Control method and system of intense pulsed light therapeutic instrument based on neural network

Abstract

The invention belongs to the technical field of control of intense pulsed light therapeutic instruments, and particularly relates to a control method and system of an intense pulsed light therapeutic instrument based on a neural network model. The method comprises the steps that real-time state information of a to-be-processed skin area of a patient is collected; generating and applying a detection stimulation signal based on the real-time state information; monitoring a patient reaction to form reaction data and generating a state evaluation result based on the reaction data; generating a treatment signal input parameter based on the state evaluation result; outputting a treatment signal to the to-be-treated skin area according to the treatment signal input parameter, and collecting a treatment feedback result after treatment is completed; and updating the neural network model based on the treatment feedback result. According to the invention, a pre-treatment safety screening mechanism is established by applying the detection stimulation signal, the output duration of the treatment signal is dynamically controlled based on the reaction data, and then the neural network model is updated by using the treatment feedback result, so that the improvement of the treatment safety and accuracy and the adaptive optimization of the control method are realized.

Description

Technical Field

[0001] This invention belongs to the field of control technology for intense pulsed light therapy devices, specifically relating to a control method and system for an intense pulsed light therapy device based on a neural network. Background Technology

[0002] Intense pulsed light (IPL) technology is used in cosmetic procedures such as skin rejuvenation and treatment of pigmented lesions. The effectiveness and safety of this technology depend on the precise control and delivery of energy. While achieving selective photothermal effects on the target tissue, it protects the surrounding normal skin tissue from thermal damage. Therefore, ensuring that every step of the treatment process is accurately executed and evaluated is of significant industry value in guaranteeing treatment results, improving patient satisfaction, and mitigating medical risks.

[0003] In the pre-treatment preparation stage of current intense pulsed light (IPL) therapy, there is a lack of suitable assessment methods for the application status of media such as cooling gels. Operators rely solely on experience to judge whether the application is complete, and cannot effectively confirm the tightness of the adhesion between the cooling gel and the skin, the uniformity of coverage, or whether sufficient pre-cooling effect has been achieved. This leads to the treatment being initiated when the adhesion is poor or the cooling is insufficient, and the local skin will face the risk of thermal damage due to the lack of effective thermal buffering, causing adverse reactions such as burns, blisters, and pigmentation. The parameter settings and execution process of existing treatment plans exhibit static and subjective characteristics. Core parameters such as treatment energy and pulse width are mostly preset based on broad skin type classifications, or fine-tuned by operators during treatment based on subjective observation. This cannot respond in real time to the dynamic physiological changes of the patient's skin under photothermal effects, thus limiting the personalization and precision of treatment plans.

[0004] In view of this, the present invention provides a control method and system for a neural network-based intense pulsed light therapy device. Summary of the Invention

[0005] The purpose of this invention is to provide a control method and system for a neural network-based intense pulsed light therapy device, which can realize the step-by-step assessment of the patient and the accurate identification of the patient's condition during the assessment process. At the same time, a set of signal inputs formed after the accurate identification can also be used to guide the working time and status of the device.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a control method for an intense pulsed light therapy device based on a neural network model, which, in response to the state assessment results generated based on the patient's skin area to be treated, executes: Treatment signal input parameters are generated based on the state assessment results. Treatment signals are output to the skin area to be treated according to the treatment signal input parameters. Treatment feedback results are collected after treatment. The neural network model is updated based on the treatment feedback results. The updated neural network model is used to generate detection stimulus signals in the subsequent control method execution. The process of responding to the status assessment results generated based on the patient's skin area to be treated includes: collecting real-time status information of the skin area to be treated; generating detection stimulus signals using a neural network model based on the real-time status information; applying the detection stimulus signals to the skin area to be treated; monitoring the patient's response to generate response data during the application of the detection stimulus signals; and generating status assessment results based on the response data.

[0007] Preferably, generating the detection stimulus signal includes: matching real-time status information with preset skin condition judgment criteria, and generating a detection stimulus signal as a non-therapeutic signal when the matching is successful; A successful match includes situations where the real-time status value extracted from the real-time status information is less than the classification measurement value, or the real-time status value is greater than the classification measurement value but the value of the status parameter does not exceed the comparison threshold.

[0008] Preferably, in response to a status assessment result generated based on the patient's skin area to be treated, the method further includes: If the duration of the response exceeds the evaluation criteria after the application of the detection stimulus signal, the state assessment result is generated as an abnormal type; when the state assessment result is an abnormal type, the output of the treatment signal is blocked.

[0009] Preferably, generating treatment signal input parameters based on the state assessment results includes: when the state assessment result is normal, acquiring the patient's baseline response state, and starting from the time point corresponding to the baseline response state, sequentially setting multiple assessment time periods, comparing the response data with the comparison parameters within each assessment time period to determine the normal output node or the stop output node, and generating treatment signal input parameters containing the normal output time period based on the normal output node and the stop output node.

[0010] Preferably, updating the neural network model based on the state assessment results includes: determining the treatment state as effective feedback, excessive feedback, or insufficient feedback based on the treatment signal intensity and the patient's response state; and adjusting the treatment signal input parameters used to generate subsequent treatment signals when the treatment state is determined to be excessive feedback or insufficient feedback.

[0011] The control system of the intense pulsed light therapy device based on a neural network model includes: The status assessment module is used to generate status assessment results for the patient's skin area to be treated. The treatment parameter generation module is configured to generate treatment signal input parameters in response to the state assessment results generated by the state assessment module. The treatment execution module is used to output treatment signals based on the treatment signal input parameters and to collect treatment feedback results; And a model update module, used to update the neural network model based on the treatment feedback results, wherein the updated neural network model is used by the state evaluation module to generate detection stimulus signals in subsequent control method execution.

[0012] Preferably, the status assessment module is configured to: acquire real-time status information of the skin area to be treated, generate and apply detection stimulus signals using a neural network model based on the real-time status information, and monitor the patient's response to generate a status assessment result.

[0013] Preferably, after the detection stimulus signal is applied, if the duration of the response exceeds the evaluation criteria, the state evaluation result is generated as an abnormal type; when the state evaluation result is an abnormal type, the output of the treatment signal is blocked.

[0014] Preferably, when the state assessment result is normal, the normal output node and the stop output node are determined based on the analysis of the response data to generate treatment signal input parameters that include the normal output period.

[0015] Preferably, the model update module is configured to: determine the treatment status based on the treatment signal intensity and patient response status contained in the treatment feedback results, and adjust the treatment signal input parameters used to generate subsequent treatment signals when the treatment status is excessive or insufficient feedback.

[0016] Beneficial effects

[0017] 1. Before outputting the treatment signal, the present invention first generates and applies a detection stimulus signal based on real-time status information, and monitors the patient's response to generate a status assessment result. If the status assessment result is abnormal, the output of the treatment signal is blocked. Then, a pre-treatment safety screening mechanism is established. By assessing the patient's response to the detection stimulus signal before treatment, patients with potential adverse reactions can be identified in advance and the output of the treatment signal can be blocked, thereby improving the safety of treatment.

[0018] 2. When generating treatment signal input parameters, this invention monitors reaction data and determines normal output nodes and stop output nodes based on the matching results of reaction data and comparison parameters. This generates treatment signal input parameters that include normal output periods, enabling dynamic and refined control of the duration of treatment signal output. By establishing a real-time termination mechanism based on patient response, it ensures that treatment energy is applied only within the patient's tolerance window, avoiding tissue overheating damage that may result from fixed-duration output.

[0019] 3. This invention collects treatment feedback results after treatment and updates the neural network model based on these results. The updated neural network model is then used to generate detection stimulus signals in subsequent control methods, thus forming a closed-loop feedback control system. By learning from each treatment feedback result, the neural network model can gradually correct its ability to assess skin condition and predict patient response, enabling the entire control system to automatically optimize treatment plans for specific patients over time, thereby achieving continuous improvement in treatment effectiveness and safety. Attached Figure Description

[0020] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system module diagram of the present invention. Detailed Implementation

[0021] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. The following are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention, which is defined by the appended claims and their equivalents.

[0022] Example 1: Please refer to Figure 1 As shown in the figure, this embodiment provides a control method for an intense pulsed light therapy device, which is used to dynamically identify and adjust the patient's skin condition, enabling personalized and precise control of pulsed light therapy. The method specifically includes the following steps: The process involves acquiring surface or temperature distribution images of the patient's skin area to be treated using an image acquisition device. These images contain visual features such as skin color, texture, and capillary distribution. The images are then input into a pre-defined image feature extraction and quantization process. This process performs a series of pre-defined calculations on the raw image data, converting it into structured numerical information. This process is based on a sample library containing a large number of labeled skin images. It uses algorithms to identify specific visual features in the images and establish a mapping relationship between image visual features and skin physiological indicators. Through this process, the acquired images are transformed into a set of structured real-time status information, which includes key indicators that characterize the skin's physiological state at the current moment, such as melanin index, hemoglobin index, and moisture content, derived from the skin images through the image feature extraction and quantization process. Then, the real-time status information is input into the data rationality verification process to verify the data quality filtering steps. Specifically, it can be judged by preset rules to check whether the data was generated within the preset time window to ensure the timeliness of the data. Then, it is checked whether all necessary indicators are complete, and values ​​that exceed the normal physiological range are removed to ensure the accuracy and reliability of subsequent calculations. By filtering out distorted or unrepresentative data, valid real-time status information is output as the basis for subsequent processing.

[0023] Furthermore, a detection stimulus signal is generated and applied to detect the skin's sensitivity or response threshold. Based on effective real-time status information and preset skin condition judgment criteria, it is determined whether it is appropriate to apply stimulation. The skin's immediate response threshold is detected by applying a safe, non-therapeutic signal. The detection stimulus signal is a non-therapeutic signal with energy below the preset therapeutic threshold. Its purpose is to induce an observable physiological response in the skin, rather than to produce a therapeutic effect. Specifically, the decision-making process for generating the signal is as follows: extract real-time state values ​​from real-time state information, such as the currently measured melanin index, and perform preliminary classification of a core skin physiological indicator through classification measurement values. When the real-time state value is less than the preset classification measurement value, such as the melanin index level being "low" or "medium", the match is considered successful. If the real-time state value is greater than or equal to the classification measurement value, then further obtain the state parameters associated with the classification measurement value. That is, when the preliminary judgment does not meet the conditions, auxiliary skin physiological indicators need to be introduced to conduct a more refined and safer secondary assessment of the skin condition. Then, the value is compared with a preset comparison threshold. The system can make a final judgment only when the relationship between the measured value of the status parameter and the threshold meets the preset conditions. Specifically, if the value of the status parameter does not exceed the comparison threshold, such as when the moisture content is higher than the preset dryness threshold, the matching is considered successful. If the value of the status parameter exceeds the comparison threshold, the matching is considered unsuccessful, and the current processing flow is terminated, thereby avoiding any stimulation to the skin in a potentially fragile state. After a successful matching, a detection stimulation signal is generated and then applied to the skin area to be treated.

[0024] Furthermore, patient responses are assessed: during the application of the detection stimulus signal, the patient's response is continuously monitored using devices such as temperature sensors and blood flow monitors to generate response data. Based on the response data, the data is analyzed and a status assessment result is generated, which is specifically divided into normal and abnormal types, to be used to decide on subsequent operational procedures. Specifically, the first assessment time required for this assessment is first calculated, which is the total dynamic calculation time for assessing the skin's response to the detection stimulus signal. The starting point is the start of the stimulus, and the ending point is the moment when the skin response tends to stabilize or reaches the safe upper limit. To achieve this, the increment of the patient's response is continuously calculated according to a fixed time interval, i.e., assessment unit, for example, each unit is 5 seconds. Such as the change in the rate of increase of skin surface temperature. If the increment is not lower than the preset increment threshold, the assessment time is continuously accumulated. It is particularly important to note that the assessment should be terminated in any of the following situations: 1. Within a predetermined number of consecutive assessment units, such as three consecutive assessment units, the increment of the response is lower than the increment threshold; 2. When the cumulative value of the response reaches the maximum cumulative value that serves as the safety upper limit, the total cumulative duration at this point is determined as the first assessment duration. Within the determined first assessment duration, the duration of the response is judged, that is, the length of time that the measured patient response exceeds the preset physiological baseline. Specifically, a set of evaluation criteria is first built in, which is the upper limit of the allowable duration of the response. When the duration of the response does not exceed the evaluation criteria, the status evaluation result is generated as normal. If the duration of the response exceeds the evaluation criteria, the status evaluation result is generated as abnormal. At this time, it is necessary to prevent the output of subsequent treatment signals and issue a prompt to the operator.

[0025] Furthermore, when the status assessment result is normal, the input parameters for generating the treatment signal are determined. First, the stable physiological data of the patient in the resting state before the application of the detection stimulus signal are acquired, and the value is determined as the patient's baseline response state. Starting from the time point corresponding to the baseline response state, multiple assessment time periods with fixed durations are set sequentially. During the treatment, a continuous time window with a fixed duration is used to periodically monitor the patient's response, for example, each segment is 10 seconds. Then, a comparison parameter is set, which is calculated by adding a preset safety tolerance to the value of the baseline response state. During the treatment, a safety threshold for judging whether the skin is overheated in real time is used. The calculation method is to add a preset "safety tolerance" to the value of the "baseline response state", for example, the baseline temperature plus 0.5 degrees Celsius. In each assessment time period, the real-time collected response data is compared with the comparison parameter. When the response data does not exceed the comparison parameter, the current time point is recorded as a normal output node, indicating that it is safe to continue applying treatment at this time. To avoid misjudgments caused by instantaneous data fluctuations, we will adopt a fault-tolerant judgment logic: when the reaction data exceeds the number of times the comparison parameters are exceeded in a certain number of consecutive evaluation time periods, the output node is determined to stop, so as to filter out false alarms caused by instantaneous noise or fluctuations. That is, the first time point in this set of evaluation time periods that exceeds the standard is determined. Then, based on the recorded normal output node and the determined stop output node, the treatment signal input parameters containing the normal output period are generated, specifically including the allowable output time period determined after real-time safety calculation and the corresponding energy setting. The time interval between a normal output node and a stopped output node is defined as the effective output period. Specifically, it is a theoretically safe treatment time window dynamically calculated based on real-time skin reaction data. Its starting point is a normal output node, and its ending point is a stopped output node. Then, to ensure absolute treatment safety, the allowable output period for the current treatment operation is calculated based on the effective output period and the maximum total output period set by the operator. This is used to finally issue the actual usable treatment duration to the execution module. The maximum total output period is the upper limit of the total duration of a single treatment set according to clinical guidelines. The final allowable output period is the smaller value between the effective output period and the remaining amount of the maximum total output period. This ensures that during treatment, the skin's real-time tolerance is adapted to while adhering to the overall safety procedures.

[0026] Furthermore, the treatment is performed and feedback is collected based on the allowed output time period and energy settings included in the treatment signal input parameters. The treatment signal is output to the skin area to be treated. After the treatment is completed, the treatment feedback results are collected to evaluate the effect of the treatment. The treatment feedback results include the treatment signal intensity and the patient's response status. The treatment signal intensity corresponds to the total energy actually output in this treatment; the patient's response status corresponds to the objective state of the skin for a period of time after the treatment, such as the erythema area and depth quantified through image analysis. Then, based on the intensity of the treatment signal and the patient's response status, a comprehensive feedback value is calculated to standardize the scalar value for evaluating treatment efficiency, reflecting the magnitude of the biological effect produced by a unit of energy input. In order to normalize the evaluation of the treatment effect, this calculation process can be implemented through a pre-set calculation method, such as dividing the quantitative value representing the patient's response status by the actual output of the treatment energy in this treatment, thereby obtaining a value representing the degree of response caused by a unit of energy. Then, the comprehensive feedback value is compared with a preset normal range, which is defined by an upper threshold and a lower threshold, representing the ideal treatment effect range. That is, the comprehensive feedback value in this range means that the treatment has achieved the expected effect. Specifically, when the comprehensive feedback value is within the normal range, the treatment status is judged as effective feedback, indicating that the treatment effect meets expectations. When the comprehensive feedback value is higher than the upper threshold of the normal range, the treatment status is judged as excessive feedback, indicating that the skin reaction caused by a unit of energy is too strong. If the comprehensive feedback value is lower than the lower threshold of the normal range, the treatment status is judged as insufficient feedback, indicating that the skin reaction caused by a unit of energy has not achieved the expected effect.

[0027] Furthermore, the parameter generation rules are updated: by taking the treatment signal input parameters, response data, and treatment feedback results that are ultimately judged as effective, excessive, or insufficient feedback from each treatment as a new set of associated data, a set of preset parameter association relationships for generating control parameters is updated. This defines the mapping rules between the input "real-time status information" and the output "treatment signal input parameters". This relationship can be dynamically updated by the treatment feedback results. The update process adjusts the correspondence between the initial state information and the optimal treatment parameters based on the latest treatment practice feedback. For example, when a treatment is judged to be "overfeeding", the update process will adjust the parameter correlation so that when a similar initial skin state is encountered in the future, the generated treatment energy will be reduced accordingly. In particular, when the treatment state is judged to be overfeeding or underfeeding, a correction process is immediately initiated. Based on the degree of deviation of the current feedback, a correction coefficient is calculated and applied to the subsequent treatment parameters to achieve rapid closed-loop correction. The correction process is an immediate closed-loop adjustment mechanism. Based on the parameter correlation, it analyzes the treatment signal input parameters that caused the adverse feedback and generates a set of corrected treatment signal input parameters. For example, if excessive feedback is detected, the process will determine an energy attenuation coefficient based on the degree of excess feedback by consulting or calculating it, and use this coefficient to reduce the subsequent treatment energy to be output. This is to correct treatment deviations in real time, rather than relying solely on long-term parameter relationship updates, thereby maximizing patient safety and treatment effectiveness in a single treatment session.

[0028] Example 2: Please refer to Figure 2 As shown, this embodiment provides a control system for an intense pulsed light therapy device based on a neural network model. By dynamically evaluating the state of the patient's skin area to be treated, a personalized treatment plan is generated. Then, based on the feedback results after treatment, the built-in neural network model is continuously optimized, thereby realizing intelligent and adaptive closed-loop control of the intense pulsed light therapy process, thereby improving the safety and effectiveness of the treatment.

[0029] The system can be physically integrated into the main control unit of the intense pulsed light therapy device, or it can be connected to the therapy device as an external computer device via a data interface.

[0030] This system includes the following modules: The status assessment module generates status assessment results for the patient's skin area to be treated, which serves as the basis for subsequent treatment decisions. In a complete operation process, it collects real-time status information of the skin area to be treated, such as skin color, melanin content, skin surface temperature, and humidity, through sensors integrated with the treatment device, such as high-resolution cameras, thermal imagers, and skin conductivity sensors.

[0031] Then, based on the collected real-time status information, the current version of the neural network model is called to generate a detection stimulus signal as a non-therapeutic signal. Specifically, the module extracts real-time status values ​​from the real-time status information, such as the melanin index, and then matches them with preset skin status judgment standards. When a match is successful, if the real-time status value is less than a preset classification measurement value, or if it is greater than the classification measurement value but the values ​​of other relevant status parameters do not exceed the comparison threshold, the skin is confirmed to be in a testable state, and the detection stimulus signal is generated. The signal energy is extremely low and insufficient to produce a therapeutic effect; its purpose is only to detect the skin's stress response.

[0032] After the detection stimulus signal is applied to the skin area to be treated, the patient's response is continuously monitored, such as subtle changes in skin pigmentation, rate of temperature rise, or sensations reported by the patient through a feedback device. This information is then quantified into response data for analysis, especially the duration of the response. If the response duration exceeds the preset assessment criteria, it indicates that the skin may be in an allergic or intolerance state, and the status assessment result will be generated as an abnormal type. When the status assessment result is an abnormal type, the output of subsequent treatment signals will be blocked to ensure patient safety. If the response is within the normal range, a status assessment result marked as normal will be generated based on the response data and passed to subsequent modules.

[0033] The treatment parameter generation module is activated after receiving the normal-type state assessment result generated by the state assessment module. Its function is to generate treatment signal input parameters to guide this treatment. By acquiring the patient's baseline response state before the treatment signal is applied, and using this as a starting point, the module sequentially sets multiple assessment time periods on the time axis.

[0034] During the output of treatment signals by the treatment execution module, response data from the status assessment module is received in real time. In each assessment period, the real-time response data is compared with a preset comparison parameter. If the response data does not reach the comparison parameter, a normal output node is determined, that is, the treatment signal is allowed to continue to be output; if the response data reaches or exceeds the comparison parameter, a stop output node is determined, indicating that the energy output should be stopped or reduced immediately.

[0035] By dynamically determining a series of normal output nodes and stop output nodes throughout the entire treatment pulse, the module ultimately generates treatment signal input parameters that include specific normal output periods. These treatment signal input parameters are a fine-grained, time-varying output control scheme, avoiding the use of a single energy value that fails to fully reflect user behavior characteristics. For example, the width, interval, and number of sub-pulses are defined to achieve real-time feedback control of the treatment process.

[0036] The treatment execution module receives the treatment signal input parameters generated by the treatment parameter generation module and outputs the treatment signal to the skin area to be treated based on the parameters. Specifically, it parses the treatment signal input parameters into specific control instructions for the hardware of the intense pulsed light therapy device.

[0037] During execution, the module controls energy release by following the normal output period defined in the treatment signal input parameters. At the same time, the module has a built-in safety interlock mechanism that will immediately prevent the output of any treatment signal when it receives an abnormal type status assessment result from the status assessment module.

[0038] After treatment, this module is responsible for collecting treatment feedback results, which is a comprehensive dataset that includes the actual output treatment signal intensity, pulse waveform, and the final patient response status, such as the final skin temperature, pigmentation changes, and delayed erythema reaction after treatment.

[0039] The model update module receives treatment feedback results collected by the treatment execution module and updates the neural network model based on these results. It analyzes the treatment signal intensity and patient response status contained in the feedback results and determines the current treatment status as effective feedback, excessive feedback, or insufficient feedback. For example, if the expected treatment endpoint is reached within a safe energy range, such as hair follicle atrophy, it is determined to be effective feedback; if the skin reaction exceeds expectations, such as excessive redness and swelling, it is excessive feedback; if the expected effect is not achieved, it is insufficient feedback.

[0040] When the treatment status is determined to be excessive or insufficient feedback, a self-optimizing parameter correction processing logic is executed. By adjusting the internal weights and parameters of the neural network model, it can more accurately correlate skin status and treatment response in future predictions. Then, based on the adjusted treatment parameter generation module, the generation strategy of subsequent treatment signals is optimized by using comparison parameters or decision rules. In the next treatment, the initial energy suggestion is reduced, or the trigger sensitivity of the stop output node is increased.

[0041] The updated neural network model will be saved and used by the state evaluation module in subsequent control methods, such as the treatment of the next skin area or the next treatment visit, to generate more accurate detection stimulus signals, thus forming a complete and continuously evolving closed-loop control system.

[0042] Through the coordinated operation of the above modules, this embodiment can achieve refined, personalized, and adaptive control of the intense pulsed light therapy process, changing the traditional mode that relies on operator experience and static preset parameters. Through data-driven real-time feedback and model iteration, the safety and predictability of the treatment are improved.

Claims

1. A control method of an intense pulsed light treatment device based on a neural network model, characterized by, In response to the state evaluation result generated according to the skin area to be treated of the patient, the following is performed: a treatment signal input parameter is generated based on the state evaluation result, a treatment signal is output to the skin area to be treated according to the treatment signal input parameter, and a treatment feedback result is collected after the treatment is completed, and the neural network model is updated based on the treatment feedback result, wherein the updated neural network model is used to generate a detection stimulation signal in subsequent control method execution; wherein, in response to the state evaluation result generated according to the skin area to be treated of the patient, the following is performed: real-time state information of the skin area to be treated is collected, a detection stimulation signal is generated using a neural network model based on the real-time state information, and the detection stimulation signal is applied to the skin area to be treated, during the application of the detection stimulation signal, the patient's reaction is monitored to form reaction data, and the state evaluation result is generated based on the reaction data.

2. The control method of the intense pulsed light treatment instrument based on a neural network model according to claim 1, wherein, Generating a detection stimulation signal includes: matching the real-time state information with a preset skin state judgment standard, and generating a detection stimulation signal as a non-treatment signal when the matching is successful; wherein, the matching success includes: the real-time state value extracted from the real-time state information is less than the classification measurement value, or the real-time state value is greater than the classification measurement value but the value of the state parameter does not exceed the comparison threshold. 3.The control method of the IPL treatment instrument based on the neural network model according to claim 1, wherein, In response to the state evaluation result generated according to the skin area to be treated of the patient, the following is also performed: After applying the detection stimulation signal, if the response duration exceeds the evaluation standard, the state evaluation result is generated as an abnormal type; when the state evaluation result is of the abnormal type, the output of the treatment signal is prevented. 4.The control method of the IPL treatment instrument based on the neural network model according to claim 1, wherein, Generating a treatment signal input parameter based on a state evaluation result includes: when the state evaluation result is of a normal type, obtaining a baseline reaction state of the patient, and sequentially setting a plurality of evaluation time periods from a time point corresponding to the baseline reaction state, comparing the reaction data with a comparison parameter in each evaluation time period to determine a normal output node or a stop output node, and generating a treatment signal input parameter containing a normal output period based on the normal output node and the stop output node. 5.The control method of the IPL treatment instrument based on the neural network model according to claim 1, wherein, Updating the neural network model based on the state evaluation result includes: based on the treatment signal intensity and the patient's response state, the treatment state is determined as effective feedback, excessive feedback or insufficient feedback, and when the treatment state is determined as excessive feedback or insufficient feedback, the treatment signal input parameter used to generate the subsequent treatment signal is adjusted.

6. A control system for an intense pulsed light therapy device based on a neural network model, characterized by It includes: a state evaluation module for generating a state evaluation result about the skin area to be treated of the patient; a treatment parameter generation module configured to generate a treatment signal input parameter in response to the state evaluation result generated by the state evaluation module; a treatment execution module for outputting a treatment signal according to the treatment signal input parameter and collecting a treatment feedback result; and a model update module for updating the neural network model based on the treatment feedback result, wherein the updated neural network model is used by the state evaluation module to generate a detection stimulation signal in subsequent control method execution.

7. The control system of the intense pulsed light therapeutic instrument based on a neural network model according to claim 6, wherein, The state evaluation module is configured to collect real-time state information of the skin area to be treated, generate and apply a detection stimulation signal using a neural network model based on the real-time state information, and monitor the patient's reaction to generate a state evaluation result.

8. The control system of the intense pulsed light therapeutic instrument based on a neural network model according to claim 7, characterized in that, If the response duration exceeds the evaluation criterion after the detection stimulus signal is applied, the state evaluation result is generated as an abnormal type; and when the state evaluation result is the abnormal type, the output of the treatment signal is prevented.

9. The control system of the intense pulsed light therapeutic instrument based on a neural network model according to claim 8, wherein, When the state evaluation result is the normal type, based on the analysis of the reaction data, a normal output node and a stop output node are determined to generate a treatment signal input parameter containing a normal output period.

10. The control system of the intense pulsed light therapeutic instrument based on a neural network model according to claim 6, wherein, The model updating module is configured to determine a treatment state based on the treatment signal intensity contained in the treatment feedback result and the patient response state, and adjust the treatment signal input parameter for generating a subsequent treatment signal when the treatment state is excessive feedback or insufficient feedback.

Images