Medical accelerator fault early warning system and method based on large language model

Abstract

A medical accelerator fault early warning system based on a large language model comprises a data acquisition module, a data processing module, a large language model prediction module and a user interaction module, and the large language model prediction module comprises an equipment historical maintenance unit, a maintenance processing collection unit, a fault prediction unit and a fault analysis unit. The large language model has strong data analysis and processing capabilities, and can better process complex medical data and equipment parameters. A large amount of historical data can be learned and analyzed by using the large language model, and rules and relationships hidden in the data are mined, so that the accuracy of fault early warning is improved. Compared with a traditional fault early warning system, the fault early warning system has higher intelligence and adaptability.

Description

Technical Field

[0001] The present invention relates to the technical field of medical accelerator early warning, and in particular to a medical accelerator fault early warning system and method based on a large language model. Background Art

[0002] Medical accelerators are devices that use high-energy rays (such as photons, electrons, protons or heavy ions) for radiation therapy. They direct these high-energy rays to tumor tissue to damage the DNA of cancer cells, thereby inhibiting tumor growth or eliminating cancer cells.

[0003] As a key device for radiotherapy, the operating status of medical accelerators directly affects the treatment effect and safety of patients. Traditional fault detection methods mainly rely on manual inspections and regular maintenance, which have problems such as lag and subjectivity. With the rapid development of artificial intelligence technology, especially the widespread application of large language models in the field of natural language processing, new solutions have been provided for the fault prediction and early warning of medical accelerators. However, there is no mature large language model system for medical accelerator fault prediction and early warning on the market. Summary of the invention

[0004] The purpose of the present invention is to provide a medical accelerator fault warning system and method based on a large language model, which solves the problems of hysteresis and subjectivity in the above-mentioned medical accelerator fault detection.

[0005] To achieve the above object, the present invention provides a medical accelerator fault warning system based on a large language model, comprising:

[0006] A data acquisition module, used to obtain the patient's medical plan information, the current equipment parameters of the medical accelerator and the historical maintenance parameters in the hospital information system;

[0007] A data processing module is used to pre-process the patient's medical plan information and the working parameters of the medical accelerator and remove abnormal data points to obtain processed patient-related medical data and medical accelerator parameter data;

[0008] Large language model prediction module, including:

[0009] The equipment history maintenance unit is used to analyze the cause of the failure of the medical accelerator through the equipment history maintenance process, mark the medical accelerator as an analysis object, analyze the parts of the analysis object with frequent failures, divide the analysis object into multiple failure areas, divide the parts of the analysis object into high-frequency failure parts, medium-frequency failure parts and low-frequency failure parts through analysis, generate high-frequency failure marks, medium-frequency failure marks and low-frequency failure marks, and mark them with corresponding failure areas;

[0010] The maintenance processing collection unit is used to organize the historical maintenance records and equipment parameters of the analysis object, analyze the causes of the failures of the maintenance processing schemes of each fault area of ​​the analysis object, divide the maintenance processing schemes into occasional maintenance schemes and specific maintenance schemes through analysis, generate occasional maintenance marks and specific maintenance marks, and collect the corresponding maintenance schemes;

[0011] The fault prediction unit is used to bring the obtained patient medical data and medical accelerator parameter data into the aging rate calculation formula, and predict the aging of the medical accelerator after the patient medical plan is adopted according to the calculation result of the aging formula, and output the aging rate result.

[0012] Its aging rate (OL rate ) is calculated as:

[0013] ;

[0014] ;

[0015] ;

[0016] ;

[0017] ;

[0018] Among them, Freq is the frequency of use of the medical accelerator, Time is the total use time of the medical accelerator, DoseType is the efficiency factor after irradiation dose classification, and Cone is the standardized dose size

[0019] Influencing factors, Load is the treatment dose executed within the set time, Repairs is the number of fault repairs of the medical accelerator, FR is the failure rate of the medical accelerator within the set time,

[0020] , ,...... are the corresponding weight coefficients,

[0021] α, β, γ, δ, and η are the corresponding nonlinear function parameters,

[0022] Thresh1, Thresh2, and Thresh3 are the thresholds of the variables, respectively. X>Thresh is a binary function. When X is greater than Thresh, the value is 1, otherwise it is 0.

[0023] The fault analysis unit is used to calculate the aging rate (OL) of each part of the analysis object. rate), the number of fault repairs (Repairs) and the failure rate (FR), predict the failure of each component of the medical accelerator after the patient's medical plan is adopted, and generate a recommended solution for the failure;

[0024] The user interaction module is used to interact with users through the interface of fault prediction results and recommended solutions, and provides prediction report generation, recommended solution query and prediction reminder functions.

[0025] Preferably, the data processing module sets the efficacy range for each type of dose by constructing a database of radiation dose and efficacy correspondence rules, and uses a decision tree, support vector machine (SVM) or neural network algorithm to take the radiation dose name in the medical plan as an input feature, and matches the efficacy factor after the known radiation dose classification as an output label, and then assigns the corresponding efficacy factor according to the category;

[0026] The data processing module analyzes the medical plan text, searches for keywords, accurately extracts the value of the radiation dose, and calculates the minimum value C of the radiation dose in the historical data. min and the maximum value C max , for the new irradiation dose C new After normalization, the standardized dose size influence factor is obtained.

[0027] Calculation formula: .

[0028] Preferably, the data processing module uses a box plot method to identify and remove abnormal values ​​of medical accelerator parameters, calculates their quartiles Q1 and Q3, and determines the upper and lower limits of the data points.

[0029] The calculation formula is as follows:

[0030] Lower limit = , upper limit = ,

[0031] Where IQR= .

[0032] Preferably, a medical accelerator fault warning method based on a large language model adopts any one of the above-mentioned medical accelerator fault warning systems based on a large language model, comprising:

[0033] S1. Obtaining the patient's medical plan information, and the current equipment parameters and historical maintenance parameters of the medical accelerator in the hospital information system;

[0034] S2. Preprocess the acquired data.

[0035] Clean and standardize the patient's medical plan information to obtain a unified data format and parameters;

[0036] For the current equipment parameters and historical maintenance parameters of the medical accelerator, the box plot method is used to identify and eliminate outliers, calculate their quartiles Q1 and Q3, and determine the upper and lower limits of the data points.

[0037] The calculation formula is as follows:

[0038] Lower limit = , upper limit = ,

[0039] Where IQR= ;

[0040] S3, calculating the fault frequency value of each fault area of ​​the medical accelerator, and classifying the fault areas by analyzing the fault conditions: high-frequency fault area, medium-frequency fault area and low-frequency fault area, and ranking each fault area according to the fault frequency value and generating corresponding marks,

[0041] The calculation formula of the frequency value F of the fault part is:

[0042] ;

[0043] Among them, T0 is the average use time of each part of the medical accelerator, N is the number of times each part of the medical accelerator is repaired for a fault, I0 is the maximum idle time allowed under normal maintenance of each part of the medical accelerator, M0 is the average repair time of each part of the medical accelerator (the average repair time for this type of fault), D0 is the regular inspection cycle of the medical accelerator, and a, b, c, d, and e are the corresponding weight coefficients respectively;

[0044] S4. Sorting out the historical maintenance records and equipment parameters of the medical accelerator, analyzing the causes of the failures of the historical maintenance solutions for each fault area of ​​the medical accelerator, dividing the maintenance solutions into occasional maintenance solutions and specific maintenance solutions through analysis, generating occasional maintenance marks and specific maintenance marks, and grouping the corresponding maintenance solutions;

[0045] S5. The high-frequency fault parts, medium-frequency fault parts and low-frequency fault parts classified in S3 are arranged according to their levels and the equipment aging rate (OL rate ), the data pre-processed by S2 is used to enter the aging rate (OL rate ) calculation formula,

[0046] Its aging rate (OL rate ) is calculated as:

[0047] ;

[0048] ;

[0049] ;

[0050] ;

[0051] ;

[0052] Among them, Freq is the frequency of use of the medical accelerator, Time is the total use time of the medical accelerator, DoseType is the efficiency factor after irradiation dose classification, Cone is the standardized dose size influencing factor, Load is the treatment dose performed within the set time, Repairs is the number of fault repairs of the medical accelerator, and FR is the fault occurrence rate of the medical accelerator within the set time.

[0053] , ,...... are the corresponding weight coefficients,

[0054] α, β, γ, δ, and η are the corresponding nonlinear function parameters,

[0055] Thresh1, Thresh2, and Thresh3 are the thresholds of the variables, respectively. X>Thresh is a binary function. When X is greater than Thresh, the value is 1, otherwise it is 0.

[0056] The aging rate of each part of the equipment is obtained through the budget. If the aging rate (OL rate ) ≥ the set threshold (threshold), then proceed to S6; if the aging rate (OL rate )<set threshold value (threshold), then proceed to S8,

[0057] The set threshold is the value of the performance aging warning line of the medical accelerator equipment. When the budget aging rate is lower than the set threshold, it means that the equipment performance is seriously affected. Otherwise, it means that the equipment performance is sufficient to meet the use requirements.

[0058] S6, analyzing the aging rate (OLrate), the number of fault repairs (Repairs) and the failure rate (FR) of each part of the object, predicting the failure of each component of the medical accelerator after the patient's medical plan is adopted and generating a recommended plan for the failure;

[0059] S7, interacting with users through the interface with fault prediction results and suggested solutions, providing prediction report generation, suggested solution query and prediction reminder functions;

[0060] S8. Save the fault prediction results and generate historical records.

[0061] Preferably, after calculating the part failure frequency value F in S3, a fuzzy membership function calculation is performed on the part again.

[0062] The calculation formula of its fuzzy membership function is:

[0063] High-frequency fault membership degree μHF(f) ;

[0064] Medium-frequency fault membership degree μMF(f) ;

[0065] Low-frequency fault membership degree μLF(f) = 1 - μHF(f) - μMF(f);

[0066] where j and k are preset parameters.

[0067] j indicates that when the number of failures of the part exceeds the threshold within the set period, the part enters a high-risk state.

[0068] k indicates the speed at which the part failure changes from normal to abnormal.

[0069] When μHF(f) > μMF(f) and μHF(f) > μLF(f), then this part will be classified as a high-frequency fault part and generate a label: high-frequency fault part [membership degree μHF(f)].

[0070] When μMF(f) > μHF(f) and μMF(f) > μLF(f), then this part will be classified as a medium-frequency fault part and generate a label: medium-frequency fault part [membership degree μMF(f)].

[0071] When μLF(f) > μHF(f) and μLF(f) > μMF(f), then this part will be classified as a low-frequency fault part and generate a label: low-frequency fault part [membership degree μLF(f)];

[0072] And rank according to the magnitude of the membership degree value.

[0073] Preferably, S6 analyzes the historical maintenance records and equipment parameters of the medical accelerator as follows:

[0074] S61. Calculate the number of fault repairs per unit usage frequency r1 = Repairs / Freq,

[0075] and the number of fault repairs per unit treatment dose r2 = Repairs / Load.

[0076] If r1 > z1 and r2 < z2, then determine this maintenance plan as an occasional maintenance plan, where z1 and z2 are preset values. If the conditions are not met, enter the next judgment;

[0077] S62. Draw the collected data of failure rate (FR) and treatment dose (Load) into a scatter plot, with treatment dose (Load) and failure rate (FR) as the vertical axis, mark the points corresponding to each set of data on the graph, and analyze the distribution of the points through the scatter plot.

[0078] If the distribution of points is in the shape of a straight line with an obvious linear relationship, it is judged as a specific maintenance plan. If the distribution of points is scattered and has no obvious linear relationship, it proceeds to the next step of judgment.

[0079] S63. Draw the collected data of failure rate (FR) and frequency of use (Freq) into a scatter plot, with frequency of use (Freq) and failure rate (FR) as the vertical axis, mark the points corresponding to each set of data on the graph, and analyze the distribution of the points through the scatter plot.

[0080] If the distribution of points is in the shape of a straight line with an obvious linear relationship, it is judged as a specific maintenance plan. If the distribution of points is scattered and does not have an obvious linear relationship, it is judged as an occasional maintenance plan.

[0081] Preferably, the threshold value calculation formula is: threshold value (threshold) ,

[0082] in, represents the average value of historical aging rate data, represents the standard deviation of historical aging rate data, is the coefficient, The value range is [1,2].

[0083] Beneficial effects of the present invention:

[0084] 1. The present invention uses a large language model to learn and analyze a large amount of historical data, dig out the rules and relationships hidden in the data, and thus improve the accuracy of fault warning. Compared with the traditional fault warning system, the fault warning system of the present invention has higher intelligence and adaptability.

[0085] 2. The aging rate formula of the present invention integrates multiple factors that affect equipment aging, such as frequency of use, usage time, irradiation dose type, etc., which makes the evaluation more comprehensive; through weight coefficients (w1 to w7) and nonlinear function parameters (α, β, γ, δ, η), the influence of each factor can be adjusted according to the characteristics of different equipment or environments to make the formula more targeted; by setting thresholds (Thresh1, Thresh2, Thresh3), the significant impact of certain factors on the aging rate after exceeding a specific value can be highlighted, so that the model can reflect the accelerated aging phenomenon under critical conditions.

[0086] 3. The early warning method of the present invention adopts a multi-factor fusion analysis method in multiple links, such as fault frequency calculation, equipment aging rate calculation, fault prediction, etc. Compared with the traditional single factor analysis method, the present invention can more comprehensively and accurately reflect the operating status, fault risk and aging of the medical accelerator. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.

[0088] Figure 1 It is a structural block diagram of the medical accelerator fault warning system of the present invention.

[0089] Figure 2 It is a structural block diagram of the large language model prediction module of the present invention.

[0090] Figure 3 It is a flow chart of the medical accelerator fault warning method of the present invention. DETAILED DESCRIPTION

[0091] Embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, but should not be construed as limiting the present invention.

[0092] like Figure 1 As shown, a medical accelerator fault warning system based on a large language model includes a data acquisition module, a data processing module, a large language model prediction module and a user interaction module, wherein the large language model prediction module includes an equipment history maintenance unit, a maintenance processing collection unit, a fault prediction unit and a fault analysis unit. The large language model has powerful data analysis and processing capabilities and can better handle complex medical data and equipment parameters. The large language model can be used to learn and analyze a large amount of historical data, and to dig out the laws and relationships hidden in the data, thereby improving the accuracy of fault warning. Compared with traditional fault warning systems, the fault warning system of the present invention has higher intelligence and adaptability.

[0093] The data acquisition module is used to obtain the patient's medical plan information in the hospital information system, and the current equipment parameters and historical maintenance parameters of the medical accelerator. The data acquisition module can obtain the patient's medical plan information in the hospital information system, the current equipment parameters and historical maintenance parameters of the medical accelerator. It ensures that the system has a comprehensive data source and provides rich basic data for subsequent fault warnings. By integrating multiple types of data, the system can more comprehensively consider the various factors that affect the failure of the medical accelerator and improve the accuracy of fault prediction.

[0094] The data processing module is used to pre-process the patient's medical plan information and the working parameters of the medical accelerator and remove abnormal data points to obtain processed patient-related medical data and medical accelerator parameter data. Through the data processing module, the patient's medical plan information and the working parameters of the medical accelerator can be pre-processed and abnormal data points can be removed. The pre-processing operation can remove data noise that may interfere with the analysis, making the subsequent analysis more accurate. The processed data is more reliable and can more realistically reflect the operating status of the medical accelerator and factors related to patient treatment, which helps to improve the prediction reliability of the entire system.

[0095] The equipment history maintenance unit is used to analyze the cause of the failure of the medical accelerator through the equipment history maintenance process, mark the medical accelerator as the analysis object, analyze the parts of the analysis object with frequent failures, divide the analysis object into multiple fault areas, divide the parts of the analysis object into high-frequency fault parts, medium-frequency fault parts and low-frequency fault parts through analysis, generate high-frequency fault marks, medium-frequency fault marks and low-frequency fault marks, and mark them with the corresponding fault areas. The equipment history maintenance unit of this early warning system divides the medical accelerator into multiple fault areas by analyzing the historical maintenance process of the medical accelerator, and further divides its parts into high-frequency fault parts, medium-frequency fault parts and low-frequency fault parts. This classification method helps to have a clearer understanding of the fault distribution of the medical accelerator. Different maintenance and prevention strategies can be adopted for fault parts of different frequencies. For example, monitoring and preventive maintenance can be strengthened for high-frequency fault parts, thereby improving the overall reliability of the medical accelerator, reducing the possibility of failure, reducing maintenance costs and increasing the service life of the equipment.

[0096] The maintenance processing collection unit is used to organize the historical maintenance records and equipment parameters of the analysis object, analyze the causes of the maintenance processing plans for each fault area of ​​the analysis object, divide the maintenance processing plans into occasional maintenance plans and specific maintenance plans through analysis, generate occasional maintenance marks and specific maintenance marks, and collect the corresponding maintenance plans. This system organizes the historical maintenance records and equipment parameters of the analysis object, analyzes the causes of the maintenance processing plans for each fault area, and divides the maintenance processing plans into occasional maintenance plans and specific maintenance plans. It can help maintenance personnel quickly identify different types of maintenance situations and improve maintenance efficiency. For occasional maintenance plans, the reasons for their contingency can be further analyzed, and for specific maintenance plans, more in-depth research and improvements can be conducted on their specific causes of failure, which helps to optimize the maintenance process and improve the overall performance of the equipment.

[0097] The fault prediction unit is used to bring the obtained patient medical data and medical accelerator parameter data into the aging rate calculation formula, and predict the aging of the medical accelerator after the patient medical plan is adopted according to the calculation result of the aging formula, and output the aging rate result. The fault prediction unit of the present invention predicts the aging of the equipment by bringing the patient medical data and the medical accelerator parameter data into the aging rate calculation formula. The aging rate calculation formula takes into account multiple factors, such as the frequency of use of the medical accelerator (Freq), the total use time (Time), the efficiency factor (DoseType) after the irradiation dose classification, the standardized dose size influence factor (Cone), the treatment dose (Load) performed within the set time, the number of fault repairs (Repairs) and the failure rate (FR). These factors cover all aspects of the operation of the medical accelerator, including the use of the equipment itself, the irradiation dose factors related to the treatment, and the maintenance history. By comprehensively considering these factors, the system can more accurately predict the aging of the equipment, discover potential fault risks in advance, and provide a strong basis for preventive maintenance.

[0098] Its aging rate (OL rate ) is calculated as:

[0099] ;

[0100] ;

[0101] ;

[0102] ;

[0103] ;

[0104] Among them, Freq is the frequency of use of the medical accelerator, Time is the total use time of the medical accelerator, DoseType is the efficiency factor after irradiation dose classification, Cone is the standardized dose size influencing factor, Load is the treatment dose performed within the set time, Repairs is the number of fault repairs of the medical accelerator, and FR is the fault occurrence rate of the medical accelerator within the set time.

[0105] , ,...... are the corresponding weight coefficients,

[0106] α, β, γ, δ, and η are the corresponding nonlinear function parameters,

[0107] Thresh1, Thresh2, and Thresh3 are the thresholds of the variables respectively. X>Thresh is a binary function. When X is greater than Thresh, the value is 1, otherwise it is 0.

[0108] The aging rate calculation formula is a weighted summation, where each factor is multiplied by the corresponding weight coefficient ( , etc.), and according to different nonlinear functions (such as , For example, for the frequency of use (Freq), the higher the frequency of use, the faster the device ages, and it This item affects the aging rate. When the usage frequency Freq is greater than the threshold Thresh1, this item will be The aging rate is affected in different ways. This function increases with the increase of Freq, and the growth rate is determined by α, which means that the higher the frequency of use, the greater the impact on the aging rate, but the growth method is nonlinear. Therefore, adding nonlinear functions to the formula can better describe these relationships and be more in line with the actual situation. Similarly, other factors such as total usage time (Time) are determined by The total use time affects the aging rate, which reflects the quadratic growth relationship between the total use time and the aging rate (when Time is greater than Thresh2), indicating that the aging rate of the equipment will accelerate with the increase of time. This formula combines multiple factors that affect equipment aging, such as frequency of use, use time, type of irradiation dose, etc., which makes the evaluation more comprehensive; through weight coefficients (w1 to w7) and nonlinear function parameters (α, β, γ, δ, η), the influence of each factor can be adjusted according to the characteristics of different equipment or environments, making the formula more targeted; by setting thresholds (Thresh1, Thresh2, Thresh3), the significant impact of certain factors on the aging rate after exceeding a specific value can be highlighted, so that the model can reflect the accelerated aging phenomenon under critical conditions.

[0109] For example, in one example, the parameters of the medical accelerator site are as follows:

[0110] Freq=300 times / week, Thresh1=200, Time=3 years, Thresh2=2 years,

[0111] DoseType=0.8 (standardized radiation dose type influence factor),

[0112] Conc=0.7 (normalized dose size factor),

[0113] Load=200 times / week, Repairs=5 times / year, Thresh3=3, FR=0.05 times / month.

[0114] set up =0.2, =0.15, =0.1, =0.15, =0.1, =0.15, =0.15;

[0115] α=0.005, β=0.05, γ=0.2, δ=0.1, η=0.002.

[0116] Then, by substituting into the aging rate formula, we can get:

[0117] OLrate=0.2×(1−e −0.005×300 )×[300>200]+0.15×(0.05×3 2 )×[3>2]+0.1×(1−0.2×0.8)+0.15×(0.1×0.7 1.5 )+0.1×(0.002×200 0.5)+0.15×5×[5>3]+0.15×0.05.

[0118] OLrate≈1.241.

[0119] In this case, the final OLrate calculation results show that the aging rate of this part is quite high and requires special attention and possible maintenance upgrades.

[0120] In addition, the method of obtaining the parameters of the nonlinear function of the present invention first collects historical data related to the performance of the medical accelerator, which should include variables such as frequency of use, time of use, type of irradiation dose, irradiation dose, treatment load, number of fault repairs and failure rate. Secondly, by drawing a scatter plot and calculating the correlation coefficient, it is possible to preliminarily understand whether there is a nonlinear trend between the variables. According to the results of the data analysis, a nonlinear function form is selected, such as an exponential function, a power function, a logarithmic function or a sigmoid function. Then, a nonlinear regression analysis is performed using a statistical software package (such as R, Python's scikit-learn library or SPSS) to estimate the parameters of the nonlinear function for finding the best fitting parameters. Next, the goodness of fit and predictive ability of the model are evaluated by cross-validation, residual analysis and determination coefficient (R²) to confirm the reliability of the model and the accuracy of parameter estimation. Finally, a sensitivity analysis is performed to evaluate the impact of parameter changes on the calculation results of the aging rate, to identify which parameters are most sensitive to the model output and to ensure the robustness of the model.

[0121] As for setting the threshold, the present invention first collects data related to equipment aging, including frequency of use, time, irradiation dose type, size, treatment load, number of maintenance times, failure rate, etc. Secondly, statistical analysis is performed on the collected data, including calculation of mean, median, standard deviation, maximum value, minimum value, etc., to understand the distribution of the data, and use correlation coefficient and other methods to analyze the correlation between different variables. Then, based on the distribution of the data, a certain quantile (such as 25th, 50th, 75th, etc.) is selected as the threshold. Next, a part of the data is used to establish an aging model, and then another part of the data is used to verify the accuracy of the model, and the threshold is adjusted according to the performance of the model. Finally, through continuous iteration, the model performance under different thresholds is compared to select the optimal threshold.

[0122] The fault analysis unit is used to calculate the aging rate (OL) of each part of the analysis object. rate), the number of fault repairs (Repairs) and the failure rate (FR), predict the fault conditions of each component of the medical accelerator after the patient's medical plan is adopted, and generate a recommended solution for the fault condition. The fault analysis unit of the present invention can more comprehensively predict the fault conditions of each component of the medical accelerator after the patient's medical plan is adopted by incorporating the key factors of the aging rate (OLrate) of each part of the analysis object, the number of fault repairs (Repairs) and the failure rate (FR). This comprehensive consideration of multiple factors avoids the limitations of single factor prediction and makes the prediction result more accurate and reliable. For example, only considering the failure rate may not accurately reflect the actual failure risk of the component, but after adding the aging rate and the number of repairs, it can be evaluated from multiple angles such as the use loss and historical maintenance of the component, so as to more accurately predict the fault condition. On the basis of predicting the fault condition of the component, the corresponding recommended solution can be generated. It provides direct guidance for the maintenance and management of the medical accelerator. For medical institutions, timely measures such as preventive maintenance and component replacement can be taken according to these recommended solutions, thereby extending the service life of the medical accelerator and reducing maintenance costs, while also ensuring the safety and reliability of medical equipment, which is conducive to improving the efficiency of medical services.

[0123] The user interaction module is used to interact with the user through the interface for the fault prediction results and the suggested solutions, and provides the functions of prediction report generation, suggestion solution query and prediction reminder. The user interaction module of the present invention displays the possible fault types, occurrence probability and other information in the form of clear tables or charts. For example, the prediction report may show that the probability of the acceleration tube failure is 30%. The suggested solutions corresponding to different fault types are listed separately to facilitate users to quickly locate. At the same time, the module can list the suggested solutions corresponding to different fault types separately to facilitate users to quickly locate; for emergency faults (such as impending acceleration tube failures), the most urgent solutions are placed in front so that users can take measures first. When an impending fault is predicted, the module will pop up a reminder window in the user interface, briefly displaying the fault type and urgency. Users can set the reminder frequency according to their needs, such as reminding once every hour, or only reminding when the probability of a fault exceeds a certain threshold (such as 50%).

[0124] In one embodiment, the data processing module sets the efficacy range for each type of dose by constructing a database of radiation dose and efficacy correspondence rules, and uses a decision tree, support vector machine (SVM) or neural network algorithm to take the radiation dose name in the medical plan as an input feature, and matches the efficacy factor after the known radiation dose classification as an output label, and then assigns the corresponding efficacy factor according to the category;

[0125] The data processing module analyzes the medical plan text, searches for keywords, accurately extracts the value of the radiation dose, and calculates the minimum value C of the radiation dose in the historical data. min and the maximum value C max , for the new irradiation dose C new After normalization, the standardized dose size influence factor is obtained.

[0126] Calculation formula: .

[0127] The data processing module of the present invention constructs a database of rules corresponding to irradiation doses and efficacy, and sets an efficacy range for each type of dose. Using advanced algorithms such as decision trees, support vector machines (SVMs) or neural networks, the irradiation dose names in the medical plan are used as input features, and the efficacy factors after irradiation dose classification are used as output labels for matching and assigning corresponding efficacy factors. This method can accurately determine the efficacy factor according to the irradiation dose name, thereby improving the accuracy of irradiation dose efficacy evaluation. For example, in complex medical plans, the efficacy of different irradiation doses may be affected by multiple factors. Through this module, the efficacy range can be accurately determined, which helps medical personnel better understand the effects of irradiation doses.

[0128] Furthermore, the data processing module accurately extracts the value of the irradiation dose by searching for keywords through parsing the medical plan text. Then, by counting the minimum value Cmin and the maximum value Cmax of the irradiation dose in the historical data, the new irradiation dose Cnew is normalized to obtain the standardized dose size influence factor. This processing method can convert irradiation doses with different value ranges into a unified standard scale, which is convenient for comparison and analysis between different medical plans or different batches of irradiation dose data. For example, when studying the impact of irradiation dose on treatment effect, the standardized dose size influence factor can more intuitively reflect the relative impact of irradiation dose, which helps medical researchers better analyze and optimize medical plans.

[0129] In one embodiment, the data processing module uses a box plot method to identify and remove abnormal values ​​of medical accelerator parameters, calculates their quartiles Q1 and Q3, and determines the upper and lower limits of the data points.

[0130] The calculation formula is as follows:

[0131] Lower limit = , upper limit = ,

[0132] Where IQR= .

[0133] The calculation method based on quartiles of the present invention does not depend on the distribution form of data (such as normal distribution, skewed distribution, etc.), and can be applied to medical accelerator working parameter data of various distribution types. For example, the energy output data of the medical accelerator may not be distributed in an ideal normal distribution due to reasons such as equipment aging and sample differences between different batches, but the box plot method can still identify outliers well. 1.5 times the IQR ensures that the real outliers can be captured while having a certain robustness. If the multiple is set too small, some normal but slightly off-center data may be misjudged as outliers. For example, in the working parameters of the medical accelerator, it is normal for the parameters to fluctuate slightly due to environmental perturbations (such as small voltage fluctuations) occasionally. If the multiple is too small, these normal fluctuations may be mistakenly eliminated. If the multiple is set too large, some real outliers may be retained, thereby affecting the subsequent judgment of the normal working range of the medical accelerator and the formulation of medical plans based on these data. For example, if a key parameter of a medical accelerator suddenly exceeds the normal range by a large margin (possibly due to equipment failure), if the multiple is too large, this abnormal value may not be identified as an abnormality and may be used for analysis, leading to incorrect conclusions.

[0134] The upper and lower limits determined in this way can reflect the overall structure of the data. In the working parameter data set of the medical accelerator, data points outside the upper and lower limits are regarded as outliers, which helps to maintain the "purity" of the data and makes subsequent analysis based on these data (such as determining the normal working range of the equipment, matching analysis with the medical plan, etc.) more accurate. For example, when analyzing the parameter stability of the medical accelerator under different working conditions, accurately eliminating outliers can more clearly see the distribution range and change trend of the parameters under normal working conditions, which helps to promptly discover potential problems that may exist in the equipment or optimize the part of the medical plan related to the equipment parameters.

[0135] A medical accelerator fault warning method based on a large language model, comprising:

[0136] S1. Obtaining the patient's medical plan information, and the current equipment parameters and historical maintenance parameters of the medical accelerator in the hospital information system;

[0137] S2. Preprocess the acquired data.

[0138] Clean and standardize the patient's medical plan information to obtain a unified data format and parameters;

[0139] For the current equipment parameters and historical maintenance parameters of the medical accelerator, the box plot method is used to identify and eliminate outliers, calculate their quartiles Q1 and Q3, and determine the upper and lower limits of the data points.

[0140] The calculation formula is as follows:

[0141] Lower limit = , upper limit = ,

[0142] Where IQR= ;

[0143] S3, calculating the fault frequency value of each fault area of ​​the medical accelerator, and classifying the fault areas by analyzing the fault conditions: high-frequency fault area, medium-frequency fault area and low-frequency fault area, and ranking each fault area according to the fault frequency value and generating corresponding marks,

[0144] The calculation formula of the frequency value F of the fault part is:

[0145] ;

[0146] Among them, T0 is the average use time of each part of the medical accelerator, N is the number of times each part of the medical accelerator is repaired for a fault, I0 is the maximum idle time allowed under normal maintenance of each part of the medical accelerator, M0 is the average repair time of each part of the medical accelerator (the average repair time for this type of fault), D0 is the regular inspection cycle of the medical accelerator, and a, b, c, d, and e are the corresponding weight coefficients respectively;

[0147] S4. Sorting out the historical maintenance records and equipment parameters of the medical accelerator, analyzing the causes of the failures of the historical maintenance solutions for each fault area of ​​the medical accelerator, dividing the maintenance solutions into occasional maintenance solutions and specific maintenance solutions through analysis, generating occasional maintenance marks and specific maintenance marks, and grouping the corresponding maintenance solutions;

[0148] S5. The high-frequency fault parts, medium-frequency fault parts and low-frequency fault parts classified in S3 are arranged according to their levels and the equipment aging rate (OL rate ), the data pre-processed by S2 is used to enter the aging rate (OL rate ) calculation formula,

[0149] Its aging rate (OL rate ) is calculated as:

[0150] ;

[0151] ;

[0152] ;

[0153] ;

[0154] ;

[0155] Among them, Freq is the frequency of use of the medical accelerator, Time is the total use time of the medical accelerator, DoseType is the efficiency factor after irradiation dose classification, Cone is the standardized dose size influencing factor, Load is the treatment dose performed within the set time, Repairs is the number of fault repairs of the medical accelerator, and FR is the fault occurrence rate of the medical accelerator within the set time.

[0156] , ,...... are the corresponding weight coefficients,

[0157] α, β, γ, δ, and η are the corresponding nonlinear function parameters,

[0158] Thresh1, Thresh2, and Thresh3 are the thresholds of the variables, respectively. X>Thresh is a binary function. When X is greater than Thresh, the value is 1, otherwise it is 0.

[0159] The aging rate of each part of the equipment is obtained through the budget. If the aging rate (OL rate ) ≥ the set threshold (threshold), then proceed to S6; if the aging rate (OL rate )<set threshold value (threshold), then proceed to S8,

[0160] The set threshold is the value of the performance aging warning line of the medical accelerator equipment. When the budget aging rate is lower than the set threshold, it means that the equipment performance is seriously affected. Otherwise, it means that the equipment performance is sufficient to meet the use requirements.

[0161] S6, analyzing the aging rate (OLrate), the number of fault repairs (Repairs) and the failure rate (FR) of each part of the object, predicting the failure of each component of the medical accelerator after the patient's medical plan is adopted and generating a recommended plan for the failure;

[0162] S7, interacting with users through the interface with fault prediction results and suggested solutions, providing prediction report generation, suggested solution query and prediction reminder functions;

[0163] S8. Save the fault prediction results and generate historical records.

[0164] The rationality of the calculation formula of the frequency value F of the fault part of the early warning method lies in that it comprehensively considers multiple key parameters of various parts of the medical accelerator. Among them, the average use time T0, the longest idle time I0, the average maintenance time M0 and the regular detection cycle D0 are all important factors affecting the fault frequency. By multiplying different weight coefficients a, b, c, d, e, the influence of each factor on the fault frequency can be adjusted according to the actual situation. For example, if the average maintenance time M0 has a greater impact on the fault frequency in the actual use of a certain medical accelerator, then the value of c can be appropriately increased so that the formula can more accurately reflect the fault frequency of the device. The method adopts a multi-factor fusion analysis method in multiple links, such as fault frequency calculation, equipment aging rate calculation, fault prediction, etc. Compared with the traditional single factor analysis method, the early warning method of the present invention can more comprehensively and accurately reflect the operating status, fault risk and aging of the medical accelerator. The large language model has powerful semantic understanding and data processing capabilities, and the method adopts a method based on the large language model, which can better process and analyze complex medical plan information, equipment parameters and other data. Compared with traditional fault warning methods, the present invention can mine deeper information and improve the accuracy and reliability of fault warning. It is also equipped with historical fault data, which can compare the latest predictions with similar historical situations to show trend changes, and enable medical staff or maintenance personnel to analyze the performance of medical accelerators and control their overall health.

[0165] In one embodiment, after the above S3 calculates the fault frequency value F of the part, the fuzzy membership function is calculated for the part.

[0166] The calculation formula of the fuzzy membership function is:

[0167] High frequency fault membership μHF(f) ;

[0168] Medium frequency fault membership μMF(f) ;

[0169] Low-frequency fault membership μLF(f)=1-μHF(f)-μMF(f);

[0170] Among them, j and k are preset parameters.

[0171] j indicates that when the number of failures of a part exceeds the threshold within a set period, the part enters a high-risk state;

[0172] k represents the speed at which the fault of a part changes from normal to abnormal.

[0173] When μHF(f)>μMF(f) and μHF(f)>μLF(f), the part will be classified as a high-frequency fault part and a label will be generated: high-frequency fault part [membership μHF(f)],

[0174] When μMF(f) > μHF(f) and μMF(f) > μLF(f), then this part will be classified as a medium - frequency fault part, and a label will be generated: Medium - frequency fault part [membership degree μMF(f)].

[0175] When μLF(f) > μHF(f) and μLF(f) > μMF(f), then this part will be classified as a low - frequency fault part, and a label will be generated: Low - frequency fault part [membership degree μLF(f)];

[0176] And they are ranked according to the magnitude of the membership degree values.

[0177] This method calculates the fuzzy membership function after calculating the fault frequency value F of the part, and comprehensively considers the preset parameters j (the part enters the high - risk state after the number of faults in the set period exceeds the threshold) and k (the speed of the part's fault transformation from normal to abnormal) and other factors to classify the fault part. This method is more comprehensive and accurate than simply classifying according to the fault frequency value. For example, although the fault frequency value of some parts is not the highest, the speed of the fault transformation from normal to abnormal is very fast or the number of faults in the set period is close to the threshold and is about to enter the high - risk state. Through the calculation of the fuzzy membership function, it can be more accurately classified as a high - frequency fault part. Ranking according to the magnitude of the membership degree values calculated by the fuzzy membership function can intuitively reflect the relative risk levels of each fault part in different fault frequency ranges (high - frequency, medium - frequency, low - frequency). This helps to take different - priority measures for fault parts of different levels when maintaining and managing medical accelerators, improving the rationality and effectiveness of resource allocation.

[0178] In one embodiment, the above - mentioned S6 analyzes the historical maintenance records and equipment parameters of the medical accelerator as follows:

[0179] S61. Calculate the number of fault repairs per unit usage frequency r1 = Repairs / Freq,

[0180] and the number of fault repairs per unit treatment dose r2 = Repairs / Load,

[0181] If r1 > z1 and r2 < z2, then this maintenance plan is judged as an accidental maintenance plan, where z1 and z2 are preset values. If the conditions are not met, proceed to the next judgment;

[0182] S62. Plot the data of the collected fault incidence rate (FR) and treatment dose (Load) as a scatter plot, with the treatment dose (Load) as the vertical axis and the fault incidence rate (FR) as the vertical axis, mark the points corresponding to each set of data on the graph, and analyze by observing the distribution of the points through the scatter plot.

[0183] If the distribution of points is in the shape of a straight line with an obvious linear relationship, it is judged as a specific maintenance plan. If the distribution of points is scattered and has no obvious linear relationship, it proceeds to the next step of judgment.

[0184] S63. Draw the collected data of failure rate (FR) and frequency of use (Freq) into a scatter plot, with frequency of use (Freq) and failure rate (FR) as the vertical axis, mark the points corresponding to each set of data on the graph, and analyze the distribution of the points through the scatter plot.

[0185] If the distribution of points is in the shape of a straight line with an obvious linear relationship, it is judged as a specific maintenance plan. If the distribution of points is scattered and does not have an obvious linear relationship, it is judged as an occasional maintenance plan. Calculate the number of fault repairs per unit frequency of use, that is, Repairs / Freq. If this ratio is very small, such as Repairs / Freq<0.001, it means that the number of fault repairs is very small during the use of a large number of equipment. Calculate the number of fault repairs per unit treatment dose, that is, Repairs / Load. If Repairs / Load<0.01, it means that fault repairs rarely occur when many treatment tasks are performed. This indicates that the fault may be a random, unpredictable, low-probability event, and the corresponding maintenance plan is therefore an occasional maintenance plan. For equipment such as medical accelerators, frequency of use and treatment dose are important indicators that reflect the operating status of the equipment. The advantages of these two formulas are that they are simple, intuitive and easy to calculate. Medical personnel or maintenance personnel can easily obtain data such as the number of fault repairs, frequency of use and treatment dose, and then obtain the values ​​of r1 and r2 through simple division operations, and then make a preliminary judgment on the type of maintenance plan based on the preset values. This simplicity enables rapid analysis of maintenance plans in practical applications, improving the efficiency of equipment maintenance management.

[0186] In one embodiment, the threshold calculation formula is: ,

[0187] in, represents the average value of historical aging rate data, represents the standard deviation of historical aging rate data, is the coefficient, The value range is [1,2].

[0188] The calculation formula for setting the threshold in the present invention comprehensively considers the average value of historical aging rate data. , Standard Deviation and coefficient ( The value range is [1,2]). Average Reflects the central trend of historical aging rate data, standard deviation It reflects the degree of dispersion of the data. The threshold can be set flexibly according to the characteristics of historical data. For example, if the historical aging rate data fluctuates greatly (large standard deviation), the coefficient can be adjusted appropriately. , a more reasonable threshold can be obtained to adapt to the changes in this data. The advantages of this formula are its simplicity and interpretability. The data required for calculation (mean and standard deviation) are relatively easy to obtain, and the formula structure is simple and easy to understand. Medical personnel can reasonably select the coefficient based on their understanding of the historical aging rate data of the equipment. This simple and effective formula can quickly and accurately determine the threshold value, provide a basis for the performance evaluation of medical accelerators, make full use of the historical operation data of the equipment, make the threshold setting more scientific and reasonable, and help improve the accuracy of the performance evaluation of medical accelerators.

[0189] What is disclosed above is only one or more preferred embodiments of the present application, and cannot be used to limit the scope of rights of the present application. Ordinary technicians in this field can understand that all or part of the processes of implementing the above embodiments and equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

Claims

1. A medical accelerator fault warning system based on a large language model, comprising: A data acquisition module, used to obtain the patient's medical plan information, current equipment parameters, historical operating parameters and historical maintenance parameters of the medical accelerator in the hospital information system; A data processing module is used to pre-process the patient's medical plan information and the working parameters of the medical accelerator and remove abnormal data points to obtain processed patient-related medical data and medical accelerator parameter data; Large language model prediction module, including: The equipment history maintenance unit is used to analyze the cause of the failure of the medical accelerator through the equipment history maintenance process, mark the medical accelerator as an analysis object, analyze the parts of the analysis object with frequent failures, divide the analysis object into multiple failure areas, divide the parts of the analysis object into high-frequency failure parts, medium-frequency failure parts and low-frequency failure parts through analysis, generate high-frequency failure marks, medium-frequency failure marks and low-frequency failure marks, and mark them with corresponding failure areas; The maintenance processing collection unit is used to organize the historical maintenance records and equipment parameters of the analysis object, analyze the causes of the failures of the maintenance processing schemes of each fault area of ​​the analysis object, divide the maintenance processing schemes into occasional maintenance schemes and specific maintenance schemes through analysis, generate occasional maintenance marks and specific maintenance marks, and collect the corresponding maintenance schemes; The fault prediction unit is used to bring the obtained patient medical data and medical accelerator parameter data into the aging rate calculation formula, and predict the aging of the medical accelerator after the patient medical plan is adopted according to the calculation result of the aging formula, and output the aging rate result. Its aging rate (OL rate ) is calculated as: ； ； ； ； ； Among them, Freq is the frequency of use of the medical accelerator, Time is the total use time of the medical accelerator, DoseType is the efficiency factor after irradiation dose classification, Cone is the standardized dose size influencing factor, Load is the treatment dose performed within the set time, Repairs is the number of fault repairs of the medical accelerator, and FR is the fault occurrence rate of the medical accelerator within the set time. , ,...... are the corresponding weight coefficients, α, β, γ, δ, and η are the corresponding nonlinear function parameters, Thresh1, Thresh2, and Thresh3 are the thresholds of the variables, respectively. X>Thresh is a binary function. When X is greater than Thresh, the value is 1, otherwise it is 0. The fault analysis unit is used to calculate the aging rate (OL) of each part of the analysis object. rate ), the number of fault repairs (Repairs) and the failure rate (FR), predicting the failure of each component of the medical accelerator after the patient's medical plan is adopted and generating a recommended solution for the failure; The user interaction module is used to interact with users through the interface of fault prediction results and recommended solutions, and provides prediction report generation, recommended solution query and prediction reminder functions.

2. A medical accelerator fault warning system based on a large language model as claimed in claim 1, characterized in that: The data processing module constructs a database of radiation dose and efficacy correspondence rules, sets efficacy ranges for each type of dose, uses decision trees, support vector machines (SVMs) or neural network algorithms, uses radiation doses in medical plans as input features, matches efficacy factors classified by known doses as output labels, and then assigns corresponding efficacy factors according to categories; The data processing module analyzes the medical plan text, searches for keywords, accurately extracts the radiation dose value, and calculates the minimum value C of the radiation dose in the historical data. min and the maximum value C max , for the new irradiation dose C new After normalization, the standardized dose size influence factor is obtained. Calculation formula: .

3. A medical accelerator fault warning system based on a large language model as claimed in claim 1, characterized in that: The data processing module uses the box plot method to identify and eliminate abnormal values ​​of medical accelerator parameters, calculates their quartiles Q1 and Q3, and determines the upper and lower limits of the data points. The calculation formula is as follows: Lower limit = , upper limit= , Where IQR= .

4. A medical accelerator fault warning method based on a large language model, using a medical accelerator fault warning system based on a large language model as claimed in any one of claims 1 to 3, characterized in that: include: S1. Obtaining the patient's medical plan information, and the current equipment parameters and historical maintenance parameters of the medical accelerator in the hospital information system; S2. Preprocess the acquired data. Clean and standardize the patient's medical plan information to obtain a unified data format and parameters; For the current equipment parameters and historical maintenance parameters of the medical accelerator, the box plot method is used to identify and eliminate outliers, calculate their quartiles Q1 and Q3, and determine the upper and lower limits of the data points. The calculation formula is as follows: Lower limit = , upper limit= , Where IQR= ; S3, calculating the fault frequency value of each fault area of ​​the medical accelerator, and classifying the fault areas by analyzing the fault conditions: high-frequency fault area, medium-frequency fault area and low-frequency fault area, and ranking each fault area according to the fault frequency value and generating corresponding marks, The calculation formula of the frequency value F of the fault part is: ； Among them, T0 is the average use time of each part of the medical accelerator, N is the number of times each part of the medical accelerator is repaired for a fault, I0 is the maximum idle time allowed under normal maintenance of each part of the medical accelerator, M0 is the average repair time of each part of the medical accelerator (the average repair time for this type of fault), D0 is the regular inspection cycle of the medical accelerator, and a, b, c, d, and e are the corresponding weight coefficients respectively; S4. Sorting out the historical maintenance records and equipment parameters of the medical accelerator, analyzing the causes of the failures of the historical maintenance solutions for each fault area of ​​the medical accelerator, dividing the maintenance solutions into occasional maintenance solutions and specific maintenance solutions through analysis, generating occasional maintenance marks and specific maintenance marks, and grouping the corresponding maintenance solutions; S5. The high-frequency fault parts, medium-frequency fault parts and low-frequency fault parts classified in S3 are arranged according to their levels and the equipment aging rate (OL rate ), the data pre-processed by S2 is used to enter the aging rate (OL rate ) calculation formula, Its aging rate (OL rate ) is calculated as: ； ； ； ； ； Among them, Freq is the frequency of use of the medical accelerator, Time is the total use time of the medical accelerator, DoseType is the efficiency factor after irradiation dose classification, Cone is the standardized dose size influencing factor, Load is the treatment dose performed within the set time, Repairs is the number of fault repairs of the medical accelerator, and FR is the fault occurrence rate of the medical accelerator within the set time. , ,...... are the corresponding weight coefficients, α, β, γ, δ, and η are the corresponding nonlinear function parameters, Thresh1, Thresh2, and Thresh3 are the thresholds of the variables, respectively. X>Thresh is a binary function. When X is greater than Thresh, the value is 1, otherwise it is 0. The aging rate of each part of the equipment is obtained through the budget. If the aging rate (OL rate ) ≥ the set threshold (threshold), then proceed to S6; if the aging rate (OL rate )<set threshold value (threshold), then proceed to S8, The set threshold is the value of the performance aging warning line of the medical accelerator equipment. When the budget aging rate is lower than the set threshold, it means that the equipment performance is seriously affected. Otherwise, it means that the equipment performance is sufficient to meet the use requirements. S6, analyzing the aging rate (OLrate), the number of fault repairs (Repairs) and the failure rate (FR) of each part of the object, predicting the failure of each component of the medical accelerator after the patient's medical plan is adopted and generating a recommended plan for the failure; S7, interacting with users through the interface with fault prediction results and suggested solutions, providing prediction report generation, suggested solution query and prediction reminder functions; S8. Save the fault prediction results and generate historical records.

5. A medical accelerator fault warning method based on a large language model as claimed in claim 4, characterized in that: After the S3 calculates the part failure frequency value F, it calculates the fuzzy membership function for the part again. The calculation formula for its fuzzy membership function is: High frequency fault membership μHF(f) ; Medium frequency fault membership μMF(f) ; The membership degree of low-frequency failure μLF(f) = 1 - μHF(f) - μMF(f); Among them, j and k are preset parameters. j indicates that when the number of failures of the part exceeds the threshold within the set period, the part enters a high-risk state. k indicates the speed at which the part failure changes from normal to abnormal. When μHF(f) > μMF(f) and μHF(f) > μLF(f), then this part will be classified as a high-frequency failure part and generate a label: high-frequency failure part [membership degree μHF(f)]. When μMF(f) > μHF(f) and μMF(f) > μLF(f), then this part will be classified as a medium-frequency failure part and generate a label: medium-frequency failure part [membership degree μMF(f)]. When μLF(f) > μHF(f) and μLF(f) > μMF(f), then this part will be classified as a low-frequency failure part and generate a label: low-frequency failure part [membership degree μLF(f)]. And each part is ranked according to the magnitude of the membership degree value.

6. A medical accelerator fault warning method based on a large language model as claimed in claim 4, characterized in that: The S6 analyzes the historical maintenance records and equipment parameters of the medical accelerator, and the steps are as follows: S61. Calculate the number of fault repairs per unit usage frequency r1 = Repairs / Freq, and the number of fault repairs per unit treatment dose r2 = Repairs / Load. If r1 > z1 and r2 < z2, then it is judged that this maintenance plan is an occasional maintenance plan, where z1 and z2 are preset values. If the conditions are not met, go to the next judgment. S62. Plot the data of the collected failure rate (FR) and treatment dose (Load) as a scatter plot, with the treatment dose (Load) and the failure rate (FR) as the vertical axis, and mark the points corresponding to each group of data on the graph. Analyze by observing the distribution of the points through the scatter plot. If the distribution of the points presents a linear shape with an obvious linear relationship, it is judged as a specific maintenance plan. If the distribution of the points is scattered and does not have an obvious linear relationship, go to the next judgment. S63. Plot the data of the collected failure rate (FR) and usage frequency (Freq) as a scatter plot, with the usage frequency (Freq) and the failure rate (FR) as the vertical axis, and mark the points corresponding to each group of data on the graph. Analyze by observing the distribution of the points through the scatter plot. If the distribution of the points presents a linear shape with an obvious linear relationship, it is judged as a specific maintenance plan. If the distribution of the points is scattered and does not have an obvious linear relationship, it is judged as an occasional maintenance plan.

7. A medical accelerator fault warning method based on a large language model as claimed in claim 4, characterized in that: The threshold calculation formula is: , in, represents the average value of historical aging rate data, represents the standard deviation of historical aging rate data, is the coefficient, The value range is [1,2].

Images