Radio frequency treatment comfort optimization method based on multi-modal data driving

Abstract

The invention discloses a radio frequency treatment comfort optimization method based on multi-modal data driving, and aims at solving the problems that in the prior art, subjective dependence of comfort evaluation is high, and systematic data support and optimization strategies are lacked. A multi-mode sensor is configured through a cloud server, treatment scheme data including thermal imaging, power output and pressure data are collected, and patient condition and environment data obtained from an authorization evaluation platform are combined. A preset comfort level analysis rule set and a multi-dimensional decision table are utilized to perform objective comfort level division and scoring on the treatment process, and based on the division result and the environment data, an ablation system technology is adopted to simulate and calculate the expected influence of different environment parameter combinations so as to determine the comfort level of the treatment process. And generating personalized environmental parameter optimization suggestions under the condition of meeting security constraints. According to the invention, objective and quantitative evaluation and intelligent environment regulation and control of the radio frequency treatment comfort are realized, and the standardization level of treatment and the patient experience are improved.

Description

Technical Field

[0001] This invention belongs to the field of data analysis technology and discloses a method for optimizing the comfort of radiofrequency therapy based on multimodal data. Background Technology

[0002] Radiofrequency ablation, as a widely used non-invasive treatment, is highly dependent on the patient's experience for its efficacy. Currently, optimizing comfort during treatment relies primarily on the individual experience of the operating physician and real-time subjective feedback from the patient, lacking an objective and quantitative assessment and control system. Treatment parameters and environmental factors are often set based on general protocols, making it difficult to dynamically adapt to individualized disease characteristics and real-time treatment responses. This results in significant fluctuations in patient comfort, impacting treatment adherence and overall satisfaction.

[0003] Existing technologies include methods that attempt to assess treatment response by monitoring single or a few physiological signals. However, these methods often view the treatment process in isolation, failing to systematically integrate and analyze the output parameters of the treatment equipment, individual patient characteristics, and treatment environment data. Their analysis typically relies on pre-set, simple threshold judgments, lacking a deep understanding of the multi-dimensional and temporal interactions of data, and thus unable to form a refined classification of comfort levels. Furthermore, existing technologies lack systematic and interpretable decision support methods for using assessment results to optimize treatment environment parameters, making it difficult to provide accurate and safe optimization suggestions.

[0004] Therefore, the industry urgently needs a method that can systematically integrate multimodal treatment process data, patient condition background, and environmental information, objectively classify comfort levels based on this data, and then intelligently generate environmental parameter optimization strategies. This method should avoid over-reliance on subjective experience and, through a data-driven technological approach, achieve standardization, personalization, and intelligent management of radiofrequency ablation comfort, thereby significantly improving the patient's treatment experience while ensuring therapeutic efficacy. Summary of the Invention

[0005] To achieve the above objectives, this application provides the following technical solution:

[0006] A multimodal data-driven method for optimizing the comfort of radiofrequency ablation includes the following steps:

[0007] S1, a multimodal sensor is configured using a preset cloud server, and radiofrequency treatment plan data is collected based on the multimodal sensor;

[0008] S2, crawl the patient's condition data and environmental data of radiofrequency treatment for each radiofrequency treatment plan from the preset radiofrequency treatment evaluation platform;

[0009] S3. Based on the radiofrequency treatment plan data and the patient's condition data for each radiofrequency treatment plan, the comfort level of radiofrequency treatment is divided to obtain the comfort level data of radiofrequency treatment for each plan.

[0010] S4. Based on the comfort data and environmental data of radiofrequency treatment for each scheme, the parameters of the radiofrequency treatment environment for that scheme are optimized using ablation system technology.

[0011] Furthermore, S1 specifically includes:

[0012] A sensor virtualization management module is deployed on the preset cloud server, and virtual sensor instances corresponding to physical multimodal sensors are created through the sensor virtualization management module;

[0013] The virtual sensor instance receives the raw sensor stream uploaded in real time by the physical multimodal sensor;

[0014] A radiofrequency treatment plan data parsing unit is set up on the preset cloud server. The parsing unit performs time-series alignment and encapsulation on the original sensor stream to form a structured treatment plan data packet.

[0015] Further, S2 includes:

[0016] Establish an authorized data interface connection with the preset radiofrequency therapy evaluation platform, and construct a batch query request based on a unique treatment plan identifier;

[0017] An asynchronous data call is initiated to the preset radiofrequency therapy evaluation platform through the authorized data interface. The target data set for the call includes a first type of data and a second type of data.

[0018] Design a data crawling scheduling strategy. This strategy initiates data calls in a non-blocking manner based on the sequence of treatment plan identifiers, and performs integrity verification and redundancy cleaning on the returned first type of data and second type of data.

[0019] The first and second types of data, after passing the verification, are stored in different sets of the non-relational database of the preset cloud server according to their correspondence with the treatment plan identifier, and a data source tag and a time validity tag are attached to each data item.

[0020] Further, S3 includes:

[0021] A comfort analysis engine is built on the preset cloud server. The comfort analysis engine first performs a data association operation, and associates and matches the structured treatment plan data package with the corresponding first type of data, namely the patient's condition data, according to the unique treatment plan identifier.

[0022] The comfort analysis engine has multiple preset comfort analysis rule sets, each rule set being configured for different types of the disease data;

[0023] For disease data containing specific skin type codes or target tissue depth prediction levels, the first comfort analysis rule set is invoked to extract the average temperature rise slope of the thermal imaging frame sequence throughout the treatment process and the cumulative time percentage of the local pressure sensing sequence exceeding a preset threshold from the structured treatment plan data package. The average temperature rise slope and the cumulative time percentage are input into a preset first multidimensional decision table, and the first comfort intermediate score is output.

[0024] For medical data containing specific chief complaints or past related treatment history, the second comfort analysis rule set is invoked. This rule set defines: from the structured treatment plan data package, the fluctuation period and amplitude characteristics of the power output waveform data stream are analyzed, the deviation integral between the power output waveform data stream and the standard stable waveform is calculated, the number of abnormal temperature diffusion frames in the treatment edge area of ​​the thermal imaging video stream is identified, the deviation integral and the number of abnormal diffusion frames are input into a preset second multidimensional decision table, and the second comfort intermediate score is output.

[0025] The comfort analysis engine establishes a scoring fusion module, which receives the first comfort intermediate score and / or the second comfort intermediate score, and performs weighted calculation and normalization on the input intermediate scores according to the weight configuration of each feature in the disease data to generate a comprehensive comfort score value for the treatment plan.

[0026] The comfort analysis engine maps the comprehensive comfort score to discrete comfort level identifiers based on a preset comfort level threshold range.

[0027] The comprehensive comfort score, the comfort level identifier, the intermediate score and key data features on which it is based are collectively encapsulated into the comfort data of the radiofrequency therapy of the treatment plan, and stored in the non-relational database.

[0028] Furthermore, before performing the data association operation, the comfort analysis engine also performs a data valid range truncation operation, specifically:

[0029] Based on the treatment area code and target tissue depth prediction level in the aforementioned disease data, the range of estimated effective treatment time is calculated.

[0030] Based on the estimated effective treatment duration range, the time-series data in the structured treatment plan data package is extracted, and thermal imaging frames, power output data, and pressure readings whose time span falls within the estimated effective treatment duration range are retained for subsequent comfort level classification analysis.

[0031] Furthermore, the construction process of the multidimensional decision table includes:

[0032] Historical treatment case data is collected in advance, including treatment plan data, disease data, and comfort evaluation tags actively provided by patients after treatment;

[0033] Based on the historical treatment case data, for different treatment scenarios and patient groups, key parameter combinations that affect comfort and their corresponding threshold boundaries are defined.

[0034] Based on the combination of key parameters and threshold boundaries, a mapping relationship between parameter input values ​​and intermediate comfort scores is established, forming the first multidimensional decision table and the second multidimensional decision table, which are then stored in the rule base of the comfort analysis engine.

[0035] Further, S4 includes:

[0036] A parameter optimization service module is deployed on the preset cloud server. The parameter optimization service module reads the comfort data of the radiofrequency therapy for a certain treatment plan and the corresponding second type of data, namely the environmental data of the radiofrequency therapy.

[0037] The parameter optimization service module uses the comfort level identifier as a status indicator of the optimization target, and uses the ambient temperature, ambient humidity, equipment body temperature, treatment bed tilt angle, and coupling agent type identifier in the environmental data as environmental parameter variables to be optimized.

[0038] The ablation system technology is specifically implemented as follows: the parameter optimization service module maintains an environmental parameter configuration space, which consists of all possible discrete or continuous value ranges of the environmental parameter variable to be optimized.

[0039] For the current treatment plan, the parameter optimization service module generates a series of parameter variation combinations in the environmental parameter configuration space, based on the currently recorded environmental data values. Each parameter variation combination is a new set of parameters obtained by making a small perturbation to the environmental parameter variables.

[0040] For each generated parameter variation combination, the expected direction and magnitude of its impact on the overall comfort score are simulated and calculated. This simulation is based on a pre-stored parameter-comfort impact knowledge graph, which reflects the correlation patterns between changes in different environmental parameters and changes in comfort scores in historical data. One or more parameter variation combinations that maximize the expected increase in the overall comfort score are selected as recommended environmental parameter optimization schemes.

[0041] Furthermore, the workflow of the parameter optimization service module includes:

[0042] The process of ranking environmental parameters by importance is as follows: analyze the key data features on which the comfort data is based, identify the features that contribute highly to the comfort classification, and then backtrack the environmental parameter variables that are most strongly associated with these features.

[0043] Based on the importance ranking of the environmental parameters, the module dynamically adjusts the generation strategy of parameter mutation combinations in the environmental parameter configuration space. For environmental parameter variables with high importance, a finer-grained perturbation step size is set near their current value to generate more mutation possibilities. For parameter variables with low importance, a coarser perturbation step size is used or their current value is kept unchanged.

[0044] The construction and updating of the parameter-comfort influence knowledge graph are independent of the parameter optimization service module, and its construction process includes:

[0045] Complete treatment records that have been linked with comfort data and environmental data are periodically extracted from the aforementioned non-relational database;

[0046] Statistical analysis was performed on the extracted records to quantify the conditional probability of each environmental parameter variable co-occurring with different comfort levels or rating intervals in different value ranges.

[0047] The conditional probabilities and the interaction information between parameters are stored in a graph structure, where nodes represent parameters or comfort states, and edges represent influence relationships and intensity weights.

[0048] When simulating the expected impact, the parameter optimization service module inputs the current parameter variation combination into the knowledge graph, reasons along the edges of the graph, finds the historical parameter pattern most similar to the current variation combination, aggregates the comfort score results corresponding to the similar patterns, calculates the expected comfort score value of the current variation combination by weighted average, and then compares it with the currently recorded comprehensive comfort score value to obtain the expected improvement.

[0049] When outputting recommended environmental parameter optimization schemes, the module also adds a confidence assessment for each recommended scheme. The confidence assessment is calculated based on the number of historical similar patterns used for reasoning in the knowledge graph and the similarity weighting value.

[0050] The parameter optimization service module encapsulates the recommended environmental parameter optimization scheme, the corresponding expected improvement in comfort, and the confidence level assessment into an optimization suggestion report, which is then provided to the treatment equipment or operator terminal through the interface of the preset cloud server.

[0051] Furthermore, the parameter optimization service module introduces a constraint check mechanism when generating parameter variation combinations, specifically:

[0052] Physical or safety constraints between environmental parameters are predefined. After each parameter variation combination is generated, its legality is verified according to the constraint condition checking mechanism. Only parameter variation combinations that pass the verification and meet the requirements of safety and physical feasibility are retained for subsequent simulation calculation and selection processes.

[0053] Furthermore, the preset radiofrequency therapy evaluation platform is a certified third-party medical efficacy evaluation community platform that allows the sharing of anonymized treatment data. All crawling operations are within the scope of the platform's user agreement and data privacy policy, and data is acquired through its application programming interface.

[0054] This invention discloses a multimodal data-driven method for optimizing the comfort of radiofrequency ablation (RFA) therapy, aiming to address the problems of strong subjective dependence and lack of systematic data support and optimization strategies in existing comfort assessments. It utilizes a cloud server configured with multimodal sensors to collect treatment protocol data, including thermal imaging, power output, and pressure data, combined with patient condition and environmental data obtained from an authorized evaluation platform. Using a pre-set set of comfort analysis rules and a multidimensional decision table, the treatment process is objectively categorized and scored for comfort. Furthermore, based on the categorization results and environmental data, ablation system technology is employed to simulate and calculate the expected impact of different combinations of environmental parameters, generating personalized environmental parameter optimization suggestions while meeting safety constraints. This invention achieves objective and quantitative assessment and intelligent environmental control of RFA comfort, contributing to improved treatment standardization and patient experience. Attached Figure Description

[0055] Figure 1 This is a flowchart illustrating a multimodal data-driven method for optimizing the comfort of radiofrequency therapy, as claimed in an embodiment of the present invention.

[0056] Figure 2 This is a second workflow diagram of a radiofrequency therapy comfort optimization method based on multimodal data-driven approach claimed in an embodiment of the present invention;

[0057] Figure 3 The third flowchart is a method for optimizing the comfort of radiofrequency therapy based on multimodal data, as claimed in an embodiment of the present invention.

[0058] Figure 4 The fourth flowchart is a method for optimizing the comfort of radiofrequency therapy based on multimodal data, as claimed in an embodiment of the present invention. Detailed Implementation

[0059] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0060] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of those features. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications in the embodiments of this application, such as up, down, left, right, front, back, etc., are only used to explain the relative positional relationships and movements between components in a specific orientation as shown in the accompanying drawings. If the specific orientation changes, the directional indications will change accordingly. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0061] References to embodiments herein mean that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0062] According to the first embodiment of the present invention, referring to Figure 1 A method for optimizing the comfort of radiofrequency therapy based on multimodal data includes the following steps:

[0063] S1, a multimodal sensor is configured using a preset cloud server, and radiofrequency treatment plan data is collected based on the multimodal sensor;

[0064] S2, crawl the patient's condition data and environmental data of radiofrequency treatment for each radiofrequency treatment plan from the preset radiofrequency treatment evaluation platform;

[0065] S3. Based on the radiofrequency treatment plan data and the patient's condition data for each radiofrequency treatment plan, the comfort level of radiofrequency treatment is divided to obtain the comfort level data of radiofrequency treatment for each plan.

[0066] S4. Based on the comfort data and environmental data of radiofrequency treatment for each scheme, the parameters of the radiofrequency treatment environment for that scheme are optimized using ablation system technology.

[0067] In this embodiment, the implementing entity is a pre-set cloud server system, and the implementation process begins with the collection of multimodal data on the treatment process.

[0068] First, the cloud server configures and manages a multimodal sensor that integrates a thermal imaging module, a power monitoring module, and a pressure sensing module. This sensor is used to simultaneously acquire thermal imaging video streams of the temperature distribution in the treatment area, real-time power signal waveform data streams from the radiofrequency treatment device, and local pressure sensing sequences showing the pressure distribution at the contact surface between the treatment head and the patient's skin during radiofrequency therapy. These raw data streams are transmitted to the cloud server in real time.

[0069] Secondly, the cloud server accesses a pre-set, independent radiofrequency therapy evaluation platform through a secure network connection. This platform stores a large amount of anonymized treatment case data. The server crawls two key supplementary data corresponding to this platform based on specific treatment plan identifiers: one is the patient's pre-treatment condition data, such as age and symptom description, and the other is the environmental data recorded at the time of treatment, such as room temperature and equipment temperature.

[0070] Next, the cloud server correlates and comprehensively analyzes the radiofrequency treatment plan data it collects with the patient data crawled from the platform; through the built-in comfort analysis logic, it evaluates and classifies the comfort of each treatment plan, generating a set of structured comfort data for each plan, which includes quantitative scores and classification labels.

[0071] Finally, for each evaluated treatment plan, the cloud server combines its comfort data with environmental data crawled from the platform, employing a parameter exploration strategy called ablation system technology. This technology systematically simulates and evaluates the potential impact of different combinations of environmental parameters on comfort, thereby identifying environmental parameter optimization suggestions that can improve the comfort of the treatment plan, such as adjusting the temperature and humidity of the treatment room or the angle of the treatment bed.

[0072] Experimental Test Overview: To verify the effectiveness of this method, a simulation experiment can be designed. In a laboratory environment, a standard radiofrequency therapy device is set up, connected to the multimodal sensor and cloud server system described in this embodiment. Simultaneously, a radiofrequency therapy evaluation platform containing historical cases is simulated, selecting several typical simulated treatment cases covering different skin types and treatment sites to execute the entire process of this method. The continuity and feasibility of the process are verified by checking whether the system can successfully complete the complete sequence from data acquisition, platform information crawling, comfort data generation to environmental parameter optimization suggestions output. The test focuses on whether the data interfaces between each step can be correctly connected and whether the logical sequence is executed, rather than the specific optimization effect.

[0073] Furthermore, S1 specifically includes:

[0074] A sensor virtualization management module is deployed on the preset cloud server, and virtual sensor instances corresponding to physical multimodal sensors are created through the sensor virtualization management module;

[0075] The virtual sensor instance receives the raw sensor stream uploaded in real time by the physical multimodal sensor;

[0076] A radiofrequency treatment plan data parsing unit is set up on the preset cloud server. The parsing unit performs time-series alignment and encapsulation on the original sensor stream to form a structured treatment plan data packet.

[0077] In this embodiment, a sensor virtualization management module runs on the cloud server. This module does not directly operate the physical hardware, but creates a completely corresponding virtual sensor instance for each physical multimodal sensor at the software level. This instance is like a digital twin of the physical sensor in the cloud, responsible for receiving and managing all data streams from the physical sensor.

[0078] The physical multimodal sensor operates continuously during treatment, generating three time-synchronized data streams:

[0079] Thermal imaging video stream: Captures infrared images of the treatment area at a fixed frame rate to form a video sequence;

[0080] Power output waveform data stream: Through the coupling circuit, the current and voltage at the output end of the radiofrequency therapy device are read in real time, the instantaneous power value is calculated and output, forming a power waveform sequence that changes over time.

[0081] Local pressure sensing sequence: The pressure sensor array on the treatment head samples at a fixed frequency and outputs the average pressure value of the treatment head in contact with the skin at each sampling point, forming a sequence of pressure changes over time.

[0082] These raw data streams are then tagged with high-precision timestamps and sent in real time over the network to the corresponding virtual sensor instances on the cloud server.

[0083] The cloud server also includes a dedicated radiofrequency therapy protocol data parsing unit. The core task of this unit is to integrate multi-source asynchronous data. It receives three timestamped raw data streams from virtual sensor instances, first performing rigorous time-series alignment to ensure precise matching of thermal imaging frames, power readings, and pressure values ​​acquired at the same time. After alignment, the parsing unit encapsulates this data into a structured treatment protocol data packet in chronological order. This data packet uses a timestamp as the primary index. For example, for a millisecond-level timestamp t, the data packet associates and stores: the thermal imaging image at time t, the instantaneous power value at time t, and the pressure sensor reading at time t. Furthermore, each data packet is assigned a globally unique treatment protocol identifier upon creation. This identifier is used throughout the entire process to associate all relevant data.

[0084] When testing this embodiment, the accuracy of data acquisition and encapsulation needs to be verified. A signal generator and a standard pressure source are used to simulate power output and pressure signals, while a standard blackbody radiation source is used to calibrate the thermal imager. The system is run for acquisition over a specified period. After completion, the generated structured data packets are checked: 1) Different time points are extracted, and the physical time alignment of the three types of data corresponding to that time point within the data packet can be verified using an external synchronization signal; 2) The integrity of the data within the data packet is checked, and there are no dropped frames or missing data; 3) The uniqueness and compliance of the treatment plan identifier for each data packet with the encoding rules are verified. A successful test is indicated by accurate timing alignment within the data packet, complete data, and a unique and valid identifier.

[0085] Furthermore, referring to Figure 2 S2 includes:

[0086] Establish an authorized data interface connection with the preset radiofrequency therapy evaluation platform, and construct a batch query request based on a unique treatment plan identifier;

[0087] An asynchronous data call is initiated to the preset radiofrequency therapy evaluation platform through the authorized data interface. The target data set for the call includes a first type of data and a second type of data.

[0088] Design a data crawling scheduling strategy. This strategy initiates data calls in a non-blocking manner based on the sequence of treatment plan identifiers, and performs integrity verification and redundancy cleaning on the returned first type of data and second type of data.

[0089] The first and second types of data, after passing the verification, are stored in different sets of the non-relational database of the preset cloud server according to their correspondence with the treatment plan identifier, and a data source tag and a time validity tag are attached to each data item.

[0090] In this embodiment, an authorized data interface based on security protocols such as OAuth is established between the cloud server and the radiofrequency therapy evaluation platform. The server first obtains a list of treatment plan identifiers to be queried from the local database. For each identifier in the batch, the system constructs a standardized query request message, which includes the identifier and the requested data range, namely, disease data and environmental data.

[0091] The system initiates data request requests to the evaluation platform asynchronously and non-blockingly through the authorized data interface. This means the server can send the next request without waiting for a response from the previous one, improving efficiency. The target data to be requested falls into two clearly defined categories:

[0092] The first type of data is patient condition data: This is not a complete medical record, but rather a summary of information extracted for comfort analysis. It consists of two parts: First, an electronic medical record summary, which includes at least the patient's age group (e.g., young adult, middle-aged), gender, the chief complaint of this treatment (e.g., skin laxity, acne scars), and relevant past treatment history (e.g., whether they have received similar radiofrequency treatments); second, a characteristic description of the treatment area, which includes at least internationally recognized skin type classification codes (e.g., Fitzpatrick classification), standardized codes for the treatment area area (e.g., small area, medium area), and the target tissue (e.g., dermis) and subcutaneous fat depth prediction level (e.g., superficial, intermediate, deep) derived from imaging or ultrasound reports.

[0093] The second category of data is environmental data from radiofrequency ablation: This is a log of objectively recorded environmental parameters during the treatment. It includes at least: real-time ambient temperature and relative humidity values ​​recorded by a digital thermometer and hygrometer in the treatment room; steady-state temperature of the treatment device recorded by sensors on its casing, reflecting the device's heat dissipation status; bed tilt angle settings recorded by a treatment bed angle sensor; and the type of coupling agent used in this treatment, such as aqueous gel, oil-based medium, and the estimated dosage level based on usage, such as small amount, standard amount, or large amount, recorded and uploaded by the therapist.

[0094] To ensure efficient and stable data acquisition, the system employs a data crawling scheduling strategy. This strategy, based on an identifier list, controls the frequency of request sending and the retry mechanism to avoid excessive load on the platform server. Upon receiving data from the platform, the system immediately performs integrity checks, verifying that all required fields are present in the returned JSON or XML data packet and cleaning up any obviously duplicated or out-of-range data.

[0095] Finally, all verified patient and environmental data are stored in two different collections of a non-relational database such as MongoDB on a cloud server, according to their respective treatment plan identifiers. Each stored data entry is accompanied by metadata, including a data source tag indicating which evaluation platform it comes from, a time validity tag indicating whether the data was recorded before or during treatment, and the data's validity period.

[0096] Experimental Test Overview: A simulated radiofrequency therapy evaluation platform test server was built, pre-stored with hundreds of records containing fictitious but medically logical patient and environmental data. Each record is associated with a unique ID. An interface for connecting to this test platform was configured on the cloud server of this method. Then, a list of known IDs was input into the system to start the crawling process. Test Points: 1) Interface Connectivity: Whether an authorized connection can be successfully established; 2) Data Accuracy: Whether the crawled data content is completely consistent with the original data stored on the test server, without tampering or loss; 3) Scheduling Strategy Effectiveness: Observe whether the system smoothly sends requests according to the preset strategy under a large number of ID requests, whether the retry mechanism is triggered, and whether it can correctly handle the platform's simulated short-term failures; 4) Storage Standardization: Check whether the crawled data is correctly stored in the corresponding set of the non-relational database, and whether the metadata is completely marked.

[0097] Further, S3 includes:

[0098] A comfort analysis engine is built on the preset cloud server. The comfort analysis engine first performs a data association operation, and associates and matches the structured treatment plan data package with the corresponding first type of data, namely the patient's condition data, according to the unique treatment plan identifier.

[0099] The comfort analysis engine has multiple preset comfort analysis rule sets, each rule set being configured for different types of the disease data;

[0100] For disease data containing specific skin type codes or target tissue depth prediction levels, the first comfort analysis rule set is invoked to extract the average temperature rise slope of the thermal imaging frame sequence throughout the treatment process and the cumulative time percentage of the local pressure sensing sequence exceeding a preset threshold from the structured treatment plan data package. The average temperature rise slope and the cumulative time percentage are input into a preset first multidimensional decision table, and the first comfort intermediate score is output.

[0101] For medical data containing specific chief complaints or past related treatment history, the second comfort analysis rule set is invoked. This rule set defines: from the structured treatment plan data package, the fluctuation period and amplitude characteristics of the power output waveform data stream are analyzed, the deviation integral between the power output waveform data stream and the standard stable waveform is calculated, the number of abnormal temperature diffusion frames in the treatment edge area of ​​the thermal imaging video stream is identified, the deviation integral and the number of abnormal diffusion frames are input into a preset second multidimensional decision table, and the second comfort intermediate score is output.

[0102] The comfort analysis engine establishes a scoring fusion module, which receives the first comfort intermediate score and / or the second comfort intermediate score, and performs weighted calculation and normalization on the input intermediate scores according to the weight configuration of each feature in the disease data to generate a comprehensive comfort score value for the treatment plan.

[0103] The comfort analysis engine maps the comprehensive comfort score to discrete comfort level identifiers based on a preset comfort level threshold range.

[0104] The comprehensive comfort score, the comfort level identifier, the intermediate score and key data features on which it is based are collectively encapsulated into the comfort data of the radiofrequency therapy of the treatment plan, and stored in the non-relational database.

[0105] In this embodiment, a core component—a comfort analysis engine—runs on the cloud server. After the engine starts, it first performs a data association operation. It reads the structured treatment plan data package and, based on the treatment plan identifier it carries, searches for and associates the corresponding first-type data condition data in the non-relational database. At this point, all the key input data for a treatment is ready.

[0106] The comfort analysis engine has multiple pre-built sets of comfort analysis rules, which are analysis logic paths designed for different patient condition characteristics.

[0107] The engine will analyze the patient's condition data. For example, if the condition data specifically emphasizes that the skin type code is heat-sensitive, such as Type I, or the target tissue depth prediction level is shallow, the engine will prioritize calling the first comfort analysis rule set. If the patient's chief complaint is a pain-sensitive condition, or if the patient's past treatment history indicates that the patient has had an adverse reaction, the engine may call the second comfort analysis rule set. Both situations may coexist and will be processed in parallel.

[0108] The execution process of the first comfort analysis rule set: This rule set primarily focuses on thermal stimulation and physical pressure. It extracts the thermal imaging frame sequence of the entire treatment process from the associated structured treatment protocol data package, calculates the average temperature rise curve of the core treatment area in all frames, and further calculates the average slope of this curve as an indicator of heat accumulation rate. Simultaneously, it extracts the local pressure sensing sequence, counts the time points when pressure readings exceed a preset discomfort threshold, and calculates the percentage of this time relative to the total treatment time, as an indicator of high-pressure duration percentage. Subsequently, the engine uses the average temperature rise slope and high-pressure duration percentage of these two indicators as input to query a preset first multidimensional decision table. This decision table defines the first comfort intermediate score corresponding to different ranges of slope and percentage combinations, for example, a score from 0 to 100. A high score indicates better comfort under this rule set.

[0109] The execution process of the second comfort analysis rule set: This rule set focuses more on the stability and safety of treatment. It analyzes the power output waveform data stream from the data packet, calculates the periodicity and stability of the waveform, and quantifies its deviation from the ideal stable square wave or sine wave to obtain a power fluctuation deviation integral value. The larger the value, the more unstable the power output. At the same time, it analyzes the thermal imaging video stream, and through image processing, identifies whether there are abnormal or unexpected temperature diffusion areas at the edge of the treatment area, which may represent uneven heat conduction or energy leakage. It also counts the number of frames with such abnormal diffusion. Similarly, the power fluctuation deviation integral and the number of abnormal diffusion frames are input into a preset second multidimensional decision table to obtain the second comfort intermediate score.

[0110] The scoring fusion module is responsible for integrating intermediate scores from different rule sets. It calculates the weights of the received intermediate scores based on the preset weights of each feature in the current condition data. For example, the first rule set has a higher weight for patients with sensitive skin, while the second rule set may have a higher weight for patients with a history of pain. Then, it performs normalization processing to convert all weighted scores into a unified comprehensive comfort score, which is ultimately a score of 0-100.

[0111] The engine predefines comfort level thresholds; for example, a composite score of ≥80 corresponds to high comfort, 60-79 to medium comfort, and <60 to low comfort. Based on the calculated composite score, the engine determines its corresponding comfort level identifier.

[0112] Ultimately, the engine encapsulates all the outputs of this analysis—the overall comfort score, comfort level identifier, intermediate score values ​​calculated from each rule set, and key data features extracted and used in the analysis, such as specific temperature rise slope values ​​and high pressure percentages—into a complete radiofrequency therapy comfort data object and stores it back into a non-relational database, linking it to the original data through the treatment plan identifier.

[0113] During experimental testing, a standard case library containing known treatment processes and patient feedback is prepared. Patient feedback from these cases is used only for final validation and does not participate in the analysis process. The system of this embodiment is used to process this set of cases. The test is divided into several levels: 1) Rule set triggering correctness: Input different disease data and check whether the engine correctly calls the expected rule set, such as triggering the first rule set for skin sensitivity cases; 2) Feature extraction accuracy: For several cases, manually calculate their average temperature rise slope, high pressure duration percentage, etc., and compare them with the results automatically extracted by the system. The error should be within an acceptable range; 3) Decision table logic rationality: Check whether the intermediate scores output by the system conform to the logic defined in the decision table. For example, a case with a rapid temperature rise and a high high pressure percentage should have a lower first comfort intermediate score; 4) Fusion and mapping: Check whether the final comprehensive comfort score and comfort level identifier are consistent with the general comfort trend in the standard case library based on complete information, including subsequent patient feedback assessments. The key to successful testing lies in the transparency of the analysis process, logical consistency, and overall relevance to external assessments, rather than precise numerical matching.

[0114] Furthermore, before performing the data association operation, the comfort analysis engine also performs a data valid range truncation operation, specifically:

[0115] Based on the treatment area code and target tissue depth prediction level in the aforementioned disease data, the range of estimated effective treatment time is calculated.

[0116] Based on the estimated effective treatment duration range, the time-series data in the structured treatment plan data package is extracted, and thermal imaging frames, power output data, and pressure readings whose time span falls within the estimated effective treatment duration range are retained for subsequent comfort level classification analysis.

[0117] In this embodiment, before the engine performs data association and rule analysis, it first uses information such as the treatment area code and target tissue depth prediction level in the disease data to call a built-in empirical formula or lookup table to estimate the range of the estimated effective treatment time for this treatment. For example, large-area, deep treatment may require a longer effective treatment time.

[0118] Then, based on this estimated time range, the engine extracts all the time-series data in the structured treatment plan data package. It only retains data segments whose timestamps fall within this estimated effective treatment duration, including thermal imaging frames, power output data points, and pressure readings. Data outside the range, such as the preparation stage before treatment and the cooling stage after treatment, will be temporarily ignored and will not participate in subsequent comfort feature extraction and calculation.

[0119] The experimental test selected several cases with significant differences in the area and depth of the treatment zones. Based on clinical experience, a reasonable effective treatment time period was manually marked for each case. The system was then run, and the estimated effective treatment duration range automatically calculated by the system based on the patient's condition data was recorded. The estimated range was compared with the manually marked range to check for significant overlap. The aim was to verify whether this extraction logic could effectively focus on the core treatment phase and eliminate interference from irrelevant data.

[0120] Furthermore, referring to Figure 3 The construction process of the multidimensional decision table includes:

[0121] Historical treatment case data is collected in advance, including treatment plan data, disease data, and comfort evaluation tags actively provided by patients after treatment;

[0122] Based on the historical treatment case data, for different treatment scenarios and patient groups, key parameter combinations that affect comfort and their corresponding threshold boundaries are defined.

[0123] Based on the combination of key parameters and threshold boundaries, a mapping relationship between parameter input values ​​and intermediate comfort scores is established, forming the first multidimensional decision table and the second multidimensional decision table, which are then stored in the rule base of the comfort analysis engine.

[0124] In this embodiment, a large amount of complete historical treatment case data is collected from the historical archive;

[0125] Expert Definition: Experienced clinicians and biomedical engineers were invited as field experts. They reviewed these historical cases, especially those with clear feedback labels. The experts analyzed which combinations of key parameters and their thresholds significantly affected patient comfort in different treatment scenarios, such as skin rejuvenation vs. fat reduction, and in patient groups, such as young oily skin vs. older dry skin.

[0126] Based on expert analysis, decision tables are constructed using parameters as dimensions. For example, the first multidimensional decision table might be a two-dimensional table, with rows representing segments of the average temperature rise slope (0-1°C / s, 1-2°C / s, >2°C / s) and columns representing segments of the high-pressure duration percentage (0-10%, 10-30%, >30%). The value in each cell is determined by experts based on the comfort feedback distribution of cases falling within this parameter range in historical cases, and assigned a first comfort intermediate score. The second table is constructed similarly. These constructed decision tables are stored in the rule base of the comfort analysis engine for online analysis and querying.

[0127] The experimental testing section focuses more on the rationality and consistency of the process. An expert group can be organized to discuss a selected batch of historical cases and formulate guidelines for constructing a decision table. Then, another expert or an independent group can attempt to construct a decision table based on these guidelines. The constructed decision table can then be used to predict offline comfort scores for another batch of historical cases. The prediction results are compared with the actual patient feedback labels in the cases to calculate the accuracy or consistency of the predictions. The goal is to verify that the prediction trend of the decision table constructed based on expert knowledge has statistically significant consistency with the actual feedback, thereby proving the effectiveness of the decision table as the basis for analysis.

[0128] Furthermore, referring to Figure 4 S4 includes:

[0129] A parameter optimization service module is deployed on the preset cloud server. The parameter optimization service module reads the comfort data of the radiofrequency therapy for a certain treatment plan and the corresponding second type of data, namely the environmental data of the radiofrequency therapy.

[0130] The parameter optimization service module uses the comfort level identifier as a status indicator of the optimization target, and uses the ambient temperature, ambient humidity, equipment body temperature, treatment bed tilt angle, and coupling agent type identifier in the environmental data as environmental parameter variables to be optimized.

[0131] The parameter optimization service module maintains an environmental parameter configuration space, which consists of all possible discrete or continuous value ranges of the environmental parameter variable to be optimized.

[0132] For the current treatment plan, the parameter optimization service module generates parameter variation combinations in the environmental parameter configuration space based on the currently recorded environmental data values. Each parameter variation combination is a new set of parameters obtained by making a small perturbation to the environmental parameter variables.

[0133] For each generated parameter variation combination, the expected direction and magnitude of its impact on the overall comfort score are simulated and calculated. This simulation is based on a pre-stored parameter-comfort impact knowledge graph, which reflects the correlation patterns between changes in different environmental parameters and changes in comfort scores in historical data. One or more parameter variation combinations that maximize the expected increase in the overall comfort score are selected as recommended environmental parameter optimization schemes.

[0134] In this embodiment, a parameter optimization service module is deployed on the cloud server. This module is triggered when optimization of a treatment plan is needed.

[0135] First, based on the input treatment plan identifier, the module reads the comfort data of the corresponding radiofrequency therapy from the database, including the comprehensive score and level identifier, as well as the corresponding second type of data, namely environmental data: ambient temperature, humidity, equipment temperature, bed tilt angle, and coupling agent type.

[0136] The module uses comfort level identifiers, such as low comfort, as the current state, and defines five specific parameters from the environmental data—ambient temperature, ambient humidity, equipment body temperature, treatment bed tilt angle, and coupling agent type identifier—as the environmental parameter variables to be optimized in this optimization.

[0137] The ablation system technology is implemented here as follows: The module maintains an environmental parameter configuration space, which is the set of all possible values ​​for the variables to be optimized. For example, the ambient temperature can be any continuous value between 20°C and 26°C, and the coupling agent type can be discrete, such as {type A, type B, type C}.

[0138] For the current solution, the module uses currently recorded environmental data values, such as ambient temperature 23°C, humidity 50%, etc., as a baseline. Then, within the configuration space, it systematically generates a series of parameter variation combinations around this baseline, generating them by making small perturbations to one or more environmental parameters. For example, combination 1: temperature +1°C, other parameters unchanged; combination 2: humidity -5%, bed tilt angle +2 degrees; combination 3: coupling agent changed to type B, equipment temperature -0.5°C, etc. The magnitude and combination of perturbations follow a preset exploration strategy.

[0139] For each generated combination of parameter variations, the module needs to predict how the comfort score might have changed if this environmental setting had been used during the initial treatment. This is accomplished through simulation calculations. The calculations rely on a pre-stored parameter-comfort impact knowledge graph, which is learned or statistically derived from a large amount of historical data. This knowledge graph characterizes the correlation patterns and strengths between different environmental parameter changes, such as increased humidity, and changes in comfort scores, such as potential increases or decreases.

[0140] The module inputs the variant combinations into the knowledge graph for reasoning, and concludes that the expected improvement in comfort score relative to the current baseline may be positive, negative, or zero.

[0141] Finally, the module selects the one or more parameter variations that are expected to provide the greatest improvement and outputs them as recommended environmental parameter optimization schemes. For example, the output suggests: increasing the ambient temperature to 24°C and using type B coupling agent, which is expected to improve comfort.

[0142] The experiment designed a set of historical cases with different environmental parameters, and their comfort data were known. During testing, one case was selected as the current solution, and the optimization module was run to check: 1) whether the module correctly read the relevant data; 2) whether the generated parameter variation combinations truly involved minor perturbations around the baseline point, and whether the combinations were diverse; 3) whether the simulation calculation process was traceable. For example, a variation combination could be randomly selected, and its influence trend could be manually estimated based on the publicly available logic of the knowledge graph to see if it was consistent with the module's output direction; 4) whether the output optimization solution was reasonable in terms of physical and clinical common sense. The focus of the test was to verify the logical consistency of the optimization process and the rationality of the recommendations, rather than verifying its absolute effect in real treatment.

[0143] Furthermore, the workflow of the parameter optimization service module includes:

[0144] The process of ranking environmental parameters by importance is as follows: analyze the key data features on which the comfort data is based, identify the features that contribute highly to the comfort classification, and then backtrack the environmental parameter variables that are most strongly associated with these features.

[0145] Based on the importance ranking of the environmental parameters, the module dynamically adjusts the generation strategy of parameter mutation combinations in the environmental parameter configuration space. For environmental parameter variables with high importance, a finer-grained perturbation step size is set near their current value to generate more mutation possibilities. For parameter variables with low importance, a coarser perturbation step size is used or their current value is kept unchanged.

[0146] The construction and updating of the parameter-comfort influence knowledge graph are independent of the parameter optimization service module, and its construction process includes:

[0147] Complete treatment records that have been linked with comfort data and environmental data are periodically extracted from the aforementioned non-relational database;

[0148] Statistical analysis was performed on the extracted records to quantify the conditional probability of each environmental parameter variable co-occurring with different comfort levels or rating intervals in different value ranges.

[0149] The conditional probabilities and the interaction information between parameters are stored in a graph structure, where nodes represent parameters or comfort states, and edges represent influence relationships and intensity weights.

[0150] When simulating the expected impact, the parameter optimization service module inputs the current parameter variation combination into the knowledge graph, reasons along the edges of the graph, finds the historical parameter pattern most similar to the current variation combination, aggregates the comfort score results corresponding to the similar patterns, calculates the expected comfort score value of the current variation combination by weighted average, and then compares it with the currently recorded comprehensive comfort score value to obtain the expected improvement.

[0151] When outputting recommended environmental parameter optimization schemes, the module also adds a confidence assessment for each recommended scheme. The confidence assessment is calculated based on the number of historical similar patterns used for reasoning in the knowledge graph and the similarity weighting value.

[0152] The parameter optimization service module encapsulates the recommended environmental parameter optimization scheme, the corresponding expected improvement in comfort, and the confidence level assessment into an optimization suggestion report, which is then provided to the treatment equipment or operator terminal through the interface of the preset cloud server.

[0153] In this embodiment, the importance ranking of environmental parameters is as follows: Before generating variant combinations, the module analyzes the current comfort data. It identifies which key data features contribute most to the calculation of the overall score. For example, system records show that the first comfort level accounts for 70% of the score, primarily determined by the temperature rise slope. Then, based on a built-in feature-environmental parameter association mapping table, the module backtracks to find the environmental parameters that most directly affect the temperature rise slope. For example, the table indicates that ambient temperature, ambient humidity, and coupling agent type are key environmental factors affecting heat dissipation and heat conduction efficiency on the skin surface; therefore, these parameters are assigned a high importance level. Equipment body temperature is next, and bed tilt angle may affect local blood circulation but has a weaker impact on heat conduction, thus its importance is lower. Based on this, the module generates a dynamic importance ranking list.

[0154] Dynamic mutation generation strategy: Based on importance ranking, the module adjusts the generated parameter mutation combinations. For highly important parameters such as ambient temperature and humidity, denser, smaller-step perturbations are set around their current values. For example, temperature is explored within ±2°C with a step size of 0.5°C, generating more diverse possibilities. For less important parameters such as bed tilt angle, larger step sizes, such as ±5 degrees, may be used, or their values ​​may be kept constant during optimization exploration to reduce the search space and improve efficiency.

[0155] Independent Construction and Update of the Knowledge Graph: This section emphasizes that the knowledge graph is an independent and continuously updated component. Its construction process involves periodically (e.g., weekly) extracting all treatment records with completed comfort assessments and associated environmental data from the database. Large-scale statistical analysis is then performed on these records. The conditional probability of each environmental parameter, such as humidity, occurring in different value ranges and co-occurring with different comfort levels, such as high comfort, is calculated. Simultaneously, the interactions between parameters are analyzed, such as whether the combined effects of high temperature and high humidity on comfort differ from their individual effects. Finally, these probabilities and relationships are stored in the form of a graph: nodes represent parameter value ranges or comfort states, edges connect influencing nodes, and the weight of the edges represents the direction and intensity of the influence.

[0156] In-depth simulation calculations: When predicting the impact of a combination of variants, the module searches the knowledge graph for historical parameter patterns most similar to the current combination—that is, combinations of environmental parameter settings that have appeared in the past. It calculates similarity based on Euclidean distance or classification matching of parameter values ​​and identifies the Top-K most similar historical patterns. Then, it aggregates the actual comfort scores corresponding to these similar historical patterns, performs a weighted average (with weights determined by similarity), and calculates the expected comfort score for the current combination of variants. Subtracting this value from the current recorded comprehensive score yields the expected improvement.

[0157] Confidence Assessment: The module attaches a confidence assessment to each recommended solution. This assessment value is primarily based on two factors: first, the number of historical similar patterns used for inference; the larger the number, the stronger the statistical basis; second, the weighted sum of the similarities of these patterns; the higher the similarity, the more reliable the inference. Confidence is presented as a percentage or in the form of high / medium / low ratings.

[0158] Optimization Recommendation Report: Finally, the module integrates and packages all outputs—recommended specific parameter adjustment schemes, the expected improvement in comfort for each scheme, and the confidence assessment for each scheme—into a structured optimization recommendation report. This report is pushed to the designated treatment device control system or the tablet terminal of the operating physician via a RESTful API or message queue on the cloud server.

[0159] Experimental Test Overview: This is a comprehensive test. Multiple cases with varying characteristics are selected for deep optimization. Key test points include: 1) Importance Ranking Rationality: For cases primarily affecting thermal discomfort, does the system prioritize temperature and humidity? For cases primarily affecting pressure discomfort, does the ranking of parameters such as bed tilt change? 2) Dynamic Variation Strategy: Observe the generated list of variation combinations and check if more refined exploration of highly important parameters has been conducted. 3) Knowledge Graph Reasoning Interpretability: For a given output suggestion, the system is required to provide a brief explanation of the reasoning basis. For example, if the suggestion is based on 32 historical cases with similar environmental settings, and their average comfort score is XX points higher than the current setting, check if this explanation is consistent with the statistical logic of the knowledge graph. 4) Report Completeness: Check if the output report includes the three core elements: the solution, the expected improvement, and the confidence level, and if the format is standardized. This test aims to verify the advanced intelligent behavior of the optimization system and the richness and interpretability of its output.

[0160] Furthermore, the parameter optimization service module introduces a constraint check mechanism when generating parameter variation combinations, specifically:

[0161] Physical or safety constraints between environmental parameters are predefined. After each parameter variation combination is generated, its legality is verified according to the constraint condition checking mechanism. Only parameter variation combinations that pass the verification and meet the requirements of safety and physical feasibility are retained for subsequent simulation calculation and selection processes.

[0162] Furthermore, the preset radiofrequency therapy evaluation platform is a certified third-party medical efficacy evaluation community platform that allows the sharing of anonymized treatment data. All crawling operations are within the scope of the platform's user agreement and data privacy policy, and data is acquired through its application programming interface.

[0163] In this embodiment, a set of environmental parameter constraint rules is predefined within the optimization service module. These rules are based on physical principles, equipment safety specifications, and clinical experience, for example:

[0164] Rule 1: When it is recommended to increase the ambient temperature, in order to prevent the equipment from overheating, the recommended range for increasing the equipment body temperature must be reduced accordingly, and it must not exceed the safety limit.

[0165] Rule 2: Certain types of coupling agents, such as oily media, are not recommended for use with settings that may cause stuffiness due to extremely high ambient humidity, because of their heat dissipation properties.

[0166] Rule 3: The tilt angle of the treatment bed must not exceed the patient's physiological tolerance limit; for example, the head must not be too much lower than the heart.

[0167] After generating each parameter mutation combination, whether it is generated by baseline perturbation or other strategies, the module immediately sends it to the constraint checking mechanism for verification. This mechanism traverses all predefined constraint rules, checking whether the current mutation combination violates any of them. Only when all checks are passed is the combination considered valid and retained, proceeding to the subsequent simulation calculation and scoring stages. Any illegal combination is discarded directly.

[0168] During testing, some extreme or illogical parameter variations were intentionally introduced, such as suggesting that the ambient temperature be set to 35°C while the equipment operates at full capacity. The optimization process was run to check whether the system automatically filtered out these obviously unreasonable or dangerous combinations during the generation of variation combinations or subsequent stages. Simultaneously, the parameter values ​​in the final output optimization suggestion report were checked to ensure they all fell within reasonable and safe ranges, thereby verifying the effectiveness of the constraint mechanism.

[0169] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0170] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0171] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Therefore, all equivalent changes, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. A method for optimizing the comfort of radiofrequency therapy based on multimodal data, characterized in that, Includes the following steps: S1, a multimodal sensor is configured using a preset cloud server, and radiofrequency treatment plan data is collected based on the multimodal sensor; S2, crawl the patient's condition data and environmental data of radiofrequency treatment for each radiofrequency treatment plan from the preset radiofrequency treatment evaluation platform; S3. Based on the radiofrequency treatment plan data and the patient's condition data for each radiofrequency treatment plan, the comfort level of radiofrequency treatment is divided to obtain the comfort level data of radiofrequency treatment for each plan. S4. Based on the comfort data and environmental data of radiofrequency treatment for each scheme, the parameters of the radiofrequency treatment environment for that scheme are optimized using ablation system technology.

2. The method for optimizing radiofrequency therapy comfort based on multimodal data as described in claim 1, characterized in that, S1 specifically includes: A sensor virtualization management module is deployed on the preset cloud server, and virtual sensor instances corresponding to physical multimodal sensors are created through the sensor virtualization management module; The virtual sensor instance receives the raw sensor stream uploaded in real time by the physical multimodal sensor; A radiofrequency treatment plan data parsing unit is set up on the preset cloud server. The parsing unit performs time-series alignment and encapsulation on the original sensor stream to form a structured treatment plan data packet.

3. The method for optimizing radiofrequency therapy comfort based on multimodal data as described in claim 2, characterized in that, The S2 includes: Establish an authorized data interface connection with the preset radiofrequency therapy evaluation platform, and construct a batch query request based on a unique treatment plan identifier; An asynchronous data call is initiated to the preset radiofrequency therapy evaluation platform through the authorized data interface. The target data set for the call includes a first type of data and a second type of data. Design a data crawling scheduling strategy. This strategy initiates data calls in a non-blocking manner based on the sequence of treatment plan identifiers, and performs integrity verification and redundancy cleaning on the returned first type of data and second type of data. The first and second types of data, after passing the verification, are stored in different sets of the non-relational database of the preset cloud server according to their correspondence with the treatment plan identifier, and a data source tag and a time validity tag are attached to each data item.

4. The method for optimizing radiofrequency therapy comfort based on multimodal data as described in claim 3, characterized in that, The S3 includes: A comfort analysis engine is built on the preset cloud server. The comfort analysis engine first performs a data association operation, and associates and matches the structured treatment plan data package with the corresponding first type of data, namely the patient's condition data, according to the unique treatment plan identifier. The comfort analysis engine has multiple preset comfort analysis rule sets, each rule set being configured for different types of the disease data; For disease data containing specific skin type codes or target tissue depth prediction levels, the first comfort analysis rule set is invoked to extract the average temperature rise slope of the thermal imaging frame sequence throughout the treatment process and the cumulative time percentage of the local pressure sensing sequence exceeding a preset threshold from the structured treatment plan data package. The average temperature rise slope and the cumulative time percentage are input into a preset first multidimensional decision table, and the first comfort intermediate score is output. For medical data containing specific chief complaints or past related treatment history, the second comfort analysis rule set is invoked. This rule set defines: from the structured treatment plan data package, the fluctuation period and amplitude characteristics of the power output waveform data stream are analyzed, the deviation integral between the power output waveform data stream and the standard stable waveform is calculated, the number of abnormal temperature diffusion frames in the treatment edge area of ​​the thermal imaging video stream is identified, the deviation integral and the number of abnormal diffusion frames are input into a preset second multidimensional decision table, and the second comfort intermediate score is output. The comfort analysis engine establishes a scoring fusion module, which receives the first comfort intermediate score and / or the second comfort intermediate score, and performs weighted calculation and normalization on the input intermediate scores according to the weight configuration of each feature in the disease data to generate a comprehensive comfort score value for the treatment plan. The comfort analysis engine maps the comprehensive comfort score to discrete comfort level identifiers based on a preset comfort level threshold range. The comprehensive comfort score, the comfort level identifier, the intermediate score and key data features on which it is based are collectively encapsulated into the comfort data of the radiofrequency therapy of the treatment plan, and stored in the non-relational database.

5. The method for optimizing radiofrequency therapy comfort based on multimodal data as described in claim 4, characterized in that, Before performing data association operations, the comfort analysis engine also performs a data validity range truncation operation, specifically: Based on the treatment area code and target tissue depth prediction level in the aforementioned disease data, the range of estimated effective treatment time is calculated. Based on the estimated effective treatment duration range, the time-series data in the structured treatment plan data package is extracted, and thermal imaging frames, power output data, and pressure readings whose time span falls within the estimated effective treatment duration range are retained for subsequent comfort level classification analysis.

6. The method for optimizing radiofrequency therapy comfort based on multimodal data as described in claim 4, characterized in that, The process of constructing the multidimensional decision table includes: Historical treatment case data is collected in advance, including treatment plan data, disease data, and comfort evaluation tags actively provided by patients after treatment; Based on the historical treatment case data, for different treatment scenarios and patient groups, key parameter combinations that affect comfort and their corresponding threshold boundaries are defined. Based on the combination of key parameters and threshold boundaries, a mapping relationship between parameter input values ​​and intermediate comfort scores is established, forming the first multidimensional decision table and the second multidimensional decision table, which are then stored in the rule base of the comfort analysis engine.

7. The method for optimizing radiofrequency therapy comfort based on multimodal data as described in claim 4, characterized in that, The S4 includes: A parameter optimization service module is deployed on the preset cloud server. The parameter optimization service module reads the comfort data of the radiofrequency therapy for a certain treatment plan and the corresponding second type of data, namely the environmental data of the radiofrequency therapy. The parameter optimization service module uses the comfort level identifier as a status indicator of the optimization target, and uses the ambient temperature, ambient humidity, equipment body temperature, treatment bed tilt angle, and coupling agent type identifier in the environmental data as environmental parameter variables to be optimized. The ablation system technology is specifically implemented as follows: the parameter optimization service module maintains an environmental parameter configuration space, which consists of all possible discrete or continuous value ranges of the environmental parameter variable to be optimized. For the current treatment plan, the parameter optimization service module generates a series of parameter variation combinations in the environmental parameter configuration space, based on the currently recorded environmental data values. Each parameter variation combination is a new set of parameters obtained by making a small perturbation to the environmental parameter variables. For each generated parameter variation combination, the expected direction and magnitude of its impact on the overall comfort score are simulated and calculated. This simulation is based on a pre-stored parameter-comfort impact knowledge graph, which reflects the correlation patterns between changes in different environmental parameters and changes in comfort scores in historical data. One or more parameter variation combinations that maximize the expected increase in the overall comfort score are selected as recommended environmental parameter optimization schemes.

8. The method for optimizing radiofrequency therapy comfort based on multimodal data as described in claim 7, characterized in that, The workflow of the parameter optimization service module includes: The process of ranking environmental parameters by importance is as follows: analyze the key data features on which the comfort data is based, identify the features that contribute highly to the comfort classification, and then backtrack the environmental parameter variables that are most strongly associated with these features. Based on the importance ranking of the environmental parameters, the module dynamically adjusts the generation strategy of parameter mutation combinations in the environmental parameter configuration space. For environmental parameter variables with high importance, a finer-grained perturbation step size is set near their current value to generate more mutation possibilities. For parameter variables with low importance, a coarser perturbation step size is used or their current value is kept unchanged. The construction and updating of the parameter-comfort influence knowledge graph are independent of the parameter optimization service module, and its construction process includes: Complete treatment records that have been linked with comfort data and environmental data are periodically extracted from the aforementioned non-relational database; Statistical analysis was performed on the extracted records to quantify the conditional probability of each environmental parameter variable co-occurring with different comfort levels or rating intervals in different value ranges. The conditional probabilities and the interaction information between parameters are stored in a graph structure, where nodes represent parameters or comfort states, and edges represent influence relationships and intensity weights. When simulating the expected impact, the parameter optimization service module inputs the current parameter variation combination into the knowledge graph, reasons along the edges of the graph, finds the historical parameter pattern most similar to the current variation combination, aggregates the comfort score results corresponding to the similar patterns, calculates the expected comfort score value of the current variation combination by weighted average, and then compares it with the currently recorded comprehensive comfort score value to obtain the expected improvement. When outputting recommended environmental parameter optimization schemes, the module also adds a confidence assessment for each recommended scheme. The confidence assessment is calculated based on the number of historical similar patterns used for reasoning in the knowledge graph and the similarity weighting value. The parameter optimization service module encapsulates the recommended environmental parameter optimization scheme, the corresponding expected improvement in comfort, and the confidence level assessment into an optimization suggestion report, which is then provided to the treatment equipment or operator terminal through the interface of the preset cloud server.

9. The method for optimizing radiofrequency therapy comfort based on multimodal data as described in claim 8, characterized in that, The parameter optimization service module also introduces a constraint check mechanism when generating parameter variation combinations, specifically: Physical or safety constraints between environmental parameters are predefined. After each parameter variation combination is generated, its legality is verified according to the constraint condition checking mechanism. Only parameter variation combinations that pass the verification and meet the requirements of safety and physical feasibility are retained for subsequent simulation calculation and selection processes.

10. A method for optimizing radiofrequency therapy comfort based on multimodal data-driven approaches according to any one of claims 1 to 9, characterized in that, The preset radiofrequency therapy evaluation platform is a certified third-party medical efficacy evaluation community platform that allows the sharing of anonymized treatment data. All crawling operations are within the scope of the platform's user agreement and data privacy policy, and data is acquired through its application programming interface.

Images