Intelligent analysis system and method for medical atomization heating solutions based on medical needs

Abstract

The present invention discloses an intelligent analysis system and method for a medical atomization heating scheme based on medical needs, and relates to the field of medical data processing. The present invention collects characteristic data of historical patients to construct a historical patient characteristic data matrix; collects characteristic data of current patients, sets a similarity threshold, calculates the similarity between the characteristic data of the current patient and the characteristic data of each historical patient, and selects a medical atomization heating scheme according to the calculation result; constructs an initial long-short-term memory network model, and uses the historical patient characteristic data to train, test and optimize the initial long-short-term memory network model to obtain a final long-short-term memory network model; uses the final long-short-term memory network model to provide a medical atomization heating scheme for the current patient; the present invention combines similarity analysis with deep learning to achieve personalized customization of the medical atomization heating scheme, automatically adjusts to adapt to different patients, improves the treatment effect, and ensures safe and effective treatment.

Description

Technical Field

[0001] The present invention belongs to the field of medical data processing, and in particular, relates to an intelligent analysis system and method for medical atomization heating solutions based on medical needs. Background Art

[0002] Traditional medical atomization technology mainly converts drug liquid into fine particles through ultrasound, high pressure or vibration so that patients can inhale it through the respiratory tract. These methods play an important role in the treatment of respiratory diseases such as asthma and chronic obstructive pulmonary disease (COPD).

[0003] The heating schemes in existing medical atomization devices are often fixed and cannot be adjusted in real time according to different drugs and patient conditions, which affects the treatment effect. As patients' requirements for the quality of medical services increase, personalized medicine has become a development trend. Medical equipment needs to be able to provide customized treatment plans based on the specific conditions of each patient. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In response to the problems in the related technology, the present invention provides an intelligent analysis system and method for medical atomization heating schemes based on medical needs. Through artificial intelligence, machine learning and other technologies, the medical atomization heating scheme is automatically adjusted according to the needs and conditions of different patients.

[0006] (II) Technical solution

[0007] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0008] S1. Set a patient basic information feature set, a patient symptom feature set, and a patient physiological feature set, and collect basic information feature data, symptom feature data, and physiological feature data of patients who have been treated and recovered using the medical atomization heating solution in the past according to the features of the patient basic information feature set, the patient symptom feature set, and the patient physiological feature set, and construct a historical patient feature data matrix;

[0009] S2, collect the basic information feature data, symptom feature data and physiological feature data of the current patient to obtain the current patient feature data set; set a similarity threshold, calculate the similarity between the feature data of the current patient and the feature data of each patient in the historical patient feature data matrix, obtain a patient similarity set, and detect whether the patient similarity set contains a patient similarity ≥ the similarity threshold; if it does, select the medical atomization heating scheme of the patient with the highest similarity to treat the current patient; if not, execute S4;

[0010] S3, constructing an initial long short-term memory network model, and setting parameters, dividing the historical patient characteristic data matrix into a historical patient characteristic data training matrix and a historical patient characteristic data test matrix, training, testing and optimizing the initial long short-term memory network model, and obtaining the final long short-term memory network model;

[0011] S4. Use the final long short-term memory network model to provide medical atomization heating solutions for current patients.

[0012] Preferably, the S1 comprises the following steps:

[0013] S11. Set the patient's basic information feature set , patient symptom feature set , patient physiological characteristics set ;

[0014] in Indicates the first i Basic information features, n Indicates the total number of basic information features; basic information features include age, weight, gender, allergy history, lung function, etc.;

[0015] in Indicates the first i Symptoms, m Indicates the total number of symptom characteristics; symptom characteristics include dyspnea, wheezing, coughing, etc.;

[0016] in Indicates the patient's physiological characteristics. i physiological characteristics, p Indicates the total number of physiological characteristics; physiological characteristics include respiratory rate, heart rate, blood oxygen saturation, etc.;

[0017] S12. According to the characteristics of the patient basic information feature set, patient symptom feature set, and patient physiological feature set, collect the basic information feature data, symptom feature data, and physiological feature data of the patients who have been treated and recovered using the medical atomization heating solution, and construct a historical patient feature data matrix A ,as follows,

[0018] ;

[0019] in Represents the first i The patient's q feature data, e Represents the total number of patients in the historical patient feature data matrix;

[0020] The above steps achieve a comprehensive quantitative analysis of the patient's condition by setting the patient's basic information, symptoms and physiological characteristics, and constructing a historical patient characteristic data matrix, which helps to formulate personalized medical atomization heating plans, improve the targetedness and effectiveness of treatment, and provide data support for clinical research and resource optimization.

[0021] Preferably, S2 comprises the following steps:

[0022] S21. Collect the basic information feature data, symptom feature data and physiological feature data of the current patient to obtain the current patient feature data set , Indicates the current patient’s i feature data, q Represents the total number of features in the current patient feature dataset;

[0023] S22. Setting similarity threshold h , compare the feature data in the current patient feature data set with the feature data of each patient in the historical patient feature data matrix; obtain the patient similarity set ,in Represents the first i The similarity of the patients e Represents the total number of patients in the patient similarity set;

[0024] S23. Set the treatment effect data threshold and detect whether the patient similarity set contains ≥ similarity threshold h similarity of patients;

[0025] If included, select the medical atomization heating scheme of the patient with the highest similarity to treat the current patient; if not included, execute S4;

[0026] The above steps ensure that the selection of treatment plans is based on similarity analysis by collecting current patient characteristic data and comparing it with historical patient characteristic data. This helps to accurately match the most appropriate medical atomization heating plan, improve the level of personalization of treatment, thereby improving treatment effects, reducing unnecessary treatment trials, and providing patients with safer and more effective treatment plans.

[0027] Preferably, the S22 comprises the following steps:

[0028] S221, using the Euclidean distance formula to calculate the distance between the feature data in the current patient data set and the feature data of each patient in the historical patient feature data matrix, to obtain the patient Euclidean distance set The formula is as follows,

[0029] ;

[0030] in, Represents the current patient characteristics to the first in the historical patient characteristics data matrix i The distance between the characteristics of each patient, Indicates the current patient’s j Features, Represents the first i The patient's j Features, q represents the total number of features;

[0031] Use the Gaussian function to convert each Euclidean distance in the patient Euclidean distance set into similarity to obtain the patient similarity set , the Gaussian function formula is as follows,

[0032] ;

[0033] in Represents the first i The similarity of the patients is the standard deviation of the Gaussian function;

[0034] The above steps can accurately quantify the similarity between the current patient and the historical patients by calculating the distance of the feature data using the Euclidean distance formula and converting it into similarity using the Gaussian function.

[0035] Preferably, S3 comprises the following steps:

[0036] S31, construct an initial long short-term memory network model, set the initial batch size, initial hidden layer size, number of initial model units and initial learning rate of the initial long short-term memory network model, set the training ratio and test ratio of the initial long short-term memory network model to l 1 , l 2 ;

[0037] S32, according to the training ratio of the initial long short-term memory network model l 1 and test ratio l 2 , dividing the historical patient characteristic data matrix into a historical patient characteristic data training matrix and a historical patient characteristic data test matrix; setting the historical patient characteristic data training label matrix and the historical patient characteristic data test label matrix according to the medical atomization heating scheme of each patient;

[0038] S33, inputting the historical patient feature data training matrix and the historical patient feature data training label matrix into the initial long short-term memory network model, training the initial long short-term memory network model, and obtaining a trained long short-term memory network model after the training is completed;

[0039] S34, inputting the historical patient feature data test matrix and the historical patient feature data test label matrix into the trained long short-term memory network model for testing, and after the test optimization is completed, obtaining the final long short-term memory network model;

[0040] The above steps achieve efficient learning and prediction of patient characteristic data by constructing and training the long short-term memory network model, which can predict the optimal medical atomization heating scheme based on historical data, ensure the accuracy of the model through testing and optimization, and ultimately improve the prediction accuracy of the treatment plan and the effectiveness of clinical application.

[0041] Preferably, the S33 comprises the following steps:

[0042] S331, setting the training error threshold, the maximum number of training iterations, and the current number of training iterations;

[0043] S332, inputting the historical patient feature data training matrix and the historical patient feature data training label matrix into the initial long short-term memory network model, and repeatedly training the initial long short-term memory network model;

[0044] S333, adjusting the initial batch size, initial hidden layer size, and number of initial model units of the initial long short-term memory network model according to the training results of each round to obtain an adjusted long short-term memory network model, and continuing to train the adjusted long short-term memory network model. When the training error is ≤ the training error threshold or the current number of iterations is ≥ the maximum number of training iterations, the training is stopped to obtain a trained long short-term memory network model;

[0045] The above steps ensure that the LSTM model stops training after reaching a preset accuracy or number of iterations by setting training parameters and dynamically adjusting the model structure, thereby improving the training efficiency and prediction performance of the model and avoiding overfitting.

[0046] Preferably, the S34 comprises the following steps:

[0047] S341. Setting accuracy threshold o 1 The historical patient feature data test matrix is ​​input into the trained long short-term memory network model for testing to obtain the test output result, and the test output result is compared with the historical tracking feature data test label matrix to obtain the test accuracy. o 2;when o 2 ≥ o 1 When , the trained long short-term memory network model is used as the final long short-term memory network model; when o 2 < o 1 , optimizing the initial learning rate of the trained long short-term memory network model to obtain an optimized long short-term memory network model, and using the optimized long short-term memory network model as the final long short-term memory network model;

[0048] The above steps verify the predictive ability of the model by setting the accuracy threshold and testing the trained LSTM network model, ensuring that the model is adopted only when it reaches or exceeds the accuracy standard; if the test accuracy does not meet the standard, the model performance is improved by optimizing the learning rate, and finally a final LSTM network model with accurate and reliable prediction is obtained to guide the decision-making of medical atomization heating solutions.

[0049] Preferably, the step of optimizing the initial learning rate of the trained long short-term memory network model in S341 to obtain the optimized long short-term memory network model comprises the following steps:

[0050] S3411, construct a particle population; set the size of the particle population to B , then the particle population is expressed as , represents the particle population i Particles;

[0051] S3412: Randomly generate an initial position set of the particle population according to the initial learning rate ; represents the number of particles in the population i The initial position of each particle;

[0052] Initialize the particle speed ,in Indicates i The speed of each particle; initialize the optimal position of each particle ,in Indicates i The best position reached by each particle; Initialize the global best position G for F Any position in; Set the maximum number of iterations to H 1 ;

[0053] S3413. For each particle, update the speed ,int Indicates the current moment, t +1 means the next moment, J is the inertia weight, c 1 and c 2 is the learning factor, r 1 and r 2 is a random number in the range of 0 to 1, express t Moment i The speed of a particle, express t Moment i The position of a particle, yes t Moment i The best position that a particle can reach is yes t Moment i The global optimal position reached by each particle;

[0054] Update Location , the updated position is used as the new learning rate to obtain the updated learning rate, and the updated learning rate is used to train the trained long short-term memory network model and evaluate the performance. If the updated position The performance is better than , then update , if the new position The performance is not better than , keep the original ; at the time t +1, traverse the historical best positions of all particles , and select the position with the best performance as the new global best position ;

[0055] S3414, repeat S3413, when o 2 ≥ o 1 Or the number of iterations reaches H 1 When , the iteration process is stopped, and the learning rate value corresponding to the global best position is output to obtain the optimal learning rate;

[0056] S3415, using the optimal learning rate as the learning rate of the trained long short-term memory network model to obtain an optimized long short-term memory network model;

[0057] The above steps optimize the initial learning rate of the long short-term memory network model by using the particle swarm optimization algorithm, further realize the fine adjustment of the model parameters, and improve the training efficiency and prediction accuracy of the model.

[0058] Preferably, S4 comprises the following steps:

[0059] S41, substituting the characteristic data in the current patient data set into the final long short-term memory network model to obtain the medical atomization heating scheme for the current patient, and using the medical atomization heating scheme for the current patient to treat the current patient;

[0060] The above steps can generate a personalized medical atomization heating plan for the current patient by inputting the current patient data set into the optimized long short-term memory network model, and perform treatment accordingly; this process realizes data-driven treatment decisions, which helps to improve the targetedness and effectiveness of treatment, while reducing uncertainty and risk in the treatment process.

[0061] A medical atomization heating scheme intelligent analysis system based on medical needs, which is used to implement the above-mentioned medical atomization heating scheme intelligent analysis method based on medical needs, including a feature set setting module, a data collection module, a similarity calculation module, a training deep learning model module, and a medical atomization heating scheme selection module;

[0062] The feature set setting module is used to set a patient basic information feature set, a patient symptom feature set, and a patient physiological feature set according to factors affecting the medical atomization heating scheme;

[0063] The data collection module is used to collect basic information feature data, symptom feature data and physiological feature data of patients who have been treated and recovered using the medical atomization heating scheme in the past according to the features in the patient basic information feature set, the patient symptom feature set and the patient physiological feature set, and construct a historical patient feature data matrix; collect the feature data of the current patient to obtain the current patient feature data set;

[0064] The similarity calculation module is used to calculate the similarity between the characteristics of the current patient and the characteristics of each patient in the historical patient characteristic data matrix to obtain a patient similarity set;

[0065] The training deep learning model module is used to train, test and optimize the initial long short-term memory network model using historical patient feature data to obtain a final long short-term memory network model;

[0066] The medical atomization heating scheme selection module is used to calculate whether the patient similarity set contains ≥ similarity threshold hsimilarity of patients; if included, the medical atomization heating scheme of the patient with the highest similarity is selected to treat the current patient; if not included, the feature data in the current patient data set is substituted into the final long short-term memory network model to obtain the medical atomization heating scheme for the current patient, and the medical atomization heating scheme for the current patient is used to treat the current patient.

[0067] (III) Beneficial effects

[0068] The present invention has the following beneficial effects:

[0069] The present invention significantly improves the personalization level of medical atomization heating solutions through an intelligent analysis system, solves the problem that traditional equipment cannot be adjusted according to the specific conditions of patients, thereby improving treatment effects and patient satisfaction; combining similarity analysis and applying deep learning models can more comprehensively consider individual differences in patients; similarity analysis can locate similar cases more quickly, and deep learning models can provide more accurate treatment plans based on patient-specific data. The combination of the two makes the treatment plan more in line with the actual needs of each patient.

[0070] The present invention can automatically adjust the medical atomization heating scheme to better adapt to different drugs and patient conditions, help optimize the treatment plan, and reduce unnecessary intervention during the treatment process.

[0071] By utilizing the long short-term memory network model, the present invention can accurately predict and optimize the medical atomization heating scheme, effectively reducing the treatment cost while extending the service life of the equipment.

[0072] By introducing the particle population algorithm to optimize the learning rate, the present invention further improves the prediction accuracy of the model, ensures the safety and effectiveness of medical atomization treatment, and provides patients with more scientific and personalized medical services.

[0073] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0075] Figure 1 It is a flow chart of the intelligent analysis system and method of medical atomization heating scheme based on medical needs of the present invention. DETAILED DESCRIPTION

[0076] The following will be combined with the drawings in the embodiments of the invention to clearly and completely describe the technical solutions in the embodiments of the invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. Based on the embodiments in the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.

[0077] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0078] Embodiment 1:

[0079] See also Figure 1 The present invention discloses an intelligent analysis method of a medical atomization heating scheme based on medical needs, comprising the following steps:

[0080] S1. Set a patient basic information feature set, a patient symptom feature set, and a patient physiological feature set, and collect basic information feature data, symptom feature data, and physiological feature data of patients who have been treated and recovered using the medical atomization heating solution in the past according to the features of the patient basic information feature set, the patient symptom feature set, and the patient physiological feature set, and construct a historical patient feature data matrix;

[0081] The S1 comprises the following steps:

[0082] S11. Set the patient's basic information feature set , patient symptom feature set , patient physiological characteristics set ;

[0083] in Indicates the first i Basic information features, n Indicates the total number of basic information features; basic information features include age, weight, gender, allergy history, lung function, etc.;

[0084] in Indicates the first i Symptoms, m Indicates the total number of symptom characteristics; symptom characteristics include dyspnea, wheezing, coughing, etc.;

[0085] in Indicates the patient's physiological characteristics. i physiological characteristics, pIndicates the total number of physiological characteristics; physiological characteristics include respiratory rate, heart rate, blood oxygen saturation, etc.;

[0086] S12. Based on the features in the patient basic information feature set, patient symptom feature set, and patient physiological feature set, collect the basic information feature data, symptom feature data, and physiological feature data of patients who have been treated and recovered using the medical atomization heating solution, and construct a historical patient feature data matrix A ,as follows,

[0087] ;

[0088] in Represents the first i The patient's q feature data, e Represents the total number of patients in the historical patient feature data matrix;

[0089] S2, collect the basic information feature data, symptom feature data and physiological feature data of the current patient to obtain the current patient feature data set; set a similarity threshold, calculate the similarity between the feature data of the current patient and the feature data of each patient in the historical patient feature data matrix, obtain a patient similarity set, and detect whether the patient similarity set contains a patient similarity ≥ the similarity threshold; if it does, select the medical atomization heating scheme of the patient with the highest similarity to treat the current patient; if not, execute S4;

[0090] The S2 comprises the following steps:

[0091] S21. Collect the basic information feature data, symptom feature data and physiological feature data of the current patient to obtain the current patient feature data set , Indicates the current patient’s i feature data, q Represents the total number of features in the current patient feature dataset;

[0092] S22. Setting similarity threshold h , compare the feature data in the current patient feature data set with the feature data of each patient in the historical patient feature data matrix; obtain the patient similarity set ,in Represents the first i The similarity of the patients e Represents the total number of patients in the patient similarity set;

[0093] The S22 comprises the following steps:

[0094] S221, using the Euclidean distance formula to calculate the distance between the feature data in the current patient data set and the feature data of each patient in the historical patient feature data matrix, to obtain the patient Euclidean distance set The formula is as follows,

[0095] ;

[0096] in, Represents the current patient characteristics to the first in the historical patient characteristics data matrix i The distance between the characteristics of each patient, Indicates the current patient’s j Features, Represents the first i The patient's j Features, q represents the total number of features;

[0097] Use the Gaussian function to convert each Euclidean distance in the patient Euclidean distance set into similarity to obtain the patient similarity set , the Gaussian function formula is as follows,

[0098] ;

[0099] in Represents the first i The similarity of the patients is the standard deviation of the Gaussian function;

[0100] S23. Set the treatment effect data threshold and detect whether the patient similarity set contains ≥ similarity threshold h similarity of patients;

[0101] If included, select the medical atomization heating scheme of the patient with the highest similarity to treat the current patient; if not included, execute S4;

[0102] S3, constructing an initial long short-term memory network model, and setting parameters, dividing the historical patient characteristic data matrix into a historical patient characteristic data training matrix and a historical patient characteristic data test matrix, training, testing and optimizing the initial long short-term memory network model, and obtaining the final long short-term memory network model;

[0103] The S3 comprises the following steps:

[0104] S31, construct an initial long short-term memory network model, set the initial batch size, initial hidden layer size, number of initial model units and initial learning rate of the initial long short-term memory network model, set the training ratio and test ratio of the initial long short-term memory network model to l1 , l 2 ;

[0105] S32, according to the training ratio of the initial long short-term memory network model l 1 and test ratio l 2 , dividing the historical patient characteristic data matrix into a historical patient characteristic data training matrix and a historical patient characteristic data test matrix; setting the historical patient characteristic data training label matrix and the historical patient characteristic data test label matrix according to the medical atomization heating scheme of each patient;

[0106] S33, inputting the historical patient feature data training matrix and the historical patient feature data training label matrix into the initial long short-term memory network model, training the initial long short-term memory network model, and obtaining a trained long short-term memory network model after the training is completed;

[0107] The S33 comprises the following steps:

[0108] S331, setting the training error threshold, the maximum number of training iterations, and the current number of training iterations;

[0109] S332, inputting the historical patient feature data training matrix and the historical patient feature data training label matrix into the initial long short-term memory network model, and repeatedly training the initial long short-term memory network model;

[0110] S333, adjusting the initial batch size, initial hidden layer size, and number of initial model units of the initial long short-term memory network model according to the training results of each round to obtain an adjusted long short-term memory network model, and continuing to train the adjusted long short-term memory network model. When the training error is ≤ the training error threshold or the current number of iterations is ≥ the maximum number of training iterations, the training is stopped to obtain a trained long short-term memory network model;

[0111] S34, inputting the historical patient feature data test matrix and the historical patient feature data test label matrix into the trained long short-term memory network model for testing, and after the test optimization is completed, obtaining the final long short-term memory network model;

[0112] The S34 comprises the following steps:

[0113] S341. Setting accuracy threshold o 1 The historical patient feature data test matrix is ​​input into the trained long short-term memory network model for testing to obtain the test output result, and the test output result is compared with the historical tracking feature data test label matrix to obtain the test accuracy.o 2 ;when o 2 ≥ o 1 When , the trained long short-term memory network model is used as the final long short-term memory network model; when o 2 < o 1 , optimizing the initial learning rate of the trained long short-term memory network model to obtain an optimized long short-term memory network model, and using the optimized long short-term memory network model as the final long short-term memory network model;

[0114] The step S341 of optimizing the initial learning rate of the trained long short-term memory network model to obtain the optimized long short-term memory network model comprises the following steps:

[0115] S3411, construct a particle population; set the size of the particle population to B , then the particle population is expressed as , represents the particle population i Particles;

[0116] S3412: Randomly generate an initial position set of the particle population according to the initial learning rate ; represents the number of particles in the population i The initial position of each particle;

[0117] Initialize the particle speed ,in Indicates i The speed of each particle; initialize the optimal position of each particle ,in Indicates i The best position reached by each particle; Initialize the global best position G for F Any position in; Set the maximum number of iterations to H 1 ;

[0118] S3413. For each particle, update the speed ,in t Indicates the current moment, t +1 means the next moment, J is the inertia weight, c 1 and c 2 is the learning factor, r1 and r 2 is a random number in the range of 0 to 1, express t Moment i The speed of a particle, express t Moment i The position of a particle, yes t Moment i The best position that a particle can reach is yes t Moment i The global optimal position reached by each particle;

[0119] Update Location , the updated position is used as the new learning rate to obtain the updated learning rate, and the updated learning rate is used to train the trained long short-term memory network model and evaluate the performance. If the updated position The performance is better than , then update , if the new position The performance is not better than , keep the original ; at the time t +1, traverse the historical best positions of all particles , and select the position with the best performance as the new global best position ;

[0120] S3414, repeat S3413, when o 2 ≥ o 1 Or the number of iterations reaches H 1 When , the iteration process is stopped, and the learning rate value corresponding to the global best position is output to obtain the optimal learning rate;

[0121] S3415, using the optimal learning rate as the learning rate of the trained long short-term memory network model to obtain an optimized long short-term memory network model;

[0122] S4. Use the final long short-term memory network model to provide a medical atomization heating solution for the current patient;

[0123] The S4 comprises the following steps:

[0124] S41. Substitute the feature data in the current patient data set into the final long short-term memory network model to obtain the current medical atomization heating scheme, and use the current medical atomization heating scheme to treat the current patient.

[0125] Embodiment 2:

[0126] A medical atomization heating scheme intelligent analysis system based on medical needs, which is used to implement the above-mentioned medical atomization heating scheme intelligent analysis method based on medical needs, including a feature set setting module, a data collection module, a similarity calculation module, a training deep learning model module, and a medical atomization heating scheme selection module;

[0127] The feature set setting module is used to set a patient basic information feature set, a patient symptom feature set, and a patient physiological feature set according to factors affecting the medical atomization heating scheme;

[0128] The data collection module is used to collect basic information feature data, symptom feature data and physiological feature data of patients who have been treated and recovered using the medical atomization heating scheme in the past according to the features in the patient basic information feature set, the patient symptom feature set and the patient physiological feature set, and construct a historical patient feature data matrix; collect the feature data of the current patient to obtain the current patient feature data set;

[0129] The similarity calculation module is used to calculate the similarity between the characteristics of the current patient and the characteristics of each patient in the historical patient characteristic data matrix to obtain a patient similarity set;

[0130] The training deep learning model module is used to train, test and optimize the initial long short-term memory network model using historical patient feature data to obtain a final long short-term memory network model;

[0131] The medical atomization heating scheme selection module is used to calculate whether the patient similarity set contains ≥ similarity threshold h similarity of patients; if included, the medical atomization heating scheme of the patient with the highest similarity is selected to treat the current patient; if not included, the feature data in the current patient data set is substituted into the final long short-term memory network model to obtain the medical atomization heating scheme for the current patient, and the medical atomization heating scheme for the current patient is used to treat the current patient.

[0132] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0133] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. An intelligent analysis method for medical atomization heating scheme based on medical needs, characterized in that: The following steps are involved: S1. Set a patient basic information feature set, a patient symptom feature set, and a patient physiological feature set, and collect basic information feature data, symptom feature data, and physiological feature data of patients who have been treated and recovered using the medical atomization heating solution in the past according to the features of the patient basic information feature set, the patient symptom feature set, and the patient physiological feature set, and construct a historical patient feature data matrix; The S1 comprises the following steps: S11. Set the patient's basic information feature set , patient symptom feature set , patient physiological characteristics set ; in a i Indicates the first i Basic information features, n Indicates the total number of basic information features; basic information features include age, weight, gender, allergy history, and lung function; in b i Indicates the first i Symptoms, m Indicates the total number of symptom characteristics; symptom characteristics include dyspnea, wheezing, and coughing; in c i Indicates the patient's physiological characteristics. i physiological characteristics, p Indicates the total number of physiological characteristics; physiological characteristics include respiratory rate, heart rate, and blood oxygen saturation; S12. According to the characteristics of the patient basic information feature set, patient symptom feature set, and patient physiological feature set, collect the basic information feature data, symptom feature data, and physiological feature data of the patients who have been treated and recovered using the medical atomization heating solution, and construct a historical patient feature data matrix A ,as follows, ； in A iq Represents the first i The patient's q feature data, e Represents the total number of patients in the historical patient feature data matrix; S2, collect the basic information feature data, symptom feature data and physiological feature data of the current patient to obtain the current patient feature data set; set a similarity threshold, calculate the similarity between the feature data of the current patient and the feature data of each patient in the historical patient feature data matrix, obtain a patient similarity set, and detect whether the patient similarity set contains a patient similarity ≥ the similarity threshold; if it does, select the medical atomization heating scheme of the patient with the highest similarity to treat the current patient; if not, execute S4; S3, constructing an initial long short-term memory network model, and setting parameters, dividing the historical patient characteristic data matrix into a historical patient characteristic data training matrix and a historical patient characteristic data test matrix, training, testing and optimizing the initial long short-term memory network model, and obtaining the final long short-term memory network model; S4. Use the final long short-term memory network model to provide medical atomization heating solutions for current patients.

2. The intelligent analysis method of medical atomization heating scheme based on medical needs according to claim 1 is characterized in that: The S2 comprises the following steps: S21, collecting basic information feature data, symptom feature data and physiological feature data of the current patient to obtain a feature data set of the current patient; S22. Setting similarity threshold h , compare the feature data in the current patient feature data set with the feature data of each patient in the historical patient feature data matrix to obtain a patient similarity set; S23. Set the treatment effect data threshold and detect whether the patient similarity set contains ≥ similarity threshold h similarity of patients; If included, select the medical atomization heating plan for the patient with the highest similarity to treat the current patient; if not included, execute S4.

3. The intelligent analysis method of medical atomization heating scheme based on medical needs according to claim 2 is characterized in that: The S22 comprises the following steps: S221, using the Euclidean distance formula to calculate the distance between the feature data in the current patient data set and the feature data of each patient in the historical patient feature data matrix, to obtain the patient Euclidean distance set , the formula is as follows, ； in, Represents the current patient characteristics to the first in the historical patient characteristics data matrix i The distance between the characteristics of each patient, f j Indicates the current patient’s j Features, A ij Represents the first i The patient's j Features, q represents the total number of features; Use the Gaussian function to convert each Euclidean distance in the patient Euclidean distance set into similarity to obtain the patient similarity set , the Gaussian function formula is as follows, ； in g i Represents the first i The similarity of the patients is the standard deviation of the Gaussian function.

4. The intelligent analysis method of medical atomization heating scheme based on medical needs according to claim 1 is characterized in that: The S3 comprises the following steps: S31, construct an initial long short-term memory network model, set the initial batch size, initial hidden layer size, number of initial model units and initial learning rate of the initial long short-term memory network model, set the training ratio and test ratio of the initial long short-term memory network model to l 1. l 2; S32, according to the training ratio of the initial long short-term memory network model l 1 and test ratio l 2. Divide the historical patient characteristic data matrix into a historical patient characteristic data training matrix and a historical patient characteristic data test matrix; set the historical patient characteristic data training label matrix and the historical patient characteristic data test label matrix according to the medical atomization heating scheme of each patient; S33, inputting the historical patient feature data training matrix and the historical patient feature data training label matrix into the initial long short-term memory network model, training the initial long short-term memory network model, and obtaining a trained long short-term memory network model after the training is completed; S34, inputting the historical patient feature data test matrix and the historical patient feature data test label matrix into the trained long short-term memory network model for testing, and after the test optimization is completed, obtaining the final long short-term memory network model.

5. The intelligent analysis method of medical atomization heating scheme based on medical needs according to claim 4 is characterized in that: The S33 comprises the following steps: S331, setting the training error threshold, the maximum number of training iterations, and the current number of training iterations; S332, inputting the historical patient feature data training matrix and the historical patient feature data training label matrix into the initial long short-term memory network model, and repeatedly training the initial long short-term memory network model; S333. Adjust the initial batch size, initial hidden layer size, and number of initial model units of the initial long short-term memory network model according to the training results of each round to obtain an adjusted long short-term memory network model, and continue to train the adjusted long short-term memory network model. When the training error is ≤ the training error threshold or the current number of iterations is ≥ the maximum number of training iterations, stop the training to obtain a trained long short-term memory network model.

6. The intelligent analysis method of medical atomization heating scheme based on medical needs according to claim 4 is characterized in that: The S34 comprises the following steps: S341. Setting accuracy threshold o 1. Input the historical patient feature data test matrix into the trained long short-term memory network model for testing, obtain the test output result, and compare the test output result with the historical tracking feature data test label matrix to obtain the test accuracy o 2. When o 2≥ o 1, the trained long short-term memory network model is used as the final long short-term memory network model; when o 2< o 1, the initial learning rate of the trained long short-term memory network model is optimized to obtain an optimized long short-term memory network model, and the optimized long short-term memory network model is used as the final long short-term memory network model.

7. The intelligent analysis method of medical atomization heating scheme based on medical needs according to claim 6 is characterized in that: The step S341 of optimizing the initial learning rate of the trained long short-term memory network model to obtain the optimized long short-term memory network model comprises the following steps: S3411, construct a particle population; set the size of the particle population to B , then the particle population is expressed as , C i represents the particle population i Particles; S3412: Randomly generate an initial position set of the particle population according to the initial learning rate ; D i represents the number of particles in the population i The initial position of each particle; Initialize the particle speed ,in E i Indicates i The speed of each particle; initialize the optimal position of each particle ,in F i Indicates i The best position reached by each particle; Initialize the global best position G for F Any position in; Set the maximum number of iterations to H 1; S3413. For each particle, update the speed ,in t Indicates the current moment, t +1 means the next moment, J is the inertia weight, c 1 and c 2 is the learning factor, r 1 and r 2 is a random number in the range of 0 to 1, express t Moment i The speed of a particle, express t Moment i The position of a particle, yes t Moment i The best position that a particle can reach is yes t Moment i The global optimal position reached by each particle; Update Location , the updated position is used as the new learning rate to obtain the updated learning rate, and the updated learning rate is used to train the trained long short-term memory network model and evaluate the performance. If the updated position The performance is better than , then update F i , if the new position The performance is not better than , keep the original F i ; at the time t +1, traverse the historical best positions of all particles F i , and select the position with the best performance as the new global best position ; S3414, repeat S3413, when o 2≥ o 1 or the number of iterations reaches H When 1, the iteration process is stopped, the learning rate value corresponding to the global best position is output, and the optimal learning rate is obtained; S3415. Using the optimal learning rate as the learning rate of the trained long short-term memory network model to obtain an optimized long short-term memory network model.

8. The intelligent analysis method of medical atomization heating scheme based on medical needs according to claim 1 is characterized in that: The S4 comprises the following steps: S41. Substitute the feature data in the current patient data set into the final long short-term memory network model to obtain the current medical atomization heating scheme, and use the current medical atomization heating scheme to treat the current patient.

9. A medical atomization heating scheme intelligent analysis system based on medical needs, used to implement a medical atomization heating scheme intelligent analysis method based on medical needs as described in any one of claims 1 to 8; characterized in that: It includes a feature set setting module, a data collection module, a similarity calculation module, a deep learning model training module, and a medical atomization heating scheme selection module.

Images