuation anomalies. After removing abnormal data, linear interpolation is used to complete the data sequence to ensure the continuity and stability of the preprocessed data.

[0040] The adaptive notch filter performs dedicated matching filtering for the fundamental and harmonic frequencies of the system's RF output. It tracks the current RF output frequency in real time, automatically locks the interference frequency band, and performs precise suppression. It does not filter out the effective signal components of skin physiological parameters. The signal-to-noise ratio is significantly improved after filtering, eliminating signal distortion, drift, and noise caused by RF energy leakage.

[0041] In this embodiment of the invention, the skin grading module constructs a multi-dimensional dynamic iterative quantitative grading system for skin radiofrequency adaptability based on the high-quality calibration parameters output by the signal preprocessing module. Based on the five core parameters after preprocessing, it is divided into six evaluation dimensions: epidermal barrier state, dermal collagen state, skin moisture content, skin sensitivity, heat tolerance, and inflammation level. Each dimension is divided into four levels, ultimately forming a 24-level skin radiofrequency adaptability grading system. Employing a lightweight fuzzy logic reasoning and incremental machine learning fusion algorithm, it completes a comprehensive hierarchical assessment, distinguishing the physiological state of the epidermis and dermis, matching the corresponding radiofrequency treatment depth and energy threshold, and identifying the radiofrequency tolerance limits and optimal treatment windows of different skin types such as sensitive skin, inflamed skin, dehydrated skin, aging skin, and skin with damaged barrier. At the same time, it adjusts the grading results in real time based on the dynamic changes in skin condition during the same treatment process, avoiding ineffective and overtreatment.

[0042] The lightweight fuzzy logic reasoning and incremental machine learning fusion algorithm is a dedicated AI model for skin radiofrequency adaptation grading. It adopts a five-level serial hierarchical structure with fully connected data transmission between layers, namely, an input layer, a fuzzification layer, a fuzzy rule reasoning layer, a defuzzification layer, and an incremental learning correction layer. The input layer directly connects to the output of the signal preprocessing module. The input data consists of five standardized calibration parameters: skin complex impedance, epidermal temperature, contact pressure, skin equivalent conductivity, and skin microcurrent. The input sampling frequency is consistent with the skin parameter acquisition rate at 3ms / time. The fuzzification layer maps each type of continuous numerical parameter to four fuzzy subsets: low, lower, higher, and high. The membership function adopts a Gaussian function, which fully covers the entire range of skin physiological parameters. The fuzzy rule reasoning layer has 24 clinically validated radiofrequency treatment-specific reasoning rules. The rule base is built based on dermatological clinical data, radiofrequency treatment safety specifications, and the tolerance characteristics of different skin types, which can complete multi-parameter coupled intelligent reasoning. The deblurring layer uses the centroid method to convert fuzzy values ​​into precise values, outputting 24-level quantized skin radiofrequency adaptation grading results. The grading results are directly transmitted to the radiofrequency adjustment module for initial parameter matching. The incremental machine learning module uses a lightweight single-layer linear perceptron network structure, taking real-time changes in skin parameters as input features and the deviation between the grading results and the actual skin condition as the loss function. The learning rate is fixed at 0.01. Incremental inference and weight update are performed once every 3 skin parameter acquisitions. A mini-batch iteration is completed every 10 sets of continuous data. The weight update threshold is set to 0.05 to ensure that the grading results can quickly respond to dynamic changes in the skin.

[0043] The six assessment dimensions are divided into four levels based on clinical skin physiological standards. The epidermal barrier status is graded based on the values ​​of skin microcurrent and impedance; the dermal collagen status is graded based on the real part of complex impedance; the skin moisture content is graded based on the imaginary part of complex impedance; the skin sensitivity is graded based on the amplitude of parameter fluctuations; the heat tolerance is graded based on the temperature response rate; and the inflammation level is graded based on the characteristics of microcurrent fluctuations and impedance changes. Each level corresponds to a specific skin physiological state.

[0044] In this embodiment of the invention, the radio frequency adjustment module receives the skin adaptation and grading results output by the skin grading module, and combines them with the real-time skin parameters synchronously transmitted by the time-division control module. It adopts a dual closed-loop multi-dimensional adjustment architecture of feedforward prediction and feedback fine-tuning to achieve the linkage and coupling adjustment of six core parameters: radio frequency, peak power, pulse width, duty cycle, duration of action, and depth of action. All six core radiofrequency parameters are within clinically validated ranges: radiofrequency frequency (1MHz-10MHz), peak power (1W-50W), pulse width (10ms-200ms), duty cycle (10%-90%), duration of action (1min-30min), and depth of action (0.5mm-5mm). The parameter linkage adjustment uses a synchronous triggering mechanism, so that when any parameter changes, the other related parameters respond synchronously.

[0045] The six core parameters of radiofrequency are adjusted in a coordinated manner according to a fixed coupling relationship. When the radiofrequency frequency increases, the depth of action is increased synchronously. When the peak power is adjusted, the corresponding duty cycle is matched synchronously. When the pulse width changes, the duration of action is adjusted in a linked manner. The parameters are bound together according to the clinical safety matching relationship, so there will be no energy imbalance problem caused by abnormal adjustment of a single parameter.

[0046] The feedforward closed loop is based on the skin radiofrequency adaptability classification results and historical treatment data, matched with the built-in clinically validated radiofrequency treatment parameter library, preset the initial radiofrequency output parameters, and matched differentiated parameter combinations for different skin classifications and treatment goals to lock in a safe treatment window. The treatment goals include anti-aging, firming, anti-inflammation, and scar repair. The feedback closed loop is based on real-time collected changes in skin parameters and performs multi-dimensional linkage fine-tuning of radiofrequency parameters, including parameter coupling logic. The feedforward closed loop has a built-in clinically validated radiofrequency treatment parameter library, which is constructed based on a large amount of clinical treatment data and the experience of dermatologists. It is classified and stored according to different skin types such as sensitive skin, healthy skin, inflamed skin, aging skin, and skin with damaged barrier. At the same time, it matches different treatment goals such as anti-aging, firming, anti-inflammation, and scar repair. After the skin classification results are output, the system completes the dual matching of type and goal and directly retrieves the optimal initial parameters.

[0047] The dual-loop multi-dimensional coupling regulation architecture adopts a collaborative working logic of feedforward interval setting and feedback fine adjustment. In the initial stage of treatment, the feedforward closed loop takes the lead, matching initial parameters with the skin grading results and clinical parameter database to determine the safe energy output range and avoid discomfort and ineffective treatment caused by abnormal initial energy. After the treatment enters the stable stage, the feedback closed loop takes the lead. The feedforward closed loop provides parameter benchmark support at the same time, and the feedback closed loop dynamically fine-tunes the core parameters according to real-time skin parameters.

[0048] When the rate of temperature rise exceeds the threshold, the duty cycle and peak power are simultaneously reduced, and the pulse width is shortened to avoid heat accumulation. When the skin impedance decreases, i.e. the water content increases, the radio frequency frequency is increased and the depth of action is increased. The duty cycle is simultaneously optimized to achieve dynamic adaptation of energy output to the real-time state of the skin. For inflamed skin, the peak power is further reduced and the pulse width is extended to reduce inflammatory stimulation.

[0049] The dual-loop multi-dimensional adjustment architecture is equipped with an intelligent adjustment algorithm. The feedforward closed-loop algorithm performs feature matching based on skin grading results and historical treatment data, and retrieves the optimal initial parameters from the clinically validated parameter library, with a matching response time of ≤1ms. The feedback closed-loop algorithm adopts PID incremental adjustment logic with a proportional coefficient of 0.8, an integral coefficient of 0.15, and a derivative coefficient of 0.05, which can complete multi-parameter linkage fine-tuning according to real-time changes in skin parameters, with an adjustment accuracy of ±1%. The algorithm establishes specific adjustment strategies for different abnormal skin conditions: when the temperature rise rate exceeds the threshold, the duty cycle and peak power are reduced and the pulse width is shortened simultaneously; when the skin impedance decreases, the radio frequency frequency is increased, the depth of action is increased, and the duty cycle is optimized simultaneously; in the inflammatory state, the peak power is further reduced by 30% and the pulse width is extended by 20%.

[0050] In this embodiment of the invention, the safety interlock module collects relevant parameters from the multi-parameter acquisition module, the time-sharing control module, and the radio frequency adjustment module in real time to construct a three-level multi-parameter linkage and prediction safety interlock system. The hardware level is executed first, and its response priority is higher than that of other modules. Based on five parameters, namely temperature, impedance, temperature rise rate, contact pressure, and skin microcurrent, it realizes the combination of predictive protection and graded protection. The safety thresholds for the three-level interlocking core parameters have all been clinically validated and set. The safety thresholds are: skin temperature 43℃, temperature rise rate 1℃ / s, skin impedance mutation threshold ±50%, contact pressure 5gf-20gf, and skin microcurrent mutation threshold ±50%. The warning level trigger condition is when the parameter is close to 80% of the threshold or two parameters deviate from the normal range. The pause level trigger condition is when the parameter reaches 100% of the threshold or three parameters are in warning. The emergency power-off level trigger condition is when the parameter exceeds the threshold by 120%, electrode contact detachment, or arc risk.

[0051] The early warning level is based on multi-parameter trend analysis, such as a continuous increase in temperature rise rate, a continuous decrease in impedance, and abnormal microcurrent fluctuations, to predict risk trends. When any parameter approaches the safety threshold, or two or more parameters deviate from the normal range, an early warning is triggered, and the RF output power is automatically reduced. At the same time, the risk is indicated through the human-machine interface to avoid thermal damage in advance. The pause level is when any parameter reaches the safety threshold, or three or more parameters trigger an early warning at the same time, the RF output is paused at the hardware level. After the skin parameters return to the safe range, the output is automatically verified and resumed. Risk trend prediction adopts a continuous parameter change rate tracking method. Based on five consecutive sets of collected data, the system continuously monitors the direction, rate and amplitude of parameter changes. When the parameter shows a unidirectional upward or downward trend that is rapidly approaching the safety threshold, the system triggers the corresponding level of protection action in advance to avoid safety risks such as thermal damage and abnormal contact.

[0052] In emergency power-off mode, when there are sudden parameter changes, poor electrode contact, or risk of electric arc, the main RF power supply and auxiliary power supply are cut off, and the output is locked. The parameter changes include sudden impedance drop, sudden temperature rise, and sudden change in microcurrent. Manual reset and skin condition detection are required before restarting. The protection data is synchronized to the self-learning module for parameter iteration.

[0053] The three-level multi-parameter linkage prediction safety interlock system is equipped with a risk prediction AI algorithm. Based on five types of parameters—temperature, impedance, temperature rise rate, contact pressure, and skin microcurrent—it performs time-series feature analysis and uses the sliding window mean method to predict parameter change trends. The window length is 5 sets of data, and the trend prediction response time is ≤1ms. The trigger thresholds for warning, pause, and emergency power-off levels are all clinically calibrated. The warning threshold is 80% of the safety threshold, the pause threshold is 100%, and the emergency power-off threshold is 120%. The algorithm can identify emergency situations such as parameter mutations, poor contact, and arc risk. The parameter mutation judgment threshold is a single-collection change rate ≥50%, and the hardware-level cutoff response time is ≤0.5ms, minimizing skin thermal damage.

[0054] When the warning level is executed, the system will simultaneously issue an audible and visual alert to remind the operator of abnormal skin condition; when the pause level is executed, the radio frequency output will completely stop, while maintaining parameter detection and data acquisition functions, waiting for the skin parameters to recover automatically; after the emergency power failure level is executed, the system will lock all operating interfaces and output functions, and the screen will display the cause of the fault. Only after the risk is eliminated, the skin condition detection is completed, and the system verification is passed can the lock be released and the system restarted.

[0055] The execution command of the hardware-level safety interlock is directly connected to the radio frequency output control circuit without going through the software algorithm parsing and data forwarding stage. When a risk signal is detected, the hardware circuit directly cuts off the radio frequency drive signal. The response priority is higher than all software control commands of the system. Even if the software runs abnormally, the hardware interlock circuit can still independently complete the protection action.

[0056] In this embodiment of the invention, the self-learning module integrates the full-process data from the multi-parameter acquisition module, time-sharing control module, signal preprocessing module, skin grading module, radiofrequency adjustment module, and safety interlock module to construct a site-specific and effect-oriented skin radiofrequency response model. It adopts an incremental machine learning and reinforcement learning fusion algorithm to automatically collect the full-process data of each treatment after each treatment, including dynamic changes in skin parameters, full record of energy output, treatment effect feedback, and safety event record. The treatment effect feedback includes the user's postoperative skin condition score and clinical effect evaluation data. The system uses a built-in Flash chip to store the entire treatment process data. The data storage format is standardized time-series text. Each data entry contains five types of information: timestamp, skin parameters, radiofrequency parameters, safety status, and effect score. The storage capacity can record 1,000 complete treatment data entries. The self-learning iteration is triggered when a single treatment is completed normally without any emergency power outages. The model adopts an incremental overwrite saving mechanism, generating a new version of the model with each iteration and automatically retaining the last 10 versions of the model for rollback verification.

[0057] Incremental machine learning employs effective data selection and update rules. The system automatically selects treatment data that has no safety warnings, no interruptions or power outages during the treatment process, and that has achieved the desired postoperative results and has a stable skin condition for model iteration. It actively removes treatment data that has abnormal interruptions, poor results, or deviated parameters to avoid bad data interfering with model accuracy and the direction of parameter library optimization.

[0058] The system incrementally iteratively updates the skin tolerance threshold, optimal treatment parameter combination, and skin response pattern. It can distinguish the skin characteristics of different parts of the face, neck, and body. The face includes the forehead, cheekbones, and jawline, and the body includes the abdomen and arms. It establishes a site-specific skin radiofrequency response sub-model and optimizes the corresponding radiofrequency parameter combination for different parts of the skin thickness, collagen density, and heat tolerance. It includes a closed-loop feedback mechanism for treatment effects, which links postoperative skin condition with treatment parameters, iteratively optimizes the parameter library, and feeds back the optimized parameters to the skin grading module and the radiofrequency adjustment module.

[0059] The site-specific skin radiofrequency response model is the core AI self-learning model of the system, adopting a four-layer modular parallel and serial hybrid structure, namely, feature extraction layer, site sub-model layer, strategy optimization layer, and feedback iteration layer. The feature extraction layer takes into account the entire treatment process data, including four core features: skin parameter time sequence, radiofrequency output parameter sequence, safety interlock event markers, and postoperative skin status score. It first performs filtering, normalization, and dimensionality unification preprocessing to eliminate differences in dimensions and values. The site sub-model layer is divided into three independent parallel sub-networks: face, neck, and body. Each sub-network adopts a lightweight convolutional neural network structure to adapt to the differentiated characteristics of skin thickness, collagen density, and heat tolerance in different sites. The batch size is set to 8, and the number of iterations per site is 100. The strategy optimization layer adopts the Q-learning reinforcement learning algorithm, with the weighted sum of the radiofrequency treatment safety coefficient and treatment effect score as the reward function. The safety coefficient has a weight of 0.6, the effect score has a weight of 0.4, the algorithm learning rate is 0.02, the discount factor is 0.9, the exploration rate is 0.1, and the reward value ≥ 0.8 is judged as the optimal parameter and stored in the database. The specific steps for model training are as follows: collect data from the entire process of a single treatment to construct a sample window; generate standardized feature vectors through the feature extraction layer; train sub-models by dividing the dataset according to body parts; optimize the decision-making strategy for radiofrequency parameters through reinforcement learning; immediately after a single treatment, perform incremental iterations to update the tolerance threshold and optimal parameters; complete model accuracy verification every 10 iterations, retain the iteration results if the verification accuracy is ≥95%, and roll back to the previous stable model if it is below 95%; the feedback iteration layer synchronously pushes the optimized parameter library to the skin grading module and the radiofrequency adjustment module.

[0060] The facial sub-model is adapted to features of thin skin, high sensitivity, and shallow collagen distribution, focusing on low power, high precision, and shallow depth of action parameters; the neck sub-model is adapted to features of loose skin, low collagen density, and moderate tolerance, focusing on gentle energy and moderate depth of action parameters; the body sub-model is adapted to features of thick skin, high tolerance, and deep collagen distribution, focusing on sufficient energy and deep depth of action parameters. The three sub-models are independent of each other.

[0061] The parameter libraries for the three sub-models of the face, neck, and body are stored independently. The system automatically calls the corresponding sub-model by recognizing the electrode contact position. When switching, the system directly loads the tolerance threshold, optimal parameter combination, and safety threshold of that part.

[0062] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0063] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A radiofrequency energy intelligent adjustment system based on rapid detection of multiple skin parameters, characterized in that, include: The multi-parameter acquisition module adopts a coaxial nested flexible integrated electrode structure, so that the radio frequency action point and the detection point are coaxially coincided. Through the flexible conductive hydrogel substrate, it adaptively adheres to skin with different curvatures and adjusts the adhesion pressure to collect five core physiological parameters: skin complex impedance, epidermal temperature, contact pressure, skin equivalent conductivity, and skin microcurrent. Based on these physiological parameters, the epidermal barrier state and skin inflammation state are identified. The time-sharing control module, based on the raw skin data acquired by the multi-parameter acquisition module, uses an adaptive time slot dynamic allocation algorithm to achieve real-time synchronization between radio frequency output and parameter acquisition; The signal preprocessing module receives synchronously transmitted multi-source skin detection signals, and completes signal interference suppression, calibration and optimization through parallel processing architecture and adaptation algorithm, and outputs calibration parameters. The skin grading module, based on preprocessed calibration parameters, constructs a multi-dimensional quantitative grading system for skin radiofrequency adaptability to complete the grading assessment of skin condition; The radio frequency modulation module combines skin grading results with real-time skin parameters and adopts a dual-closed-loop multi-dimensional coupling modulation architecture to achieve linkage and adaptive adjustment of core radio frequency parameters, dynamically match different treatment goals, and ensure that energy output is adapted to the real-time state of the skin. The safety interlock module collects relevant parameters from each module in real time, constructs a three-level multi-parameter linkage and prediction safety interlock system, realizes graded protection based on core parameters, avoids thermal damage through risk trend prediction, and synchronizes protection data to the self-learning module for system parameter iteration. The self-learning module integrates data from all modules throughout the entire process, constructs a site-specific and effect-oriented skin radiofrequency response model, and iteratively updates the skin tolerance threshold and optimal treatment parameters through a fusion learning algorithm, optimizing parameter combinations for different skin characteristics.

2. The radiofrequency energy intelligent adjustment system based on rapid multi-parameter detection of skin according to claim 1, characterized in that, The multi-parameter acquisition module has a coaxial nested flexible integrated electrode structure with a circular radio frequency output main electrode at the center and a ring-shaped nested double-layer sensing unit around it. The inner layer is a bipolar complex impedance detection electrode, and the outer layer is an NTC temperature sensing unit and a thin film pressure sensing unit. The skin microcurrent sensing unit is integrated into the gap between the impedance detection electrodes to achieve coaxial coincidence between the radio frequency action point and the detection point. The electrodes use a flexible conductive hydrogel substrate, which can adaptively conform to the skin surface with different curvatures, including the face, neck, and joints. The collected parameters cover the core physiological indicators that affect the effectiveness and safety of radiofrequency treatment, and can identify epidermal barrier damage and skin inflammation.

3. The radiofrequency energy intelligent adjustment system based on rapid multi-parameter detection of skin according to claim 2, characterized in that, The adaptive time slot dynamic allocation algorithm of the time-division control module can dynamically adjust the ratio of output time slots to detection time slots according to the RF output pulse width and the rate of change of skin parameters. During the inter-pulse gap of the RF output, the hardware cuts off the RF output link and opens the electromagnetic shielding circuit, and simultaneously connects the multi-parameter acquisition link to complete parameter acquisition. During the output time slot, the hardware cuts off the acquisition link and simultaneously connects the RF output link, and the acquired skin parameters are transmitted to the signal preprocessing module in real time.

4. The radiofrequency energy intelligent adjustment system based on rapid multi-parameter detection of skin according to claim 3, characterized in that, The parallel processing architecture of the signal preprocessing module can perform skin inflammation signal feature extraction, extracting inflammatory factor-related features from changes in skin microcurrent and imaginary part of complex impedance, and realizing the identification of skin inflammation status. The core processing flow includes adaptive co-frequency interference suppression, pressure compensation normalization, rapid outlier removal, real-time temperature drift and time drift calibration, and inflammatory signal feature extraction, which can eliminate signal deviation and abnormal interference, and output calibration parameters to the skin grading module.

5. The radiofrequency energy intelligent adjustment system based on rapid multi-parameter detection of skin according to claim 4, characterized in that, The multi-dimensional skin radiofrequency adaptability grading system of the skin grading module is based on five pre-processed core parameters, divided into six assessment dimensions: epidermal barrier state, dermal collagen state, skin moisture content, skin sensitivity, heat tolerance, and inflammation level. Each dimension is divided into four levels, forming a 24-level skin radiofrequency adaptability grading system. It adopts a lightweight fuzzy logic reasoning and incremental machine learning fusion algorithm to complete the full-dimensional grading assessment, which can distinguish the physiological state of the epidermis and dermis, match the corresponding radiofrequency treatment depth and energy threshold, identify the radiofrequency tolerance limit and optimal treatment window of different skin types, and adjust the grading results in real time according to the dynamic changes in skin state during treatment.

6. The radiofrequency energy intelligent adjustment system based on rapid multi-parameter detection of skin according to claim 5, characterized in that, The dual-closed-loop multi-dimensional coupled adjustment architecture of the radio frequency modulation module is a dual-closed-loop architecture combining feedforward prediction and feedback fine-tuning. It can realize the linkage and coupling adjustment of six core parameters: radio frequency, peak power, pulse width, duty cycle, duration of action, and depth of action. The feedforward closed loop is based on the skin radio frequency adaptability classification results and historical treatment data, matches the built-in clinically validated radio frequency treatment parameter library, presets the initial radio frequency output parameters, and locks the safe treatment window. The feedback closed loop performs multi-dimensional linkage fine-tuning of radio frequency parameters based on real-time collected changes in skin parameters.

7. The radiofrequency energy intelligent adjustment system based on rapid multi-parameter detection of skin according to claim 6, characterized in that, The three-level, multi-parameter linkage and predictive safety interlock system of the safety interlock module is executed at the hardware level with a higher response priority than other modules. Based on five core parameters—temperature, impedance, temperature rise rate, contact pressure, and skin microcurrent—it combines predictive protection with graded protection. The graded protection includes warning level, pause level, and emergency power-off level, which can trigger corresponding protective actions according to the degree of parameter abnormality. The protection data is synchronized to the self-learning module for parameter iteration.

8. The radiofrequency energy intelligent adjustment system based on rapid multi-parameter detection of skin according to claim 7, characterized in that, The self-learning module employs a fusion algorithm of incremental machine learning and reinforcement learning. After each treatment, it automatically collects data from the entire treatment process and incrementally updates the skin tolerance threshold, optimal treatment parameter combination, and skin response patterns. It can distinguish the skin characteristics of different areas, including the face, neck, and body, and establish a site-specific skin radiofrequency response sub-model. It optimizes the corresponding radiofrequency parameter combination for different areas based on skin thickness, collagen density, and heat tolerance. A treatment effect feedback loop is provided, which associates the postoperative skin condition with the treatment parameters, iteratively optimizes the parameter library, and feeds back the optimized parameters to the skin grading module and the radiofrequency adjustment module.

9. The radiofrequency energy intelligent adjustment system based on rapid multi-parameter detection of skin according to claim 8, characterized in that, The multi-parameter acquisition module automatically adjusts the contact pressure between the electrode and the skin through real-time feedback from the thin-film pressure sensing unit; the parameter acquisition rate can be preset and adjusted to achieve full acquisition of five core physiological parameters.

10. The radiofrequency energy intelligent adjustment system based on rapid multi-parameter detection of skin according to claim 9, characterized in that, The feedback closed loop of the radio frequency adjustment module can perform linkage adjustment according to the real-time changes in skin parameters. When the temperature rise rate exceeds the threshold, the duty cycle and peak power are reduced simultaneously, and the pulse width is shortened. When the skin impedance decreases, the radio frequency is increased, the depth of action is increased, and the duty cycle is optimized simultaneously. For skin inflammation, the peak power is further reduced and the pulse width is extended.

Images