Remote pelvic inflammation treatment system combined with artificial intelligence diagnosis and medium

Abstract

The invention relates to a remote pelvic inflammation treatment system combined with artificial intelligence diagnosis and a medium, and relates to the technical field of medical treatment, the system comprises at least one memory storing a computer program, and at least one processor is configured to implement a remote pelvic inflammation treatment method when executing the computer program, the method comprises the following specific steps: obtaining quantitative indexes of treatment effects of similar groups referred this time in different dimensions of treatment schemes after using remote pelvic inflammation treatment equipment, and determining different dimensions of treatment scheme suggestions according to preset formulating basis of different dimensions of treatment scheme suggestions, the dimensions of the treatment scheme suggestions comprise drug treatment suggestions, physical treatment suggestions and lifestyle adjustment suggestions; and according to the associated remote pelvic inflammation treatment equipment parameters suggested by the treatment schemes of different dimensions, adjusting the remote pelvic inflammation treatment equipment parameters. The pelvic inflammation diagnosis and treatment method has the advantages that remote pelvic inflammation diagnosis and treatment are achieved in combination with artificial intelligence, and the method is accurate, efficient and high in individuation.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to a remote pelvic inflammatory disease treatment system and medium combined with artificial intelligence diagnosis. Background Art

[0002] Pelvic inflammatory disease is a common gynecological disease that brings many discomforts and health risks to patients.

[0003] At present, the treatment of pelvic inflammatory disease mainly relies on traditional medical equipment and methods. Pelvic inflammatory disease treatment equipment includes ultrasonic therapy devices, infrared therapy devices, etc. Doctors judge the condition through the patient's symptom description, gynecological examination and some routine physiological parameter tests, and use drug therapy, physical therapy and other methods for treatment.

[0004] There are some drawbacks to traditional treatments for pelvic inflammatory disease. First, traditional medical equipment and methods are insufficient in terms of diagnostic accuracy and personalized treatment. Doctors mainly rely on experience to make diagnoses, which is prone to misdiagnosis and missed diagnoses. Second, traditional treatments lack effective integration and analysis of patients' treatment history records, and cannot provide precise treatment based on the patient's specific situation. Summary of the invention

[0005] In order to combine artificial intelligence to achieve remote pelvic inflammatory disease diagnosis and treatment, which is accurate, efficient and highly personalized, the present application provides a remote pelvic inflammatory disease treatment system and medium combined with artificial intelligence diagnosis.

[0006] In the first aspect, the present application provides a remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis, which adopts the following technical solutions: A remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis, comprising: Remote pelvic inflammatory disease treatment equipment; at least one processor; At least one memory storing a computer program, at least one processor configured to implement a remote pelvic inflammatory disease treatment method when executing the computer program, the specific steps are as follows: Obtaining the patient's pelvic images, physiological parameters, and textual information describing the patient's symptoms, wherein the physiological parameters include body temperature, heart rate, and blood pressure; According to the mapping relationship between the preset categories of comprehensive information on pelvic inflammatory disease of patients and the feature vector construction scheme, the feature vector construction schemes matching the comprehensive information on pelvic inflammatory disease of different categories are analyzed and determined, and the feature vector construction schemes for the comprehensive information on pelvic inflammatory disease of different categories of patients are executed to construct feature vectors for obtaining the comprehensive information on pelvic inflammatory disease of different categories of patients, wherein the preset categories of comprehensive information on pelvic inflammatory disease of patients include the patient's pelvic images, physiological parameters and text information describing the patient's symptoms; The feature vectors of the comprehensive information of pelvic inflammatory disease of different categories of patients are constructed and spliced ​​together to form a feature vector of the comprehensive information of pelvic inflammatory disease of the patient; Input the feature vector of the patient's pelvic inflammatory disease comprehensive information into a pre-trained diagnostic suggestion analysis model, and output a diagnostic suggestion, which includes the diagnosis result of pelvic inflammatory disease and the cause analysis; Obtain the patient's treatment history records, integrate the patient's pelvic inflammatory disease comprehensive information and treatment history data, perform data preprocessing and extract key features, and input the key features of this patient into the pre-built patient feature model, calculate the existing patient group whose similarity with this patient reaches the preset similarity threshold, and use it as the similar group for reference; Obtain the quantitative indicators of the treatment effects of the remote pelvic inflammatory disease treatment equipment in different dimensions for the similar groups of this reference, and determine the treatment plan recommendations in different dimensions based on the preset basis for formulating the treatment plan recommendations in different dimensions. The dimensions of the treatment plan recommendations include drug treatment recommendations, physical therapy recommendations, and lifestyle adjustment recommendations; Adjust the remote pelvic inflammatory disease treatment device parameters according to the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan recommendations in different dimensions.

[0007] By adopting the above technical solutions, the feature vector is constructed by integrating pelvic images, physiological parameters and symptom descriptions, and diagnostic suggestions are accurately output. By calculating similar groups and referring to their treatment effects, multi-dimensional treatment plan recommendations are determined, and then the equipment parameters are adjusted. This not only improves the accuracy of diagnosis and the targeting of treatment, but also realizes personalized treatment according to the individual conditions of patients, providing more efficient and accurate remote treatment services for patients with pelvic inflammatory disease, improving patients' treatment experience and rehabilitation effects, and is expected to play an important role in the field of remote treatment of pelvic inflammatory disease.

[0008] Optionally, the feature vectors of the comprehensive information of pelvic inflammatory disease of different categories of patients are constructed and concatenated together to form a feature vector of the comprehensive information of pelvic inflammatory disease of the patient, including: According to the preset weights of the comprehensive information of pelvic inflammatory disease of different categories of patients, the constructed and obtained feature vectors of the comprehensive information of pelvic inflammatory disease of different categories of patients are spliced ​​to form a feature vector of the comprehensive information of pelvic inflammatory disease of the patients.

[0009] By adopting the above technical solution, the feature vectors of different categories of information can be spliced ​​by preset weights to more accurately reflect the comprehensive situation of the patient's pelvic inflammatory disease. The weight setting of different information can highlight key information, such as the importance of images in judging the condition. The comprehensive feature vector formed in this way helps to improve the accuracy and pertinence of diagnosis, and provide a strong basis for the subsequent formulation of more accurate treatment plans.

[0010] Optionally, the weights of comprehensive information on pelvic inflammatory disease in different categories of patients are obtained by: Collect the comprehensive information of pelvic inflammatory disease from different patients, and use the preset correlation coefficient analysis method to calculate the correlation coefficient between the comprehensive information of pelvic inflammatory disease of different categories and the diagnosis results of pelvic inflammatory disease, and accumulate the sum of the correlation coefficients; The ratio of the correlation coefficient of different categories of pelvic inflammatory disease comprehensive information to the sum of the correlation coefficients is used as the weight of the corresponding category of pelvic inflammatory disease comprehensive information.

[0011] By adopting the above technical solution and determining the weight through correlation coefficient analysis, the degree of correlation between various types of information and the diagnosis of pelvic inflammatory disease can be objectively reflected. The proportion of important information in the comprehensive feature vector is more reasonable, and the accuracy of diagnosis is improved. It helps to provide a more accurate basis for subsequent treatment plans and make the parameter adjustment of remote pelvic inflammatory disease treatment equipment more scientific. It can better adapt to the conditions of different patients, improve the treatment effect, and provide reliable method support for remote treatment of pelvic inflammatory disease.

[0012] Optionally, obtaining the preset similarity threshold includes: The weights for obtaining comprehensive information on pelvic inflammatory disease in different categories of patients; According to the correspondence between the weight combination of the comprehensive information of pelvic inflammatory disease in different categories of patients and the preset threshold of similarity, the preset threshold of similarity is analyzed and determined.

[0013] By adopting the above technical solution and determining the threshold according to the weight of different categories of information, we can more accurately find groups similar to the current patient. This makes the reference value of similar groups higher and provides a more accurate basis for treatment plans. The weight reflects the importance of information. Combining the weight to determine the threshold can ensure the similarity of similar patients in key information and improve the targeted treatment.

[0014] Optionally, if the dimension of the treatment plan suggestion is a drug treatment suggestion, according to the preset basis for formulating treatment plan suggestions of different dimensions, the treatment plan suggestions of different dimensions are determined to include: To count the various drugs used by the similar groups in this reference during the use of remote pelvic inflammatory disease treatment equipment, and to determine the frequency of use of each drug in the similar groups, as well as the symptom relief rate and side effect incidence rate of each drug in the similar groups; The comprehensive score of each drug is calculated based on the symptom relief effect and side effect incidence of each drug in similar groups and the preset comprehensive score formula; The drug with the highest comprehensive score is recommended as the selected drug, and the dosage and treatment course statistics of patients in the similar group who use the selected drug are conducted. The average dosage of the similar group is used as the patient's dosage this time, and the average treatment course of the similar group is used as the patient's treatment course this time; The drug with the highest comprehensive score, the patient's current medication dosage, and the patient's current treatment course are summarized as a drug treatment recommendation.

[0015] By adopting the above technical solutions and analyzing the drug usage of similar groups, we can scientifically determine the drug with the highest comprehensive score and the reasonable dosage and course of treatment. This improves the accuracy and pertinence of drug selection and reduces the risk of side effects. Based on similar group data, we can provide patients with personalized treatment plans to improve treatment effects.

[0016] Optionally, if the dimension of the treatment plan suggestion is a physical therapy suggestion, according to the preset basis for formulating treatment plan suggestions of different dimensions, the treatment plan suggestions of different dimensions are determined to include: Obtain detailed data on similar groups after using various physical therapy methods, including pain relief and improvement rate of inflammatory indicators, including C-reactive protein and white blood cell count; The preset physical therapy scoring formula was used to calculate the comprehensive scores of different physical therapies. The physical therapy scoring formula was as follows: comprehensive score = pain relief rate × 0.4 + C-reactive protein improvement rate × 0.3 + white blood cell count improvement rate × 0.3; According to the comprehensive scoring results, the physical therapy with the highest comprehensive score was selected as the physical therapy plan selected this time; Statistics of physical therapy parameters for patients in similar groups who used the selected physical therapy regimen; The average physical therapy parameters of a similar group of references are used as the optimal physical parameters for using the selected physical therapy plan, and the optimal physical parameters and the selected physical therapy plan are used as the physical therapy recommendations for this time.

[0017] By adopting the above technical solutions, the best physical therapy plan and parameters are determined based on similar group data, which improves the scientificity and pertinence of the treatment. The comprehensive score is calculated by quantitative indicators to ensure the effectiveness of the selected plan. The statistical physical therapy parameters provide patients with accurate references and improve the treatment effect.

[0018] Optionally, steps prior to obtaining detailed data on similar groups of people after using various physical therapy methods are included, as follows: Analyze whether the quantitative effect value of a single physical therapy for patients reaches the preset quantitative effect value within the preset time range; If not, obtain detailed data on similar groups after using various physical therapy methods and proceed to the next steps; If yes, obtain detailed data on similar groups using different combinations of physical therapy methods; The preset combined treatment scoring formula was used to calculate the combined treatment scores of different physical therapy methods. The comprehensive score = pain relief rate × 0.4 + C-reactive protein improvement rate × 0.3 + white blood cell count improvement rate × 0.3 + combined treatment effect coefficient × weight; According to the comprehensive scoring results, the combination of physical therapy methods with the highest comprehensive score was selected as the physical therapy plan selected this time; According to the reasonable range of physical therapy parameters, a combination method is used to generate different parameter combinations; The generated parameter combination is input into the established parameter and effect relationship model, the therapeutic effect of the parameter combination is output, and the parameter combination with the best therapeutic effect is selected as the best physical parameters for using the selected physical therapy plan, and the best physical parameters and the selected physical therapy plan are used as the physical therapy recommendations for this time.

[0019] By adopting the above technical solutions, the effect of a single physical therapy is first determined, and data is obtained in a targeted manner and the combined treatment score is calculated to determine the plan. The optimal parameters are selected by using parameter combinations and models, which improves the scientificity and accuracy of the physical therapy plan. It can better adapt to the patient's condition and improve the treatment effect.

[0020] Optionally, referring to average physical therapy parameters of a similar population as optimal physical parameters for use of the selected physical therapy regimen includes: Obtaining the patient's personalized medical information and analyzing whether the patient's personalized medical information meets the preset personalized medical condition information, wherein the personalized medical information includes special physical conditions, differences in disease severity, and differences in treatment goals; If yes, then the adjustment range of the physical therapy parameters is analyzed and determined based on the correspondence between the detailed results contained in the personalized medical information and the adjustment range of the physical therapy parameters, and the optimal physical parameters of the selected physical therapy plan are re-determined based on the average physical therapy parameters of similar groups and the adjustment range of the physical therapy parameters; If not, the average physical therapy parameters of a reference similar group are taken as the optimal physical parameters using the selected physical therapy regimen.

[0021] By adopting the above technical solution, the patient's condition is fully considered and has significant benefits. If the patient has personalized medical information, the parameters can be adjusted specifically to improve the adaptability and effectiveness of the treatment. If there is no personalized information, the average parameters are used to ensure the rationality of the solution. It not only meets the needs of special patients, but also provides a reliable reference for general patients.

[0022] Optionally, according to the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan suggestions of different dimensions, adjusting the remote pelvic inflammatory disease treatment device parameters includes: Analyze whether the patient has pre-set special physical conditions or risk factors; If yes, the parameters determined by the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan suggestions of different dimensions are sent to the terminal held by the relevant person in charge, and the parameters confirmed by the terminal held by the relevant person in charge are used as the parameters of the remote pelvic inflammatory disease treatment device; If not, adjust the remote pelvic inflammatory disease treatment device parameters according to the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan recommendations in different dimensions.

[0023] In a second aspect, the present application provides a computer storage medium, which adopts the following technical solution: A computer storage medium includes a program that can be loaded and executed by a processor to implement a remote pelvic inflammatory disease treatment method in a remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis as described in the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of a remote pelvic inflammatory disease treatment method in a remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis in an embodiment of the present application.

[0025] Figure 2 This is a schematic diagram of the process of obtaining the weights of comprehensive information on pelvic inflammatory disease in different categories of patients according to another embodiment of the present application.

[0026] Figure 3 It is a schematic diagram of the process of obtaining a preset similarity threshold according to another embodiment of the present application.

[0027] Figure 4 In another embodiment of the present application, if the dimension of the treatment plan suggestion is a drug treatment suggestion, a flowchart of determining the treatment plan suggestions in different dimensions is provided according to the preset basis for formulating the treatment plan suggestions in different dimensions. DETAILED DESCRIPTION

[0028] The present application is further described in detail below in conjunction with the accompanying drawings.

[0029] Reference Figure 1 , a remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis disclosed in the present application, including a remote pelvic inflammatory disease treatment device, at least one processor and at least one memory storing a computer program, at least one processor is configured to implement a remote pelvic inflammatory disease treatment method when executing the computer program, and the specific steps are as follows: Step S100, obtaining the patient's pelvic images, physiological parameters and text information describing the patient's symptoms.

[0030] First of all, it should be noted that the remote pelvic inflammatory disease treatment equipment includes a pelvic image acquisition module, a physiological parameter monitoring module, and a symptom description input module. The pelvic image acquisition module can obtain the patient's pelvic image, the physiological parameter monitoring module can monitor the patient's physiological parameters in real time, and the symptom description input module allows the patient to input symptom descriptions.

[0031] Among them, pelvic images: images that reflect the internal structure and condition of the pelvis generated by specific image acquisition technology. For example, ultrasound images can show the shape, size, position, and presence of lesions, such as lumps and effusions, of pelvic organs such as the uterus, ovaries, and fallopian tubes. The acquisition of pelvic images mainly depends on the pelvic image acquisition module contained in the remote pelvic inflammatory disease treatment device. When the patient uses the remote pelvic inflammatory disease treatment device, the pelvic image acquisition module is correctly placed in a specific part of the body (usually the lower abdomen). After the module is started, it emits ultrasonic waves and receives the reflected signals, and generates an ultrasonic image of the pelvis through internal signal processing and image reconstruction algorithms. These image data will be transmitted to the storage unit of the device in real time and uploaded to the cloud server through the remote communication module.

[0032] Physiological parameters include body temperature, heart rate and blood pressure. Body temperature: refers to the temperature of the human body, which is one of the important physiological indicators reflecting the health of the human body. Heart rate: refers to the number of times the heart beats per minute. The heart rate of a normal adult at rest is generally 60-100 beats / minute. Blood pressure: refers to the pressure exerted on the blood vessel wall when blood flows in the blood vessels. It is usually divided into systolic pressure (high pressure) and diastolic pressure (low pressure). The normal blood pressure range for adults is generally 90-139mmHg for systolic pressure and 60-89mmHg for diastolic pressure.

[0033] Body temperature is obtained as follows: The temperature sensor built into the physiological parameter monitoring module can be used to obtain body temperature by contact or non-contact measurement. Heart rate is obtained as follows: The patient places a finger or other body part on the sensor through the photoelectric sensor equipped with the physiological parameter monitoring module. The sensor emits light of a specific wavelength to illuminate the skin, and the heart rate is calculated by detecting the changes in the absorption and reflection of light by the blood. Blood pressure is obtained as follows: The patient ties a cuff to the arm and other parts of the body, and the physiological parameter monitoring module controls the inflation and deflation of the cuff. The pressure sensor configured in the physiological parameter monitoring module detects the pressure changes in the cuff to calculate the blood pressure value.

[0034] The patient's symptom description text information: This is the patient's description of his or her own pelvic inflammatory disease-related symptoms, including but not limited to the nature of lower abdominal pain (such as dull pain, bloating, tingling, etc.), the degree of pain (mild, moderate, severe, etc.), the frequency of pain (occasionally, frequently, continuously, etc.), abnormal leucorrhea (color, odor, texture, etc.), abnormal menstruation (increased or decreased menstrual flow, irregular menstrual cycle, etc.) and other accompanying symptoms (such as fever, fatigue, nausea, etc.).

[0035] The method of obtaining symptom description text information is as follows: patients can describe their symptoms in detail by text input through the user interface of the remote pelvic inflammatory disease treatment system. For example, patients find the symptom description page in the mobile phone application of the system and describe each symptom one by one, such as "persistent dull pain in the lower abdomen, the pain level is moderate. The amount of vaginal discharge increases, the color is yellow, and there is a peculiar smell. The menstrual cycle has been irregular in the past two months, and the menstrual volume has decreased. Occasionally, there is a feeling of fever and fatigue." The system will store and process this text information.

[0036] Step S200, based on the preset mapping relationship between the patient's pelvic inflammatory disease comprehensive information category and the feature vector construction scheme, analyze and determine the feature vector construction scheme that matches the different categories of pelvic inflammatory disease comprehensive information, and execute the feature vector construction scheme of the different categories of patient's pelvic inflammatory disease comprehensive information to construct and obtain the feature vector of the different categories of patient's pelvic inflammatory disease comprehensive information.

[0037] Among them, the preset patient pelvic inflammatory disease comprehensive information category includes the patient's pelvic images, physiological parameters and the patient's symptom description text information.

[0038] Specifically, the feature vector construction scheme for pelvic images is as follows: First, the acquired pelvic images are enhanced to make them more suitable for network input. Then, the enhanced image is input into the pre-trained ResNet model. The ResNet model automatically extracts features from the image and obtains feature maps at different levels. Next, feature fusion technology is used to fuse shallow features with deep features. Feature fusion can be achieved using a simple splicing method or a more complex attention mechanism. Finally, the fused features are converted into a feature vector of fixed length as the feature vector of the pelvic image. For example, the fused feature map is converted into a feature vector of length 512 through global average pooling and a fully connected layer.

[0039] The feature vector construction scheme of physiological parameters is as follows: First, the time series data of the three parameters of body temperature, heart rate and blood pressure are Z-score standardized to have the same scale. Then, the DTW algorithm is used to align the time series of physiological parameters of different patients to eliminate the offset on the time axis. For example, if the lengths of the heart rate time series of two patients are different, the DTW algorithm can be used to find the best alignment path between them so that the heart rate values ​​at the corresponding time points are comparable.

[0040] Next, PCA dimensionality reduction is performed on the aligned physiological parameter data. Assuming that the physiological parameter data after standardization and alignment is a three-dimensional matrix (number of patients × number of time steps × 3, where 3 represents the three parameters of body temperature, heart rate and blood pressure), it can be reduced to a two-dimensional matrix (number of patients × number of principal components) through PCA. Finally, the principal component data after dimensionality reduction are combined into a feature vector as the feature vector of the physiological parameter. For example, if two principal components are selected, the length of the physiological parameter feature vector of each patient is 2.

[0041] The scheme for constructing the feature vector of symptom description text information is as follows: First, input the symptom description text into the pre-trained BERT model. The BERT model will automatically encode the text and obtain the vector representation of each word. Then, an attention mechanism is introduced on the output layer of the BERT model. For example, a multi-head attention mechanism can be used to allow the model to automatically learn the importance weights between different words and highlight the key information related to pelvic inflammatory disease symptoms. Finally, the word vectors weighted by the attention mechanism are average pooled or max pooled to obtain a fixed-length feature vector as the feature vector of the symptom description text. For example, after being processed by BERT and the attention mechanism, a feature vector with a length of 768 is obtained.

[0042] The process of analyzing and determining the feature vector construction scheme that matches different categories of comprehensive information on pelvic inflammatory disease and executing it is as follows: the system can automatically identify the category of information when receiving the patient's comprehensive information on pelvic inflammatory disease based on the preset mapping relationship. For example, when a set of medical imaging data is received, the system determines that this is a pelvic imaging category and executes the corresponding feature vector construction scheme, that is, using image analysis and deep learning algorithms to extract image features and construct image feature vectors. If the received values ​​are body temperature, heart rate, and blood pressure, the feature vector construction scheme for physiological parameters is executed, and the numerical values ​​are normalized and combined into vectors. For symptom description text information, the corresponding natural language processing scheme is also executed to construct a text feature vector.

[0043] Step S300, concatenating the constructed and acquired feature vectors of the comprehensive information of pelvic inflammatory disease of patients of different categories to form a feature vector of the comprehensive information of pelvic inflammatory disease of the patients.

[0044] Specifically, the above operation is based on the fact that different types of information can reflect the patient's pelvic inflammatory disease condition from different angles. Through splicing, various features can be integrated to provide a more comprehensive and accurate basis for subsequent diagnosis and treatment. For example, pelvic images can intuitively display the structure and pathological conditions of the organs in the pelvic cavity; physiological parameters can reflect the patient's overall physiological state; and symptom description texts reflect the patient's subjective feelings and specific symptom manifestations. By splicing these different types of feature vectors together, it is possible to comprehensively consider the patient's objective physical condition and subjective symptoms, and improve the pertinence of diagnosis and treatment.

[0045] The process of forming a feature vector of comprehensive information of pelvic inflammatory disease of a patient is as follows: according to the preset weights of comprehensive information of pelvic inflammatory disease of different categories of patients, the constructed and obtained feature vectors of comprehensive information of pelvic inflammatory disease of different categories of patients are spliced ​​to form a feature vector of comprehensive information of pelvic inflammatory disease of a patient.

[0046] More specifically, the weight is determined first: the importance of each feature vector in the splicing process is determined according to the preset weights of the comprehensive information of pelvic inflammatory disease in different categories of patients. The determination of weights can be based on a variety of factors, such as the contribution of different information to the accuracy of pelvic inflammatory disease diagnosis, the importance analysis of various types of information in historical data, etc. For pelvic imaging feature vectors, since they can directly reflect the pathological conditions in the pelvic cavity, they may be given a relatively high weight. For example, the weight of the pelvic imaging feature vector is set to 0.4. The physiological parameter feature vector reflects the overall physiological state of the patient and has a certain reference value for the diagnosis of pelvic inflammatory disease. Its weight can be set to 0.3. The symptom description text information feature vector reflects the patient's subjective feelings and specific symptoms, and the weight can be set to 0.3.

[0047] Then, linear splicing is performed as follows: according to the determined weights, the feature vectors of the comprehensive information of pelvic inflammatory disease of different categories of patients are linearly spliced. Specifically, for each patient, the pelvic image feature vector, physiological parameter feature vector and symptom description text information feature vector are multiplied by their respective weights and then added together to form a comprehensive feature vector.

[0048] Assume that the pelvic image feature vector of a patient is , the physiological parameter feature vector is (Assuming that the physiological parameter feature vector is three-dimensional, corresponding to the feature values ​​of body temperature, heart rate and blood pressure respectively), the symptom description text information feature vector is Then, the feature vector of the patient's pelvic inflammatory disease comprehensive information is It can be calculated by the following formula: .

[0049] That is, each element of the pelvic image feature vector is multiplied by 0.4, each element of the physiological parameter feature vector is multiplied by 0.3, and each element of the symptom description text information feature vector is multiplied by 0.3. Then, the corresponding elements of the three weighted vectors are added together to obtain the final comprehensive feature vector.

[0050] Step S400, inputting the feature vector of the patient's pelvic inflammatory disease comprehensive information into a pre-trained diagnosis suggestion analysis model, and outputting a diagnosis suggestion, which includes the diagnosis result of pelvic inflammatory disease and the cause analysis.

[0051] Among them, the pre-trained diagnostic recommendation analysis model adopts a model architecture that combines the convolutional neural network (CNN) and the long short-term memory network (LSTM) in the deep neural network. CNN is good at processing spatial features such as images and can extract key structural and lesion features from pelvic images; LSTM is suitable for processing sequence data and can effectively analyze the changes of physiological parameters over time and the semantic sequence in the symptom description text.

[0052] The process of training data collection and preprocessing is as follows: 1. Data collection: Data of pelvic inflammatory disease patients in the past five years were collected from the electronic medical record systems of several large general hospitals and specialized hospitals. This includes detailed pelvic images (such as ultrasound images, CT scan images, etc.), physiological parameter records (continuous monitoring data of body temperature, heart rate, and blood pressure), and symptom description texts accurately filled in by patients. Ensure that the data covers patients with pelvic inflammatory disease of different age groups, different disease severity, and different etiology types to improve the generalization ability of the model. 2. Data preprocessing: For pelvic images, image standardization is performed, including image resizing, grayscale normalization, and other operations, so that the model can better process image data. At the same time, annotation tools are used to annotate the lesion areas in the images to provide supervision information for the model. The physiological parameter data are numerically standardized so that they are distributed in a specific interval, such as mapping the values ​​of body temperature, heart rate, and blood pressure to the [0,1] interval. For continuously monitored physiological parameter data, time series analysis is performed to extract key statistical features, such as mean, standard deviation, maximum, minimum, etc. The symptom description text is processed by natural language. First, the text is cleaned to remove irrelevant information such as punctuation marks and stop words. Then, word embedding technology (such as Word2Vec or GloVe) is used to convert the text into a numerical vector representation so that the model can process text data.

[0053] The model training process is as follows: 1. Model architecture construction: The CNN part consists of multiple convolutional layers, pooling layers, and fully connected layers. The convolutional layer is used to extract local features of pelvic images, the pooling layer is used to reduce the feature dimension, and the fully connected layer maps the extracted features to a vector of fixed length. The LSTM part receives physiological parameter time series data and symptom description text vector sequence. The LSTM unit can memorize long-term time dependencies, so as to better analyze the changing trends of physiological parameters and the semantic information in the symptom description text. The feature vectors output by CNN and LSTM are concatenated, and then further feature fusion and classification are performed through several fully connected layers.

[0054] The training parameters are set as follows: Optimization algorithm selection: The Adam optimizer performs well in practical applications. For example, in image classification tasks, the Adam optimizer can quickly converge to good results. Compared with other optimization algorithms, it can handle large-scale data sets and complex models more effectively. Taking an image classification model based on a convolutional neural network as an example, the Adam optimizer can achieve a high accuracy in a shorter time.

[0055] The learning rate is set as follows: Assume that the initial learning rate is set to 0.001. During the training process, a learning rate decay strategy can be adopted. For example, after a certain number of training rounds, the learning rate is reduced to half of the original. If the model converges quickly at the beginning, but gradually stabilizes as the training progresses, you can consider reducing the learning rate to further optimize the model. Another method is to use an adaptive learning rate algorithm such as Adagrad or Adadelta. Taking natural language processing tasks as an example, when processing text data, different words may have different importance and update frequencies. The adaptive learning rate algorithm can automatically adjust the learning rate based on the gradient history information of each parameter to better adapt to the characteristics of the data.

[0056] The batch size is set as follows: For a medium-sized dataset, the batch size can be set to 32 or 64. If hardware resources permit, you can try a larger batch size, such as 128 or 256, to improve training efficiency. However, too large a batch size may lead to a decrease in the generalization ability of the model. For example, in a medical image analysis task, a smaller batch size allows the model to better learn the differences between different patients and improve the generalization ability of the model. In a large-scale image classification task, a larger batch size can speed up training while also achieving better performance.

[0057] The number of training rounds is set as follows: The number of training rounds is determined based on the size of the dataset and the complexity of the model. Generally speaking, for smaller datasets and simple models, the number of training rounds can be set to dozens to hundreds of rounds; for larger datasets and complex models, thousands of rounds of training may be required. For example, in a handwritten digit recognition task, using a simple neural network model, it may only take dozens of rounds of training to achieve a high accuracy rate. In a large-scale object detection task, using a deep neural network model may require thousands of rounds of training to converge.

[0058] Model evaluation and adjustment are as follows: 1. Data division: It is common practice to divide the data into training set, validation set and test set. For example, 70% of the data can be used as training set, 20% of the data as validation set, and 10% of the data as test set. In a medical diagnosis task, in order to ensure the generalization ability of the model, the data of different patients need to be randomly assigned to different sets. At the same time, it is necessary to ensure that the data in each set is representative and covers different conditions and symptoms. 2. Performance indicator selection: Accuracy is one of the most commonly used performance indicators, but in the case of some unbalanced data sets, other indicators such as recall and F1 value may need to be considered. For example, in a disease detection task, if the incidence of the disease is low, focusing only on accuracy may result in insufficient detection ability of the model for a few categories. Take a binary classification problem as an example, assuming that there are fewer positive samples. If the model only focuses on accuracy, it may predict all samples as negative classes, thereby obtaining a higher accuracy. But in fact, such a model has poor detection ability for positive samples. In this case, recall and F1 value can better reflect the performance of the model. 3. Overfitting treatment: Regularization technology can effectively prevent overfitting. For example, L1 regularization can make the weights of the model more sparse, that is, some weights become zero, thereby reducing the complexity of the model. L2 regularization can make the weights of the model smoother and avoid excessive weight values. Dropout is another commonly used method to prevent overfitting. During the training process, the outputs of some neurons are randomly set to zero, forcing the model to learn more robust features. Taking a deep neural network as an example, using Dropout in each layer can significantly improve the generalization ability of the model. 4. Model adjustment strategy: If the performance of the model on the validation set no longer improves or overfitting occurs, you can try to adjust the architecture or parameters of the model. For example, you can increase the number of layers of the model, increase the number of neurons, adjust the activation function, etc. Taking an image classification task as an example, if the accuracy of the model on the validation set no longer improves, you can try to increase the number of convolutional layers or increase the number of neurons in the fully connected layer to improve the expressiveness of the model. At the same time, you can adjust parameters such as learning rate and batch size to optimize the training process of the model.

[0059] The output diagnostic suggestions are as follows: Pelvic inflammatory disease diagnosis result: The model outputs a binary classification result, indicating whether the patient has pelvic inflammatory disease. For example, the output is "1" for pelvic inflammatory disease, and "0" for no pelvic inflammatory disease. Or it outputs a probability value, indicating the possibility that the patient has pelvic inflammatory disease, such as "the probability of having pelvic inflammatory disease is 75%."

[0060] Cause analysis: Based on the model's analysis of different features, it outputs possible causes. For example, if the model finds a specific pathological pattern in the pelvic images, physiological parameters show elevated inflammation indicators, and the symptom description mentions a recent surgical history, it may output "the cause may be pelvic inflammatory disease caused by postoperative infection."

[0061] Step S500, obtain the patient's treatment history record, integrate the patient's pelvic inflammatory disease comprehensive information and treatment history data, perform data preprocessing and extract key features, and input the key features of this patient into a pre-constructed patient feature model, calculate the existing patient group whose similarity with this patient reaches a preset similarity threshold, and use it as a similar group for reference this time.

[0062] Treatment history records include information such as various treatment methods for pelvic inflammatory disease that patients have received in the past, treatment time, and treatment effects. For example, it may include the name, dosage, and course of treatment of drugs that have been used, physical therapy methods received (such as hot compresses, ultrashort waves, etc.), and the improvement of symptoms after treatment. Treatment history records can be obtained as follows: The patient's treatment history record can be obtained from the hospital's electronic medical record system. If the patient has been treated in different hospitals, the medical records can be integrated through the medical information sharing platform or the medical records provided by the patient himself. For example, when using the remote pelvic inflammatory disease treatment system, patients need to upload their medical records in various hospitals or fill out a treatment history questionnaire. The system extracts relevant information from the hospital system through the data interface, or professionals enter and organize the information provided by the patient.

[0063] The method for integrating the comprehensive information of pelvic inflammatory disease and treatment history data of patients is as follows: the patient's current pelvic image, physiological parameters, symptom description text information and treatment history are integrated to form a comprehensive data set. For example, this information can be stored in a structured data table, where each row represents a patient and each column corresponds to a different information type, such as pelvic image feature vector, physiological parameter vector, symptom description text vector, treatment history record field, etc.

[0064] Data preprocessing involves the following contents, as follows: 1. Data cleaning: remove noise data and outliers. For example, if there are obviously unreasonable values ​​in physiological parameters (such as high or low body temperature, abnormal heart rate, etc.), they can be screened and corrected by comparing with the normal range. For images of poor quality or unrecognizable in pelvic imaging, they can be re-collected or marked as unavailable. 2. Data standardization: Standardize different types of data to have the same scale and distribution. For example, normalize the numerical data in physiological parameters and treatment history records so that their range is between [0,1]; vectorize the symptom description text information to make it consistent in length. 3. Missing value processing: For cases with missing values, interpolation, mean filling and other methods can be used for processing. For example, if the specific time of a treatment is missing in a patient's treatment history record, the missing value can be estimated by interpolating the time before and after the treatment.

[0065] Extracting key features involves the following contents, as follows: 1. Feature selection method: Filtering method: Screening based on the statistical characteristics of the features. For example, the correlation between each feature and the diagnosis result of pelvic inflammatory disease can be calculated, and the features with higher correlation can be selected. For example, by calculating indicators such as the Pearson correlation coefficient or mutual information, physiological parameters and imaging features that are highly correlated with the severity of pelvic inflammatory disease can be screened out. Wrapping method: Use machine learning algorithms for feature selection. For example, recursive feature elimination (RFE) can be used to select the feature subset with the least impact on model performance by gradually deleting features and evaluating model performance. Taking a classification model based on support vector machine (SVM) as an example, all features are first used for training, and then one feature is gradually deleted, the model is retrained and the performance is evaluated until the best feature subset is found. Embedding method: Automatic feature selection during model training. For example, a logistic regression model based on L1 regularization can be used. L1 regularization will make the coefficients of some features become zero, thereby achieving feature selection. During the training process, the model automatically learns which features are more important for the diagnosis and treatment of pelvic inflammatory disease. 2. Key feature examples: Key features extracted from pelvic images may include the size, shape, and texture features of the lesion area. For example, if the image shows fluid accumulation in the pelvis, the volume of the fluid accumulation can be used as a key feature. Key features in physiological parameters may include fluctuations in body temperature, trends in heart rate, and stability of blood pressure. For example, if a patient's body temperature continues to rise during treatment, it may indicate worsening inflammation.

[0066] Key features in the symptom description text may include the degree, frequency, nature of pain, specific manifestations of abnormal vaginal discharge, etc. For example, if the patient describes the pain as severe and stabbing, and the frequency is high, this may be an important feature. Key features in the treatment history record may include effective medications used, the length of treatment, the degree of symptom improvement after treatment, etc. For example, if a certain medication has been effective in the patient's past treatment, then this medication may be a key feature.

[0067] The pre-built patient feature model can use the K-Means clustering algorithm, which is an unsupervised learning algorithm used to divide data points into K clusters so that each data point belongs to the cluster represented by the cluster center closest to it. The specific steps are as follows: 1. Determine the number of clusters K: The appropriate number of clusters can be determined through multiple experiments and evaluations. For example, you can start with a smaller K value, observe the rationality and interpretability of the clustering results, and gradually increase the K value until a better effect is achieved. You can also determine the K value by estimating the number of possible pelvic inflammatory disease types or disease states based on prior knowledge or clinical experience. 2. Initialize the cluster center: Randomly select the feature vectors of K patients as the initial cluster center. These feature vectors can include pelvic imaging features, physiological parameter features, symptom description text features, and treatment history features. 3. Assign data points: For each patient, calculate the distance between its feature vector and each cluster center, and you can use metrics such as Euclidean distance and cosine distance. Assign the patient to the cluster represented by the nearest cluster center. 4. Update cluster center: recalculate the average value of all patient feature vectors in each cluster as the new cluster center. Repeat steps 3 and 4 until the cluster center no longer changes significantly or the preset number of iterations is reached.

[0068] The example is as follows: Assume that there are feature data of 100 patients, and K=3 is determined after multiple experiments. First, the feature vectors of three patients are randomly selected as the initial cluster centers. Then the distance between each patient and the three cluster centers is calculated, and the patient is assigned to the nearest cluster. For example, the feature vector of patient A is closest to cluster center 1, so patient A is assigned to cluster 1. Then the cluster center is updated, and the average value of the feature vectors of all patients in cluster 1 is calculated as the new cluster center 1. Repeat this process until the cluster center is stable. Finally, three clusters are obtained, each cluster represents a different type or condition of pelvic inflammatory disease, such as mild pelvic inflammatory disease, moderate pelvic inflammatory disease, and severe pelvic inflammatory disease.

[0069] Similarity group calculation uses similarity measurement method, which is used to calculate the similarity between two patient feature vectors. If the feature vectors are high-dimensional and have different scales, consider using cosine similarity because it is not affected by the vector scale. If the feature vectors are low-dimensional and have the same scale, use Euclidean distance. Calculate similarity measurement value: For each existing patient and this patient, calculate the similarity measurement value between their feature vectors. For example, if cosine similarity is used, calculate the dot product of the two feature vectors divided by the product of their modulus lengths to get the cosine similarity value. Construct similarity matrix: Store the similarity measurement values ​​between all patients in a matrix to facilitate subsequent similar group calculation. Example: Suppose there are five patients with feature vectors A, B, C, D, and E. Select cosine similarity as the similarity measurement method. Calculate the cosine similarity between patient A and other patients, assuming that the similarity between A and B is 0.7, the similarity between A and C is 0.8, the similarity between A and D is 0.6, and the similarity between A and E is 0.5. Similarly, the similarities between other patients are calculated to obtain a 5×5 similarity matrix.

[0070] A similar group refers to a group of existing patients that have a high degree of similarity to the current patient. By calculating the similarity metric, it is possible to determine which existing patients are most similar to the current patient, and use their group as a reference similar group.

[0071] The steps to confirm similar groups are as follows: 1. Set the preset threshold of similarity: Determine a suitable preset threshold of similarity based on the actual situation. This threshold can be determined by experiment or experience, and is usually between [0,1]. For example, the threshold can be set to 0.8, indicating that only when the similarity metric of two patients is greater than 0.8, they are considered to belong to a similar group. 2. Compare similarity metrics: Compare the feature vector of this patient with the feature vector of each existing patient to calculate the similarity metric. If the similarity metric is greater than or equal to the preset threshold of similarity, the existing patient is added to the similar group. 3. Determine similar groups: All existing patients who meet the similarity conditions are formed into similar groups. This group can be used as a similar group for reference this time and for subsequent treatment plan recommendations. Example: Assume that the feature vector of this patient is F, and the feature vectors of existing patients are A, B, C, D, and E. Set the preset threshold of similarity to 0.8. Calculate the cosine similarity between F and A, B, C, D, and E. Assume that the similarity between F and B is 0.85, the similarity between F and C is 0.9, the similarity between F and D is 0.75, and the similarity between F and E is 0.7. Then the similar group consists of patients B and C, because their similarity with this patient is greater than or equal to 0.8.

[0072] Step S600, obtain the quantitative indicators of the treatment effect of the similar groups of this reference after using the remote pelvic inflammatory disease treatment equipment in different dimensions of treatment plans, and determine the treatment plan recommendations in different dimensions based on the preset basis for formulating the treatment plan recommendations in different dimensions. The dimensions of the treatment plan recommendations include drug treatment recommendations, physical therapy recommendations, and lifestyle adjustment recommendations.

[0073] Treatment plans in different dimensions: including drug treatment suggestions, physical therapy suggestions, lifestyle adjustment suggestions, etc. Drug treatment may involve different types of antibiotics, anti-inflammatory drugs, etc.; physical therapy may include hot compresses, ultrashort wave therapy, etc.; lifestyle adjustments include diet adjustments, exercise suggestions, regular work and rest, etc.

[0074] Quantitative indicators of treatment effect: specific numerical indicators used to measure the effectiveness of treatment plans. For example, for drug treatment, it can be the time of symptom relief, the degree of decrease in inflammatory indicators, etc.; for physical therapy, it can be the degree of pain relief, the amount of pelvic fluid reduction, etc.; for lifestyle adjustments, it can be patient compliance indicators, the degree of improvement in physical immunity, etc.

[0075] The process of obtaining the quantitative indicators of the treatment effect of the similar groups of this reference in different dimensions of the treatment plan after using the remote pelvic inflammatory disease treatment equipment is as follows: First, the treatment records of similar group patients are extracted from the database of the remote pelvic inflammatory disease treatment equipment. These records include information such as the time the patient used the treatment equipment, the treatment parameter settings, the changes in physiological parameters during the treatment process, and the improvement of symptoms. Then, the treatment records are analyzed to calculate the quantitative indicators of the treatment effect of treatment plans in different dimensions. For example, for drug treatment, the average symptom relief time and the decrease in inflammatory indicators of similar groups of patients after using specific drugs can be counted; for physical therapy, the volume of pelvic effusion reduced and the proportion of lesion area reduced can be calculated by comparing the pelvic images before and after treatment; for lifestyle adjustments, the patient's compliance data and changes in immunity indicators can be obtained through questionnaires or monitoring equipment.

[0076] If the dimension of treatment plan recommendation is lifestyle adjustment recommendation, then the lifestyle adjustment recommendation is obtained as follows: Obtain the lifestyle adjustment data of the similar groups used in this reference after using the remote pelvic inflammatory disease treatment equipment, including detailed records of dietary changes, exercise conditions, sleep improvement, etc. and corresponding quantitative indicators of treatment effects. For example, statistics on the improvement of inflammatory indicators after increasing the intake of vitamin-rich foods in similar groups, the degree of pain relief after adhering to specific types of exercise, and the improvement of overall physical condition after adjusting sleep.

[0077] Use statistical methods to establish a data analysis model between quantitative indicators and lifestyle adjustment measures. Here, you can consider using a multiple linear regression model and fit the statistical model using the collected data. Estimate the regression coefficients and determine the parameters of the model using methods such as the least squares method. To verify the model, you can use methods such as cross-validation and leave-one-out validation to evaluate the accuracy and stability of the model. At the same time, check whether the assumptions of the model are met, such as normality, independence, and homoscedasticity of the residuals.

[0078] According to the results of the data analysis model, the contribution of each lifestyle adjustment measure to the treatment effect of pelvic inflammatory disease is calculated. The contribution of each lifestyle adjustment measure can be calculated using the following methods: First, calculate the square of the standardized regression coefficient of each independent variable. Then, divide the square of the standardized regression coefficient of each independent variable by the sum of the squares of the standardized regression coefficients of all independent variables to obtain the proportion of the independent variable that explains the dependent variable, that is, the contribution. The regression coefficient represents the impact of a unit change in an independent variable on the dependent variable when other variables remain unchanged. The standardized regression coefficient is obtained by standardizing the regression coefficient, which eliminates the impact of the variable unit and allows direct comparison between different variables.

[0079] The lifestyle adjustment measures are ranked according to their contribution. For example, if regular aerobic exercise contributes 35%, adequate sleep contributes 30%, increased intake of vitamin-rich foods contributes 20%, and reduced intake of spicy foods contributes 15%, then regular aerobic exercise and adequate sleep can be listed as high-priority adjustment measures, increased intake of vitamin-rich foods as medium priority, and reduced intake of spicy foods as low priority.

[0080] Based on the correspondence between lifestyle adjustment measures and relevant recommendations, specific lifestyle recommendations are formed and ranked.

[0081] High-priority measures: For regular aerobic exercise, develop a specific exercise plan for the patient. Determine the appropriate type, frequency, and duration of exercise based on the patient's physical condition and exercise ability. For example, it is recommended to perform at least three moderate-intensity aerobic exercises a week, such as jogging, swimming, or cycling, for more than 30 minutes each time. At the same time, provide exercise safety precautions, such as warming up before exercise and stretching after exercise to avoid injuries caused by excessive exercise. For adequate sleep, help patients develop good sleeping habits. It is recommended that patients maintain a fixed sleep time every day and create an environment that is conducive to sleep, such as keeping the bedroom quiet, dark, and cool, and avoiding the use of electronic devices before bedtime. If the patient has sleep problems, some relaxation techniques, such as deep breathing, meditation, or a warm bath, can be provided to help them fall asleep.

[0082] Medium priority measures: Provide specific dietary advice to patients on increasing their intake of foods rich in vitamins. Make a list of foods rich in vitamins, such as citrus fruits, green leafy vegetables, nuts, etc., and advise patients to increase their intake of these foods in their daily diet. A weekly meal plan can be developed to ensure that patients can get enough vitamins. At the same time, remind patients to pay attention to the combination and cooking methods of food to maximize the retention of nutrients in food.

[0083] Low priority measures: For reducing spicy food intake, explain to patients the possible effects of spicy foods on PID and advise them to gradually reduce the consumption of spicy foods. Some suggestions for alternative foods, such as bland vegetables, fruits, and whole grains, can be provided. At the same time, patients are encouraged to add other foods that are beneficial to PID in their diet, such as foods rich in protein and antioxidants.

[0084] Step S700, adjusting the remote pelvic inflammatory disease treatment device parameters according to the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan recommendations of different dimensions.

[0085] Treatment plan recommendations in different dimensions: including drug treatment recommendations, physical therapy recommendations, lifestyle adjustment recommendations, etc. These recommendations are made based on the patient's condition and the treatment experience of similar groups, aiming to provide patients with personalized treatment plans. Remote pelvic inflammatory disease treatment equipment parameters: refers to the various parameters that can be adjusted in the remote pelvic inflammatory disease treatment equipment, and the settings of these parameters will directly affect the treatment effect. For example, the power, frequency, time and other parameters of physical therapy equipment, and the dosage, speed and other parameters of drug delivery equipment.

[0086] Remote pelvic inflammatory disease treatment equipment can automatically adjust parameters based on treatment plan recommendations in different dimensions. The device's built-in algorithm can automatically calculate the appropriate parameter settings and adjust them according to the requirements of the treatment plan. For example, the device can automatically adjust physical therapy parameters such as power and frequency based on the patient's condition and treatment history to achieve the best treatment effect.

[0087] Reference Figure 2 , the weights of comprehensive information of pelvic inflammatory disease in different categories of patients are obtained by: Step S310, collecting comprehensive pelvic inflammatory disease information of different patients, and using a preset correlation coefficient analysis method to calculate the correlation coefficients between the comprehensive pelvic inflammatory disease information of different categories and the diagnosis results of pelvic inflammatory disease, and accumulating the sum of the correlation coefficients.

[0088] Pelvic inflammatory disease comprehensive information: including the patient's pelvic images, physiological parameters (body temperature, heart rate, blood pressure) and symptom description text information. Correlation coefficient analysis method: a statistical method used to measure the strength of the linear relationship between two variables. In this step, it is used to determine the degree of correlation between different categories of pelvic inflammatory disease comprehensive information and pelvic inflammatory disease diagnosis results. Pelvic inflammatory disease diagnosis results: the conclusion of whether a patient has pelvic inflammatory disease determined by professional doctors' diagnosis, laboratory tests, etc., usually divided into two results: "suffering from pelvic inflammatory disease" and "not suffering from pelvic inflammatory disease".

[0089] The specific analysis process of the correlation coefficient is as follows: For the pelvic imaging category: First, the collected pelvic images are subjected to feature extraction, for example, the size, shape, texture and other features of the lesion area in the image can be extracted. Then, the correlation coefficient between these features and the diagnosis results of pelvic inflammatory disease is calculated. Methods such as the Pearson correlation coefficient and the Spearman correlation coefficient can be used. Assuming that the Pearson correlation coefficient is used, for each patient's pelvic imaging feature vector and the diagnosis result of pelvic inflammatory disease (represented by 0 and 1), the covariance and standard deviation between them are calculated to obtain the correlation coefficient. If the correlation coefficient between certain features in the pelvic imaging and the diagnosis result of pelvic inflammatory disease is high, it means that these features are strongly associated with the occurrence of pelvic inflammatory disease.

[0090] For the physiological parameter category: Similarly, the physiological parameters (body temperature, heart rate, blood pressure) are first processed, which may include data standardization, outlier processing and other steps. Then, the correlation coefficient between the processed physiological parameter data and the diagnosis result of pelvic inflammatory disease is calculated. For example, when calculating the correlation coefficient between body temperature and the diagnosis result of pelvic inflammatory disease, the body temperature data of all patients can be combined into one vector, and the diagnosis result of pelvic inflammatory disease can be combined into another vector, and the calculation formula of the correlation coefficient can be used. If the body temperature has a significantly different distribution in patients with pelvic inflammatory disease and those without pelvic inflammatory disease, then the correlation coefficient between body temperature and the diagnosis result of pelvic inflammatory disease will be higher.

[0091] For the symptom description text information category: perform natural language processing on the symptom description text and convert the text into a numerical vector representation. You can use the bag-of-words model, TF-IDF algorithm, or deep learning text processing method. Next, calculate the correlation coefficient between the text vector and the pelvic inflammatory disease diagnosis result. For example, by calculating the similarity measure (such as cosine similarity) between the text vector and the pelvic inflammatory disease diagnosis result vector and converting it into the form of a correlation coefficient. If certain symptoms occur frequently in patients with pelvic inflammatory disease and less frequently in patients without pelvic inflammatory disease, then the text vectors corresponding to these symptoms will have a higher correlation coefficient with the pelvic inflammatory disease diagnosis result.

[0092] Sum of correlation coefficients: The sum of the correlation coefficients between the different categories of pelvic inflammatory disease comprehensive information and the pelvic inflammatory disease diagnosis results. This value is used to calculate the weight of each category of information in the subsequent calculation. The calculation process is as follows: After calculating the correlation coefficients of the three categories of pelvic imaging, physiological parameters, and symptom description text information respectively, add them together to get the sum of the correlation coefficients. For example, if the correlation coefficient between pelvic imaging and pelvic inflammatory disease diagnosis results is r1, the correlation coefficient of physiological parameters is r2, and the correlation coefficient of symptom description text information is r3, then the sum of the correlation coefficients is r1+r2+r3.

[0093] Step S320: The ratio of the correlation coefficients of different categories of comprehensive information on pelvic inflammatory disease to the sum of the correlation coefficients is used as the weight of the comprehensive information on pelvic inflammatory disease of the corresponding category.

[0094] For the pelvic imaging category: Assume that the correlation coefficient between the pelvic imaging and the diagnosis result of pelvic inflammatory disease calculated in step S310 is r1, and the sum of the correlation coefficients is s (s=r1+r2+r3, where r2 is the correlation coefficient of the physiological parameters and r3 is the correlation coefficient of the symptom description text information). Then the weight of the pelvic imaging category is w1=r1 / s. For example, if r1=0.6, r2=0.3, r3=0.1, then s=0.6+0.3+0.1=1, and the weight of the pelvic imaging category is w1=0.6 / 1=0.6.

[0095] For physiological parameter categories: Similarly, if the correlation coefficient between physiological parameters and the diagnosis of pelvic inflammatory disease is r2, the sum of the correlation coefficients is s. The weight of the physiological parameter category is w2=r2 / s. For example, in the above example, r2=0.3, then the weight of the physiological parameter category is w2=0.3 / 1=0.3.

[0096] For the symptom description text information category: let the correlation coefficient between the symptom description text information and the diagnosis result of pelvic inflammatory disease be r3, and the sum of the correlation coefficients be s. The weight of the symptom description text information category is w3=r3 / s. In this example, r3=0.1, so the weight of the symptom description text information category is w3=0.1 / 1=0.1.

[0097] Reference Figure 3 , the acquisition of the preset similarity threshold includes: Step S510, obtaining the weights of comprehensive information on pelvic inflammatory disease of patients of different categories.

[0098] Step S520, analyzing and determining the preset similarity threshold according to the corresponding relationship between the weight combination of the comprehensive information of pelvic inflammatory disease of different categories of patients and the preset similarity threshold.

[0099] Weight combination: A combination of weights of comprehensive information of pelvic inflammatory disease in different categories of patients. For example, if there are three categories of pelvic imaging, physiological parameters and symptom description text information, and their weights are w1, w2 and w3 respectively, then the weight combination can be expressed as (w1, w2, w3).

[0100] Similarity preset threshold: It is used to determine the critical value of the similarity between the existing patient group and the current patient. When the similarity measurement value between the existing patient and the current patient reaches or exceeds this threshold, the group to which the existing patient belongs is considered to be the similar group for this reference.

[0101] First, establish the correspondence between different weight combinations and similarity preset thresholds. This correspondence can be determined in the following way: Based on experience and expert knowledge: Medical experts determine the appropriate similarity preset threshold under different weight combinations based on their clinical experience and understanding of pelvic inflammatory disease diagnosis. For example, if the weight of pelvic imaging is higher, it means that imaging information is more important in diagnosis. At this time, a higher similarity preset threshold may need to be set to ensure that the similar groups found have a high similarity in imaging features.

[0102] Data analysis and experiments: Through the analysis and experiments of a large amount of patient data, the treatment effect and diagnostic accuracy of similar groups under different weight combinations are observed to determine the best similarity preset threshold. For example, a portion of patient data can be randomly selected, different weight combinations and similarity preset thresholds can be tried, the characteristics of similar groups under different circumstances and the effectiveness of treatment plans can be compared, and the weight combination and similarity preset threshold that can produce the best results can be selected.

[0103] Then, according to the weight combination of the comprehensive information of pelvic inflammatory disease of different categories of patients currently obtained, the corresponding similarity preset threshold is searched in the established correspondence. Assuming that the current weight combination is (w1, w2, w3), by querying the correspondence table or using the pre-established function model, the corresponding similarity preset threshold is determined to be t. Finally, the determined similarity preset threshold t is applied to subsequent calculations and analysis to determine the similar group for this reference.

[0104] Reference Figure 4 If the dimension of the treatment plan recommendation is a drug treatment recommendation, according to the preset basis for formulating treatment plan recommendations of different dimensions, the treatment plan recommendations of different dimensions include: Step S610, counting various drugs used by the similar group of this reference during the use of the remote pelvic inflammatory disease treatment device, and determining the use frequency of each drug in the similar group and the symptom relief rate and side effect incidence rate of each drug in the similar group.

[0105] Specifically, the statistical method is as follows: extract the treatment records of similar groups of patients from the database of remote pelvic inflammatory disease treatment equipment. These records include information such as the name, dosage, and time of use of the drugs used by the patients during the use of the treatment equipment. The treatment records are sorted and classified, the same drugs are classified into one category, and the number of times each drug is used is counted. For example, if there are 100 patients in a similar group, 50 of whom used drug A, 30 used drug B, and 20 used drug C, then the number of times drug A was used is 50, the number of times drug B was used is 30, and the number of times drug C was used is 20.

[0106] Frequency of use: refers to the ratio of the number of times a drug is used in a similar group to the total number of patients. Frequency of use can reflect the popularity and prevalence of the drug in a similar group. The calculation method is as follows: Frequency of use = number of times a drug is used / total number of patients in a similar group. For example, in the above example, the frequency of use of drug A is 50 / 100=0.5, the frequency of use of drug B is 30 / 100=0.3, and the frequency of use of drug C is 20 / 100=0.2.

[0107] Symptom relief rate: refers to the proportion of patients whose pelvic inflammatory disease symptoms are relieved after using a certain drug. Symptom relief can be measured by patient self-report, doctor's assessment or specific symptom indicators. The symptom relief rate is determined as follows: For each patient who uses a certain drug, record the symptoms of pelvic inflammatory disease before and after using the drug. Symptom rating scales or other assessment methods can be used to quantify the severity of symptoms. Calculate the proportion of patients with symptom relief to the total number of patients using the drug, which is the symptom relief rate of the drug. For example, if 50 patients use drug A, and 40 of them have their symptoms relieved, then the symptom relief rate of drug A is 40 / 50=0.8.

[0108] Side effect incidence: refers to the proportion of patients who experience side effects after using a certain drug. Side effects can include allergic reactions, gastrointestinal discomfort, dizziness and other adverse reactions. The determination method is as follows: For each patient who uses a certain drug, observe whether they experience side effects during the use of the drug, and record the type and severity of the side effects. Calculate the proportion of patients who experience side effects to the total number of patients using the drug, which is the side effect incidence of the drug. For example, if 50 patients use drug A, and 10 of them experience side effects, then the side effect incidence of drug A is 10 / 50=0.2.

[0109] Step S620, calculating the comprehensive score of each drug according to the symptom relief effect and side effect incidence of each drug in similar groups and a preset comprehensive scoring formula.

[0110] Design of comprehensive scoring formula: Considering the different importance of symptom relief effect and side effect incidence, different weights can be assigned to them. For example, the weight of symptom relief effect can be set to w1, the weight of side effect incidence can be set to w2, and w1+w2=1. The comprehensive scoring formula can be expressed as: comprehensive score = w1×symptom relief effect-w2×side effect incidence. Among them, the symptom relief effect and side effect incidence can be expressed by specific values, such as the percentage of symptom relief rate and side effect incidence. The specific calculation process is as follows: For each drug, first determine the specific values ​​of its symptom relief effect and side effect incidence. For example, assume that the symptom relief rate of drug A is 80% and the side effect incidence is 10%. Then, determine the values ​​of w1 and w2 according to the preset weights. Assume w1=0.7, w2=0.3. Substitute the values ​​into the comprehensive scoring formula for calculation: comprehensive score = 0.7×80%-0.3×10%=56%-3%=53%.

[0111] In step S630, the drug with the highest comprehensive score is recommended as the selected drug, and dosage statistics and treatment course statistics are performed on patients in the similar group who use the selected drug. The average dosage of the reference similar group is used as the patient's dosage for this time, and the average treatment course of the reference similar group is used as the patient's treatment course for this time.

[0112] Medication dosage statistics: Statistics on the medication dosage of patients in similar groups who use the selected medication to determine the commonly used dosage range of the medication in similar situations. Treatment course statistics: Statistics on the treatment course of patients in similar groups who use the selected medication to understand how long the medication usually needs to be used to achieve a good therapeutic effect.

[0113] The specific steps are as follows: Determine the drug with the highest comprehensive score: Review the comprehensive score of each drug calculated in step S620, and select the drug with the highest score as the selected drug. For example, if the comprehensive score of drug A is 80 points, the comprehensive score of drug B is 70 points, and the comprehensive score of drug C is 60 points, then drug A is selected as the recommended drug.

[0114] Medication dosage statistics: Screen patients who used selected drugs from the treatment records of similar groups. Record the specific dosage of selected drugs used by these patients. For example, the dosage of drug A may have multiple specifications, such as 500 mg per time, 1000 mg per time, etc.

[0115] The frequency of use of different doses was counted. For example, in a similar group, 60% of patients used a dose of 500 mg each time, and 40% of patients used a dose of 1000 mg each time.

[0116] Treatment course statistics: Similarly, find patients who used the selected drug from the treatment records of similar groups. Record the treatment course of these patients using the selected drug. The treatment course can be expressed in days, weeks, or other time units. Calculate the average treatment course. For example, the treatment courses of patients using drug A in the similar group are 7 days, 10 days, 14 days, etc., and the average treatment course is (7+10+14) ÷ 3 = 10.33 days.

[0117] Step S640, summarizing the drug with the highest comprehensive score, the patient's current medication dosage, and the patient's current treatment course as a medication treatment recommendation.

[0118] If the dimension of the treatment plan recommendation is physical therapy recommendation, according to the preset basis for formulating treatment plan recommendations of different dimensions, the treatment plan recommendations of different dimensions include: Step S6a0, obtaining detailed data of similar groups after using various physical therapy methods, the detailed data including pain relief degree and improvement rate of inflammatory indicators, and the inflammatory indicators include C-reactive protein and white blood cell count.

[0119] Detailed data were obtained by extracting physical therapy records of similar groups of patients from the database of remote pelvic inflammatory disease treatment equipment. These records should include the physical therapy method used by the patient, the time, frequency, intensity and other parameters of the treatment, as well as the pain assessment results and inflammatory index detection data before and after treatment.

[0120] Pain relief: This measures the degree to which a patient's pain symptoms are reduced after receiving physical therapy. This can be determined through patient self-report, physician assessment, or using a specific pain assessment tool. This can be determined as follows: Patient self-report: Before and after physical therapy, ask patients to rate their pain using a numerical rating scale (e.g., 0-10, 0 for no pain, 10 for the most severe pain) or descriptive words (e.g., mild pain, moderate pain, severe pain, etc.). The difference between the pain scores before and after treatment is calculated as an indicator of pain relief. For example, if a patient's pain score before treatment is 8 and after treatment is 4, then the pain relief is 4.

[0121] Improvement rate of inflammatory indicators: By comparing the changes in the patient's inflammatory indicators (C-reactive protein, white blood cell count) before and after physical therapy, the effect of physical therapy on inflammation control is measured. The higher the improvement rate of inflammatory indicators, the better the control effect of physical therapy on inflammation. Calculation method: C-reactive protein improvement rate: First, obtain the patient's C-reactive protein test values ​​before and after physical therapy. Then, calculate the improvement rate, the formula is (C-reactive protein value before treatment - C-reactive protein value after treatment) ÷ C-reactive protein value before treatment × 100%. For example, the patient's C-reactive protein value before treatment is 50mg / L, and after treatment it is 30mg / L, then the C-reactive protein improvement rate is (50-30) ÷ 50 × 100% = 40%.

[0122] White blood cell count improvement rate: Similarly, obtain the patient's white blood cell count test values ​​before and after physical therapy. The formula for calculating the improvement rate is (white blood cell count before treatment - white blood cell count after treatment) ÷ white blood cell count before treatment × 100%. For example, if the patient's white blood cell count before treatment is 12×10 9 / L, after treatment it is 8×10 9 / L, then the improvement rate of white blood cell count is (12-8) ÷ 12 × 100% ≈ 33.3%.

[0123] Step S6b0, using a preset physical therapy scoring formula to calculate and obtain the comprehensive scores of different physical therapies, the physical therapy scoring formula is as follows: comprehensive score = pain relief rate × 0.4 + C-reactive protein improvement rate × 0.3 + white blood cell count improvement rate × 0.3.

[0124] Assume that for a certain physical therapy method, we have determined through step S6a0 that the pain relief rate is 60%, the C-reactive protein improvement rate is 40%, and the white blood cell count improvement rate is 30%. Then, according to the physical therapy scoring formula, calculation is performed: comprehensive score = 60% × 0.4 + 40% × 0.3 + 30% × 0.3 = 0.24 + 0.12 + 0.09 = 0.45 (i.e. 45%).

[0125] Step S6c0, according to the comprehensive scoring result, select the physical therapy with the highest comprehensive score as the physical therapy plan selected this time.

[0126] The specific selection process is as follows: First, review the comprehensive scores of various physical therapy methods calculated in step S6b0. Suppose we consider three physical therapy methods, namely hot compress, ultrashort wave therapy and pelvic floor muscle rehabilitation training, and their comprehensive scores are 0.45, 0.52 and 0.48 respectively. Then, compare the size of these comprehensive scores. In this example, ultrashort wave therapy has the highest comprehensive score of 0.52. Finally, ultrashort wave therapy is determined to be the physical therapy plan selected this time. This means that when formulating treatment recommendations for this patient, ultrashort wave therapy is recommended as a physical therapy method.

[0127] Step S6d0, performing statistics of physical therapy parameters for patients in a similar group who use the selected physical therapy regimen.

[0128] The treatment records of patients in a similar group who used the selected physical therapy program were extracted from the database of the telepelvic inflammatory treatment device. These records should include information such as the specific parameters of the physical therapy and the patient's treatment response.

[0129] The extracted treatment records are sorted and classified, and patients with the same parameter settings are grouped together, and the number of patients in each group is counted. For example, if there are 100 similar groups of patients who use ultrashort wave therapy as a physical therapy program, 50 of them have treatment parameters of 30 minutes per treatment, 3 times a week, and medium intensity, 30 have treatment parameters of 20 minutes per treatment, 4 times a week, and low intensity, and 20 have treatment parameters of 40 minutes per treatment, 2 times a week, and high intensity, then the number of patients in these three groups can be counted separately.

[0130] Step S6e0, using the average physical therapy parameters of a similar group as the optimal physical parameters for the selected physical therapy program, and using the optimal physical parameters and the selected physical therapy program as the physical therapy recommendation for this time.

[0131] Reference to the average physical therapy parameters of similar groups: The average value calculated by statistically analyzing the physical therapy parameters of patients in similar groups who used the selected physical therapy program. These parameters include treatment time, frequency, intensity, etc., which serve as reference values ​​for this patient using the physical therapy program.

[0132] Tele-PID treatment also includes the following steps prior to obtaining detailed data on similar groups of people using various physical therapy approaches: Step S6A0, analyzing whether the effect quantification value of the patient's single physical therapy within the preset time range reaches the preset effect quantification value. If not, execute step S6B0; if yes, execute step S6C0.

[0133] Preset time range: A time period determined based on the general treatment cycle of pelvic inflammatory disease and clinical experience, used to evaluate the effect of a single physical therapy on a patient within that time period. For example, it can be set to two weeks, one month, etc.

[0134] Single physical therapy: refers to the treatment of patients with pelvic inflammatory disease using only one physical therapy method, such as hot compress, ultrashort wave physical therapy, pelvic floor muscle rehabilitation training, etc.

[0135] Effect quantification value: a specific numerical indicator used to measure the effect of physical therapy. It can be quantified based on multiple aspects such as the degree of pain relief and the improvement rate of inflammatory indicators. For example, the degree of pain relief can be quantified using a digital scoring method, and the improvement rate of inflammatory indicators can be calculated by detecting changes in indicators such as C-reactive protein and white blood cell count. Preset effect quantification value: a standard value determined based on clinical experience and treatment goals, used to determine whether the effect of a single physical therapy has met expectations.

[0136] The specific analysis of the effect quantification value is as follows: Based on the collected data, the patient's effect quantification value is calculated. For each evaluation indicator, the calculation is performed according to the predetermined quantification method. For example, if the three indicators of pain relief degree, C-reactive protein improvement rate and white blood cell count improvement rate are selected, the difference in the patient's pain score before and after treatment, and the improvement rates of C-reactive protein and white blood cell count are calculated respectively.

[0137] Compare the effect quantification value with the preset effect quantification value: Compare the calculated effect quantification value with the preset effect quantification value. If the effect quantification value reaches or exceeds the preset effect quantification value, it is considered that the single physical therapy is effective; otherwise, it is considered that the single physical therapy is not effective. For example, if the preset effect quantification value is that the pain score is reduced by more than 3 points, the C-reactive protein improvement rate is more than 30%, and the white blood cell count improvement rate is more than 20%. If the patient's pain score is reduced by 4 points, the C-reactive protein improvement rate is 35%, and the white blood cell count improvement rate is 25%, then the single physical therapy is considered effective, and step S6C0 is executed; if the patient's effect quantification value does not reach the preset effect quantification value, step S6B0 is executed.

[0138] Step S6B0, obtaining detailed data of similar groups after using various physical therapy methods, and continuing to subsequent steps.

[0139] Step S6C0, obtaining detailed data of similar groups after using different combinations of physical therapy methods.

[0140] Combination of different physical therapy methods: Combining two or more physical therapy methods in the hope of achieving better therapeutic effects. For example, hot compress can be combined with ultrashort wave therapy, or pelvic floor muscle rehabilitation training can be combined with magnetic therapy.

[0141] Step S6D0, using the preset combined treatment scoring formula, calculate the combined treatment scores of different physical therapy method combinations, and the comprehensive score = pain relief rate × 0.4 + C-reactive protein improvement rate × 0.3 + white blood cell count improvement rate × 0.3 + combined treatment effect coefficient × weight.

[0142] Combined treatment scoring formula: a mathematical formula used to comprehensively evaluate the effects of different physical therapy combinations. Combined treatment effect coefficient: used to measure the synergistic effect of physical therapy combinations. This coefficient can be determined based on clinical experience, experimental data or statistical analysis. If there is good synergy between the physical therapy combinations, the combined treatment effect coefficient will be higher; conversely, if there is no obvious synergy between the combinations, the combined treatment effect coefficient will be lower.

[0143] Assume that for a certain combination of physical therapy methods, we have determined through the previous steps that the pain relief rate is 70%, the C-reactive protein improvement rate is 50%, the white blood cell count improvement rate is 40%, and the combined treatment effect coefficient is 0.2 (assuming that the combined treatment effect of this combination is better based on clinical experience), and the weight distribution is 0.4 for the pain relief rate, 0.3 for the C-reactive protein improvement rate, 0.3 for the white blood cell count improvement rate, and 0.2 for the combined treatment effect coefficient.

[0144] Calculated according to the combined treatment scoring formula: comprehensive score = 70% × 0.4 + 50% × 0.3 + 40% × 0.3 + 0.2 × 0.2 = 0.28 + 0.15 + 0.12 + 0.04 = 0.59 (i.e. 59%).

[0145] Step S6E0, based on the comprehensive scoring result, select the physical therapy method combination with the highest comprehensive score as the physical therapy plan selected this time.

[0146] Step S6F0, using a combination method to generate different parameter combinations according to a reasonable range of physical therapy parameters.

[0147] Reasonable range: The range of values ​​of physical therapy parameters determined based on clinical experience, medical research and equipment instructions. Within this range, physical therapy methods are safe and effective, and exceeding this range may lead to adverse consequences. Combination method: By combining different physical therapy parameters, a variety of possible parameter combination schemes are generated. These combination schemes can be used for subsequent evaluation and selection to determine the best combination of physical therapy parameters.

[0148] Suppose we have three physical therapy parameters: treatment time (15-30 minutes), treatment frequency (2-4 times a week), and treatment intensity (low, medium, high). Use exhaustive method to generate parameter combinations: treatment time is 15 minutes, treatment frequency is 2 times a week, and treatment intensity is low. Treatment time is 15 minutes, treatment frequency is 2 times a week, and treatment intensity is medium. Treatment time is 15 minutes, treatment frequency is 2 times a week, and treatment intensity is high. ... Treatment time is 30 minutes, treatment frequency is 4 times a week, and treatment intensity is high.

[0149] Step S6G0, input the generated parameter combination into the established parameter and effect relationship model, output the therapeutic effect of the parameter combination, and select the parameter combination with the best therapeutic effect as the best physical parameters for using the selected physical therapy plan, and use the best physical parameters and the selected physical therapy plan as the physical therapy recommendations for this time.

[0150] Parameter-effect relationship model: A model based on data and algorithms that is used to predict the effects of physical therapy under different parameter combinations. This model can be trained and optimized based on historical data and known physical therapy principles. The model will use its internal algorithms and data to predict the physical therapy effects under the input parameter combination. For example, if the model is based on a machine learning algorithm, it will predict the effect of the current parameter combination based on the treatment effects of similar parameter combinations in historical data.

[0151] The treatment effect of the output parameter combination is as follows: The model will output the predicted value of the treatment effect of each parameter combination. This predicted value can be a specific value, such as the predicted value of indicators such as pain relief rate and C-reactive protein improvement rate; it can also be a classification result, such as "good effect", "moderate effect", "poor effect", etc.

[0152] Select the best treatment parameter combination: Compare the treatment effects of all parameter combinations output by the model. The best treatment parameter combination can be determined based on specific indicators or comprehensive scores.

[0153] For example, the comprehensive score of each parameter combination can be calculated, and the comprehensive score can be weighted according to multiple indicators such as pain relief rate, C-reactive protein improvement rate, and white blood cell count improvement rate. The parameter combination with the highest comprehensive score is selected as the parameter combination with the best treatment effect. Alternatively, a threshold can be set, and the parameter combination with a treatment effect prediction value exceeding the threshold is selected as the parameter combination with the best treatment effect.

[0154] The average physical therapy parameters of a similar population were referenced as the optimal physical parameters for the selected physical therapy regimen including: Step S6e1, obtaining the patient's personalized medical information, and analyzing whether the patient's personalized medical information meets the preset personalized medical condition information, wherein the personalized medical information includes special physical conditions, differences in disease severity, and differences in treatment goals. If yes, execute step S6e2; if no, execute step S6e3.

[0155] Special physical conditions: Allergic constitution: Determine the type and severity of allergic substances through allergen testing. For example, it is divided into mild allergy (mild reaction to a common allergen, such as skin itching), moderate allergy (obvious reaction to multiple common allergens, such as rash, breathing difficulties, etc.), and severe allergy (strong reaction to multiple allergens, which may lead to shock). Different degrees of quantitative scores are assigned, such as 1 point for mild allergy, 2 points for moderate allergy, and 3 points for severe allergy. Suffering from other chronic diseases: Quantify according to the severity of the disease and the impact on physical therapy. For example, heart disease is divided into stable stage (heart function level I-II, less impact on physical therapy, quantified as 2 points), unstable stage (heart function level III-IV, greater impact on physical therapy, quantified as 4 points); diabetes is divided into well-controlled (stable blood sugar, less impact on physical therapy, quantified as 1 point), poorly controlled (large blood sugar fluctuations, greater impact on physical therapy, quantified as 3 points), etc. In special physiological periods: such as pregnancy (divided into early, middle and late stages according to the pregnancy stage, quantified as 2 points, 3 points and 4 points respectively), lactation period (quantified as 2 points).

[0156] Differences in severity of illness: Quantification is performed using the pelvic inflammatory disease symptom scoring scale, including pain level (0-10 points), abnormal secretion level (0-5 points), fever (2 points for fever, 0 points for no fever), etc. The scores of each indicator are added together to obtain the total score of the severity of the illness. For example, a total score of 0-5 points is mild, which is quantified as 1 point; 6-10 points is moderate, which is quantified as 2 points; 11 points and above are severe, which is quantified as 3 points.

[0157] Differences in treatment goals: Relieve pain as soon as possible: quantified according to the patient's urgency for pain relief, divided into very urgent (quantified as 3 points), relatively urgent (quantified as 2 points), and generally urgent (quantified as 1 point). Long-term recovery and relapse prevention: quantified according to the patient's acceptance of long-term treatment and the importance of relapse prevention, divided into highly important (quantified as 3 points), relatively important (quantified as 2 points), and generally important (quantified as 1 point).

[0158] The quantitative standards for preset personalized medical condition information are as follows: For different personalized medical information, specific quantitative thresholds are set as preset conditions. For example: Special physical conditions: When the quantitative score of allergic body weight reaches 2 points or more, the quantitative score of chronic diseases reaches 3 points or more, and the quantitative score of special physiological periods reaches 2 points or more, it is considered that the preset personalized medical conditions are met. Difference in severity of illness: When the quantitative score of severity of illness reaches 2 points or more, it is considered that the preset personalized medical conditions are met. Difference in treatment goals: When the quantitative score of relieving pain as soon as possible reaches 2 points or more or the quantitative score of long-term rehabilitation and prevention of recurrence reaches 2 points or more, it is considered that the preset personalized medical conditions are met.

[0159] The quantitative score of the patient's personalized medical information obtained is compared with the quantitative standard of the preset personalized medical condition information. If the patient's quantitative score reaches or exceeds the preset standard, the patient's personalized medical information is considered to have reached the preset personalized medical condition information. For example, the patient has mild allergies (quantitative score of 1 point), the severity of the disease is moderate (quantitative score of 2 points), and he urgently wants to relieve the pain as soon as possible (quantitative score of 2 points). At this time, since the quantitative score of the difference between the severity of the disease and the treatment goal reaches the preset standard, it is considered that the patient's personalized medical information has reached the preset personalized medical condition information.

[0160] Step S6e2, analyzes and determines the adjustment range of the physical therapy parameters based on the correspondence between the detailed results contained in the personalized medical information and the adjustment range of the physical therapy parameters, and redetermines the optimal physical parameters of the selected physical therapy plan based on the average physical therapy parameters of similar groups and the adjustment range of the physical therapy parameters.

[0161] Detailed results of personalized medical information: the specific personalized medical information of the patient determined in step S6e1, such as the specific type and degree of special physical conditions, the quantitative value of the severity of the disease, the specific manifestations of differences in treatment goals, etc.

[0162] Adjustment range of physical therapy parameters: The degree to which physical therapy parameters are adjusted based on personalized medical information. Different personalized medical information corresponds to different adjustment ranges to ensure that the physical therapy plan is suitable for the patient's individual situation and can achieve better treatment results.

[0163] According to the detailed results of the patient's personalized medical information determined in step S6e1, the corresponding adjustment range is found. For example, if the patient has a mild allergic constitution and a moderate condition, and hopes to relieve pain as soon as possible, then the personalized medical information of these three aspects is comprehensively considered to determine that the treatment intensity is reduced by 10%, and the treatment time and frequency are adjusted according to the moderate condition and the need to relieve pain as soon as possible. The treatment time can be the same as the average parameter or slightly increased, and the frequency can be increased by 10%-15%.

[0164] Step S6e3, taking the average physical therapy parameters of a similar group as the optimal physical parameters for using the selected physical therapy program.

[0165] According to the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan recommendations in different dimensions, the remote pelvic inflammatory disease treatment device parameters are adjusted including: Step S710, analyzing whether the patient has a preset special physical condition or risk factor; Pre-set special physical conditions: refers to specific physical conditions that may affect the effectiveness or safety of remote pelvic inflammatory disease treatment equipment, such as severe heart disease, allergic constitution, coagulation dysfunction, etc.

[0166] Risk factors include but are not limited to certain medications the patient is taking that may interact with the treatment device, and the patient's recent surgical history.

[0167] The specific analysis process is as follows: 1. Collect patient information: Obtain detailed information about the patient's physical condition through questionnaires filled out by the patient, doctor's consultations, and related examination reports. 2. Compare the collected patient information with the preset special physical conditions and risk factors one by one. For example, if the preset special physical conditions include heart disease, then check whether the patient has a diagnosis record of heart disease, whether there are heart-related symptoms (such as palpitations, chest tightness, etc.), and whether the electrocardiogram shows abnormalities. For risk factors, such as the interaction between the medication being taken and the treatment device, consult the drug instructions and relevant medical literature to determine whether the medication the patient is taking may affect the effectiveness or safety of the treatment device.

[0168] Step S720: if yes, the parameters determined according to the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan suggestions of different dimensions are sent to the terminal held by the relevant person in charge, and the parameters confirmed by the terminal held by the relevant person in charge are used as the parameters of the remote pelvic inflammatory disease treatment device; The person in charge can be a doctor, therapist, or equipment manager. They will receive parameter information and a description of the patient's special physical condition on the terminal. The person in charge will evaluate and confirm the parameters based on the patient's specific situation. They will consider the patient's special physical condition and risk factors to determine whether the parameters sent are appropriate. If adjustments are required, they can modify the parameters on the terminal and confirm the final parameter settings. These confirmed parameters will be used as the actual parameters of the remote pelvic inflammatory disease treatment device.

[0169] Step S730, if no, adjust the remote pelvic inflammatory disease treatment device parameters according to the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan recommendations of different dimensions.

[0170] The present application also provides a computer storage medium, including a computer storage medium that can be loaded and executed by a processor to implement the following Figures 1 to 4 Procedures for each step involved in the process.

[0171] The embodiments of this specific implementation method are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Therefore, all equivalent changes made based on the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis, characterized in that: include: Remote pelvic inflammatory disease treatment equipment; at least one processor; At least one memory storing a computer program, at least one processor configured to implement a remote pelvic inflammatory disease treatment method when executing the computer program, the specific steps are as follows: Obtaining the patient's pelvic images, physiological parameters, and textual information describing the patient's symptoms, wherein the physiological parameters include body temperature, heart rate, and blood pressure; According to the mapping relationship between the preset categories of comprehensive information on pelvic inflammatory disease of patients and the feature vector construction scheme, the feature vector construction schemes matching the comprehensive information on pelvic inflammatory disease of different categories are analyzed and determined, and the feature vector construction schemes for the comprehensive information on pelvic inflammatory disease of different categories of patients are executed to construct feature vectors for obtaining the comprehensive information on pelvic inflammatory disease of different categories of patients, wherein the preset categories of comprehensive information on pelvic inflammatory disease of patients include the patient's pelvic images, physiological parameters and text information describing the patient's symptoms; The feature vectors of the comprehensive information of pelvic inflammatory disease of different categories of patients are constructed and spliced ​​together to form a feature vector of the comprehensive information of pelvic inflammatory disease of the patient; Input the feature vector of the patient's pelvic inflammatory disease comprehensive information into a pre-trained diagnostic suggestion analysis model, and output a diagnostic suggestion, which includes the diagnosis result of pelvic inflammatory disease and the cause analysis; Obtain the patient's treatment history records, integrate the patient's pelvic inflammatory disease comprehensive information and treatment history data, perform data preprocessing and extract key features, and input the key features of this patient into the pre-built patient feature model, calculate the existing patient group whose similarity with this patient reaches the preset similarity threshold, and use it as the similar group for reference; Obtain the quantitative indicators of the treatment effects of the remote pelvic inflammatory disease treatment equipment in different dimensions for the similar groups of this reference, and determine the treatment plan recommendations in different dimensions based on the preset basis for formulating the treatment plan recommendations in different dimensions. The dimensions of the treatment plan recommendations include drug treatment recommendations, physical therapy recommendations, and lifestyle adjustment recommendations; Adjust the remote pelvic inflammatory disease treatment device parameters according to the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan recommendations in different dimensions.

2. A remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis according to claim 1, characterized in that: The feature vectors of the comprehensive information of pelvic inflammatory disease of different categories of patients are constructed and spliced ​​together to form a feature vector of the comprehensive information of pelvic inflammatory disease of patients, including: According to the preset weights of the comprehensive information of pelvic inflammatory disease of different categories of patients, the constructed and obtained feature vectors of the comprehensive information of pelvic inflammatory disease of different categories of patients are spliced ​​to form a feature vector of the comprehensive information of pelvic inflammatory disease of the patients.

3. A remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis according to claim 2, characterized in that: The weights of comprehensive information on pelvic inflammatory disease in different categories of patients are obtained by: Collect the comprehensive information of pelvic inflammatory disease from different patients, and use the preset correlation coefficient analysis method to calculate the correlation coefficient between the comprehensive information of pelvic inflammatory disease of different categories and the diagnosis results of pelvic inflammatory disease, and accumulate the sum of the correlation coefficients; The ratio of the correlation coefficient of different categories of pelvic inflammatory disease comprehensive information to the sum of the correlation coefficients is used as the weight of the corresponding category of pelvic inflammatory disease comprehensive information.

4. A remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis according to claim 3, characterized in that: The acquisition of the preset similarity threshold includes: The weights for obtaining comprehensive information on pelvic inflammatory disease in different categories of patients; According to the correspondence between the weight combination of the comprehensive information of pelvic inflammatory disease in different categories of patients and the preset threshold of similarity, the preset threshold of similarity is analyzed and determined.

5. A remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis according to any one of claims 1 to 4, characterized in that: If the dimension of the treatment plan recommendation is a drug treatment recommendation, according to the preset basis for formulating treatment plan recommendations of different dimensions, the treatment plan recommendations of different dimensions include: To count the various drugs used by the similar groups in this reference during the use of remote pelvic inflammatory disease treatment equipment, and to determine the frequency of use of each drug in the similar groups, as well as the symptom relief rate and side effect incidence rate of each drug in the similar groups; The comprehensive score of each drug is calculated based on the symptom relief effect and side effect incidence of each drug in similar groups and the preset comprehensive score formula; The drug with the highest comprehensive score is recommended as the selected drug, and the dosage and treatment course statistics of patients in the similar group who use the selected drug are conducted. The average dosage of the similar group is used as the patient's dosage this time, and the average treatment course of the similar group is used as the patient's treatment course this time; The drug with the highest comprehensive score, the patient's current medication dosage, and the patient's current treatment course are summarized as a drug treatment recommendation.

6. A remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis according to any one of claims 1 to 4, characterized in that: If the dimension of the treatment plan recommendation is physical therapy recommendation, according to the preset basis for formulating treatment plan recommendations of different dimensions, the treatment plan recommendations of different dimensions include: Obtain detailed data on similar groups after using various physical therapy methods, including pain relief and improvement rate of inflammatory indicators, including C-reactive protein and white blood cell count; The preset physical therapy scoring formula was used to calculate the comprehensive scores of different physical therapies. The physical therapy scoring formula was as follows: comprehensive score = pain relief rate × 0.4 + C-reactive protein improvement rate × 0.3 + white blood cell count improvement rate × 0.3; According to the comprehensive scoring results, the physical therapy with the highest comprehensive score was selected as the physical therapy plan selected this time; Statistics of physical therapy parameters among similar groups of patients using a specific physical therapy regimen; The average physical therapy parameters of a similar group of references are used as the optimal physical parameters for using the selected physical therapy plan, and the optimal physical parameters and the selected physical therapy plan are used as the physical therapy recommendations for this time.

7. A remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis according to claim 6, characterized in that: It also includes the steps that precede obtaining detailed data on similar groups of people after using various physical therapy methods, as follows: Analyze whether the quantitative effect value of a single physical therapy for patients reaches the preset quantitative effect value within the preset time range; If not, obtain detailed data on similar groups after using various physical therapy methods and proceed to the next steps; If yes, obtain detailed data on similar groups using different combinations of physical therapy methods; The preset combined treatment scoring formula was used to calculate the combined treatment scores of different physical therapy methods. The comprehensive score = pain relief rate × 0.4 + C-reactive protein improvement rate × 0.3 + white blood cell count improvement rate × 0.3 + combined treatment effect coefficient × weight; According to the comprehensive scoring results, the combination of physical therapy methods with the highest comprehensive score was selected as the physical therapy plan selected this time; According to the reasonable range of physical therapy parameters, a combination method is used to generate different parameter combinations; The generated parameter combination is input into the established parameter and effect relationship model, the therapeutic effect of the parameter combination is output, and the parameter combination with the best therapeutic effect is selected as the best physical parameters for using the selected physical therapy plan, and the best physical parameters and the selected physical therapy plan are used as the physical therapy recommendations for this time.

8. The remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis according to claim 6, characterized in that: The average physical therapy parameters of a similar population were referenced as the optimal physical parameters for the selected physical therapy regimen including: Obtaining the patient's personalized medical information and analyzing whether the patient's personalized medical information meets the preset personalized medical condition information, wherein the personalized medical information includes special physical conditions, differences in disease severity, and differences in treatment goals; If yes, then the adjustment range of the physical therapy parameters is analyzed and determined based on the correspondence between the detailed results contained in the personalized medical information and the adjustment range of the physical therapy parameters, and the optimal physical parameters of the selected physical therapy plan are re-determined based on the average physical therapy parameters of similar groups and the adjustment range of the physical therapy parameters; If not, the average physical therapy parameters of a reference similar group are taken as the optimal physical parameters using the selected physical therapy regimen.

9. The remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis according to claim 6, characterized in that: According to the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan recommendations in different dimensions, the remote pelvic inflammatory disease treatment device parameters are adjusted including: Analyze whether the patient has pre-set special physical conditions or risk factors; If yes, the parameters determined by the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan suggestions of different dimensions are sent to the terminal held by the relevant person in charge, and the parameters confirmed by the terminal held by the relevant person in charge are used as the parameters of the remote pelvic inflammatory disease treatment device; If not, adjust the remote pelvic inflammatory disease treatment device parameters according to the remote pelvic inflammatory disease treatment device parameters associated with the treatment plan recommendations in different dimensions.

10. A computer storage medium, characterized in that: It includes a program that can be loaded and executed by a processor to implement a remote pelvic inflammatory disease treatment method in a remote pelvic inflammatory disease treatment system combined with artificial intelligence diagnosis as described in any one of claims 1 to 9.

Images