A prediction method and device based on clinical pathway model

Abstract

The present invention discloses a prediction method and device based on a clinical pathway model, and relates to the field of medical technology. The method includes: obtaining characteristic information of a diabetic patient, and determining a training data set according to the characteristic information; establishing an LSTM neural network, and using the training data set to train the LSTM neural network to obtain a blood glucose value prediction model; selecting target characteristic data with a preset number of occurrence frequencies higher than a threshold value during the training process of the blood glucose value prediction model; and comparing the predicted values ​​obtained by modifying the target characteristic data in the characteristic information and inputting them into the blood glucose value prediction model with the normal predicted values ​​determined by the characteristic information, so as to determine the key features for predicting the blood glucose value of the diabetic patient from the target characteristic data. The present invention solves the problem in the prior art that a large amount of characteristic information data of the patient is required when predicting the patient's blood glucose value, resulting in high resource consumption of the hospital.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to a prediction method and device based on a clinical pathway model. Background Art

[0002] Clinical pathway refers to a targeted, authoritative and sequential diagnosis and treatment pathway developed by experts for a certain disease or surgery based on experience or actual data. It contains a large amount of characteristic information that can help doctors with treatment. For example, when facing patients with type 2 diabetes, the characteristic information of the patient in the clinical pathway can be used to predict blood sugar levels.

[0003] However, in the existing technology, blood glucose levels are predicted directly through all the characteristic information of patients contained in the existing clinical pathway, so a large amount of characteristic information needs to be collected before making a specific prediction. The information may also include characteristics that have little effect on the prediction of blood glucose levels, increasing the resource consumption of the hospital. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a prediction method and device based on a clinical pathway model, aiming to solve the problem in the prior art that a large amount of characteristic information data of patients is required when predicting the blood glucose level of patients, resulting in high resource consumption of hospitals.

[0005] The embodiment of the present invention is implemented as follows:

[0006] A prediction method based on a clinical pathway model, the method comprising:

[0007] Acquire characteristic information of the diabetic patients from diabetic patients who have used a preset clinical pathway, and determine a training data set based on the characteristic information;

[0008] Establishing an LSTM neural network, and using the training data set to train the LSTM neural network to obtain a blood sugar value prediction model;

[0009] Selecting a preset number of target feature data with a frequency higher than a threshold value during the training process of the blood glucose level prediction model;

[0010] The target feature data in the feature information are modified and input into the blood glucose level prediction model to obtain predicted values, which are compared with normal predicted values ​​determined by the feature information, so as to determine the key features for predicting the blood glucose level of the diabetic patient from the target feature data.

[0011] Furthermore, in the above-mentioned prediction method based on the clinical pathway model, the step of selecting a preset number of target feature data with an occurrence frequency higher than a threshold value during the training process of the blood glucose value prediction model includes:

[0012] The LSTM neural network is trained a preset number of times using the training data set, and the training results of each training are obtained. The target number of times that the training results meet the preset conditions is determined from all the training times, and the preset number of target feature data whose occurrence frequency is higher than the threshold value during the target number of training is obtained.

[0013] Furthermore, in the above-mentioned prediction method based on the clinical pathway model, the step of modifying the target feature data in the feature information and inputting the modified target feature data into the blood glucose level prediction model to obtain a predicted value and comparing it with a normal predicted value determined by the feature information to determine a key feature for predicting the blood glucose level of the diabetic patient from the target feature data includes:

[0014] A difference between the predicted value and a normal predicted value determined by the feature information is determined, and the target feature data whose difference is within a preset range is determined as the key feature.

[0015] Furthermore, in the above-mentioned prediction method based on the clinical pathway model, the step of determining the training data set according to the characteristic information includes:

[0016] The characteristic information is merged and cleaned, and then the characteristic information is processed for exceptions to determine missing features of missing data in the characteristic information;

[0017] The missing features are filled with the historical data average of the missing features, and then the feature information is normalized.

[0018] Furthermore, in the above-mentioned prediction method based on the clinical pathway model, after the step of determining the key features for predicting the blood glucose level of the diabetic patient from the target feature data, the step further includes:

[0019] Inputting the characteristic information and key characteristic data as input data into the blood glucose level prediction model to obtain a characteristic blood glucose level and a key blood glucose level;

[0020] Respectively obtaining the root mean square errors of the characteristic blood sugar value and the key blood sugar value, and comparing the root mean square errors of the characteristic blood sugar value and the key blood sugar value;

[0021] If the root mean square error between the characteristic blood glucose value and the key blood glucose value is within a preset range, it is determined that the key feature is used to accurately predict the blood glucose value of the diabetic patient.

[0022] Furthermore, in the above-mentioned prediction method based on the clinical pathway model, the step of obtaining characteristic information of the diabetic patients from the diabetic patients who have used the preset clinical pathway includes:

[0023] First, find the target clinical pathway whose number of pathways is higher than the preset threshold from the preset clinical pathway database;

[0024] A final clinical pathway with a number of visits higher than a preset threshold is selected from the target clinical pathway as the preset clinical pathway.

[0025] Furthermore, in the above-mentioned prediction method based on the clinical pathway model, after the step of determining the key features for predicting the blood glucose level of the diabetic patient from the target feature data, the step further includes:

[0026] The key feature data of the diabetic patient are obtained, and the key feature data are input into the blood glucose level prediction model to obtain the blood glucose level of the diabetic patient.

[0027] Another object of the present invention is to provide a prediction device based on a clinical pathway model, the device comprising:

[0028] An acquisition module, used to acquire characteristic information of the diabetic patients from the diabetic patients who have used the preset clinical pathway, and determine a training data set according to the characteristic information;

[0029] A training module, used to establish an LSTM neural network, and train the LSTM neural network using the training data set to obtain a blood sugar value prediction model;

[0030] A selection module, used for selecting a preset number of target feature data with an occurrence frequency higher than a threshold value during the training process of the blood glucose level prediction model;

[0031] A prediction module is used to modify the target feature data in the feature information and input the resulting predicted value into the blood glucose value prediction model, and compare it with the normal predicted value determined by the feature information, so as to determine the key features for predicting the blood glucose value of the diabetic patient from the target feature data.

[0032] Furthermore, in the above-mentioned prediction device based on the clinical pathway model, the prediction module is specifically used for:

[0033] A difference between the predicted value and a normal predicted value determined by the feature information is determined, and the target feature data whose difference is within a preset range is determined as the key feature.

[0034] Another object of the present invention is to provide a readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the steps of any one of the methods described above.

[0035] Another object of the present invention is to provide an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the program.

[0036] The present invention obtains characteristic information of diabetic patients from diabetic patients who have used a preset clinical pathway, and determines a training data set according to the characteristic information; establishes an LSTM neural network, and uses the training data set to train the LSTM neural network to obtain a blood glucose value prediction model; selects a preset number of target characteristic data with an occurrence frequency higher than a threshold value during the training process of the blood glucose value prediction model; and compares the predicted values ​​obtained by modifying the target characteristic data in the characteristic information and inputting them into the blood glucose value prediction model with the normal predicted values ​​determined by the characteristic information, so as to determine the key features for predicting the blood glucose value of diabetic patients from the target characteristic data, predict the blood glucose value by the key features, and reduce the collection of characteristic information data. The present invention solves the problem that a large amount of characteristic information data of patients is required when predicting the blood glucose value of patients in the prior art, resulting in high resource consumption of hospitals. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a flow chart of a prediction method based on a clinical pathway model in a first embodiment of the present invention;

[0038] Figure 2 It is a structural block diagram of a prediction device based on a clinical pathway model in the third embodiment of the present invention.

[0039] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0040] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0041] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0043] The following will describe in detail how to predict the patient's blood glucose level through key features in conjunction with specific embodiments and drawings, thereby reducing the amount of data collected when predicting the blood glucose level.

[0044] Embodiment 1

[0045] See also Figure 1 , which shows a prediction method based on a clinical pathway model in a first embodiment of the present invention, and the method includes steps S10 to S13.

[0046] Step S10, acquiring characteristic information of the diabetic patients from the diabetic patients who have used the preset clinical pathway, and determining a training data set according to the characteristic information.

[0047] Among them, first of all, it is necessary to collect sample data. Specifically, the patient's characteristic information can be collected from patients who have used the preset clinical pathway in multiple large hospitals, including the patient's age, the patient's illness duration, the food intake of the day, etc. In the embodiment of the present invention, it is mainly aimed at type 2 diabetes, and sample data of type 2 diabetes can be collected. Specifically, in order to improve the breadth and representativeness of the data, in some optional embodiments of the present invention, the target clinical pathway with a number of pathways higher than the preset threshold can be first found from the preset clinical pathway database, that is, the target clinical pathway with a large number of uses can be first found; the final clinical pathway with a number of people higher than the preset threshold is selected from the target clinical pathway as the preset clinical pathway.

[0048] In the specific implementation, the resource scheduling of clinical pathways can be optimized to improve the utilization rate of medical resources. First, according to the above rules, the clinical pathways with more clinical pathways in the current hospital are determined, and the clinical pathways with higher number of patients are selected. Secondly, according to the order relationship between the medical order tasks in the clinical pathway, and the relationship between the number of medical resources required for each medical order task, a clinical pathway resource scheduling model is constructed.

[0049] Thirdly, the particle swarm algorithm is used to optimize the scheduling model. The particle swarm algorithm is an optimization algorithm that simulates the foraging behavior of bird flocks. It is suitable for solving complex scheduling and resource allocation problems. The goal is to find a resource scheduling solution with the shortest treatment days and the largest number of treated patients. In order to avoid the particle swarm algorithm from falling into local optimality, in some optional embodiments of the present invention, the individual position update method of the wolf in the gray wolf optimization algorithm can be integrated into the particle swarm algorithm. This fusion can enhance the global search capability of the algorithm and increase the probability of finding the best resource scheduling solution.

[0050] Specifically, the resource scheduling optimization model can be expressed as:

[0051] ; (1)

[0052] ; (2)

[0053] Among them, formula (1) is the objective function, which means minimizing the total number of treatment days, and formula (2) is the constraint condition, which means that the total demand for resources cannot exceed the given total resources. T represents the total number of treatment days, t i represents the treatment time of each task, r i Indicates the resources required for each task, R Indicates the total resource limit.

[0054] More specifically, the wolf individual position update method in the gray wolf optimization algorithm can be integrated into the particle swarm algorithm as follows:

[0055] First, initialize the particle swarm, set the initial position and speed of the particles, initialize the three optimal individual positions of the Grey Wolf Optimization Algorithm, calculate the fitness value of each particle, and update its individual optimal position and global optimal position. Update the three optimal positions of the Grey Wolf Optimization Algorithm, calculate the speed and position of each particle according to the speed and position update formula of the particle swarm algorithm, and calculate the updated position of each particle in combination with the update strategy of the Grey Wolf Optimization Algorithm to determine whether the maximum number of iterations or other termination conditions have been reached. If not, continue to iterate; otherwise, output the optimal solution to obtain the final clinical pathway.

[0056] Step S11, establishing an LSTM neural network, and using the training data set to train the LSTM neural network to obtain a blood sugar value prediction model.

[0057] Among them, the blood sugar value prediction model is trained by using the model training method. The blood sugar value prediction model masters the law of blood sugar value prediction, so that the blood sugar value can be accurately predicted. Specifically,

[0058] Blood sugar levels are affected by many factors, such as circadian rhythm, diet, hormone levels in the body, drug effects, etc. Therefore, changes in blood sugar levels are continuous and there is a certain correlation between different time points. LSTM has the ability to remember and learn patterns and trends in historical data. It can learn the relationship between blood sugar levels in different time steps from historical blood sugar data and use these relationships to predict future blood sugar levels.

[0059] When the embodiment of the present invention is implemented, it is necessary to simultaneously predict the patient's blood glucose level in the next 30 minutes, 60 minutes, 90 minutes, and 120 minutes, so a multi-task learning model is needed. LSTM can be easily expanded into a multi-task learning model, which can simultaneously predict blood glucose levels at multiple time points, thereby better coping with the variability and complexity of blood glucose value prediction. Therefore, a multi-task LSTM algorithm model is used to predict the blood glucose level of patients in the clinical pathway of type 2 diabetes.

[0060] Step S12, selecting a preset number of target feature data with an occurrence frequency higher than a threshold during the training process of the blood sugar level prediction model.

[0061] Among them, the LSTM neural network is trained for a preset number of times using the training data set, and the training results of each time are obtained. The target number of times that the training results meet the preset conditions is determined from all the training times, and the target feature data with a preset number of occurrences higher than the threshold value during the target number of trainings is obtained. In the process of model training, the feature information used for model training can be used as the target feature data with a preset number of occurrences higher than the threshold value. For example, the blood glucose value prediction model is trained 500 times using the patient's feature data, and the 20 trainings with the best performance are selected. The performance can be sorted according to a certain evaluation index (such as accuracy, F1 score, etc.), and the frequency of occurrence of each patient feature is counted for the selected 20 training results. The number of occurrences of each feature is added up to obtain the total number of occurrences of each feature in these 20 trainings, and the features are sorted according to the total number of occurrences, and the features with the number exceeding the threshold value are selected, for example, the first 15 features with the largest number of occurrences are selected. In addition, in some optional embodiments of the present invention, the number of all feature data during training can also be obtained, and the number of target feature data to be selected is determined according to the number of types of all feature data used according to the set ratio.

[0062] Step S13, respectively modifying the target feature data in the feature information and inputting the resulting predicted values ​​into the blood glucose level prediction model, and comparing the predicted values ​​with the normal predicted values ​​determined by the feature information, so as to determine the key features for predicting the blood glucose level of the diabetic patient from the target feature data.

[0063] Among them, the target feature data in the feature information is modified and input into the blood glucose value prediction model to obtain the predicted value of blood glucose. For example, one modification is made, and the feature information without data modification is input into the blood glucose value prediction model to obtain a normal blood glucose prediction value. The two are compared. The greater the difference, the greater the influence of the feature on the blood glucose prediction value, which is the key feature.

[0064] Specifically, the difference between the predicted value and the normal predicted value determined by the feature information can be determined, and the target feature data with the difference within a preset range can be determined as the key feature, wherein the preset range can be set according to the actual situation. For example, when the difference is greater than a certain threshold, the target feature data corresponding to the difference is determined as the key feature, so that the key feature that meets the conditions can be selected from the target feature data, and the blood glucose value can be predicted by the key feature. For example, the key feature data of the diabetic patient is obtained, and the key feature data is input into the blood glucose value prediction model to predict the blood glucose value of the diabetic patient.

[0065] In addition, in some optional embodiments of the invention, in order to further verify the accuracy of key features in predicting blood glucose values, the feature information and key feature data are respectively input as input data into the blood glucose value prediction model to obtain the characteristic blood glucose value and the key blood glucose value; the root mean square errors of the characteristic blood glucose value and the key blood glucose value are respectively obtained, and the root mean square errors of the characteristic blood glucose value and the key blood glucose value are compared; if the root mean square errors of the characteristic blood glucose value and the key blood glucose value are within a preset range, it is determined that the key feature can be used to accurately predict the blood glucose value of the diabetic patient.

[0066] Among them, after determining the key features, the root mean square error obtained by using the patient's key feature information as input data and the root mean square error obtained by using all the patient's feature information as input data are compared, thereby proving that the key features can be used to predict the patient's blood sugar level during the treatment process of the clinical pathway of type 2 diabetes.

[0067] In summary, the prediction method based on the clinical pathway model in the above embodiment of the present invention obtains the characteristic information of diabetic patients from diabetic patients who have used the preset clinical pathway, and determines the training data set according to the characteristic information; establishes an LSTM neural network, and uses the training data set to train the LSTM neural network to obtain a blood glucose value prediction model; selects the target characteristic data with a preset number of occurrence frequencies higher than the threshold value during the training process of the blood glucose value prediction model; and compares the predicted values ​​obtained by modifying the target characteristic data in the characteristic information and inputting them into the blood glucose value prediction model with the normal predicted values ​​determined by the characteristic information, so as to determine the key features for predicting the blood glucose value of diabetic patients from the target characteristic data, predict the blood glucose value by the key features, and reduce the collection of characteristic information data. The problem that a large amount of characteristic information data of patients is required in the prior art when predicting the blood glucose value of patients, resulting in high resource consumption of hospitals, is solved.

[0068] Embodiment 2

[0069] This embodiment also proposes a prediction method based on a clinical pathway model. The difference between the prediction method based on a clinical pathway model in this embodiment and the prediction method based on a clinical pathway model in Example 1 is that:

[0070] Step S10 includes:

[0071] The characteristic information is merged and cleaned, and then the characteristic information is processed for exceptions to determine missing features of missing data in the characteristic information;

[0072] The missing features are filled with the historical data average of the missing features, and then the feature information is normalized.

[0073] Among them, the feature information is preprocessed by merging, cleaning, normalizing, etc., and useless information is cleaned, such as the patient's diagnosis information, the patient's attending physician information, etc. The feature information that is useful for this article is retained, such as the patient's age, the patient's illness duration, the patient's blood sugar value measured every 15 minutes, and other information. After cleaning the useless information, the data needs to be processed abnormally. Some of the collected information has not been fully collected. For example, some patients did not record their food intake in detail, and some heart rate data will be lost due to equipment reasons. For missing data, fill it with the average value of historical or most patients. Finally, the Max-Min method is used to normalize the data.

[0074] In summary, the prediction method based on the clinical pathway model in the above embodiment of the present invention obtains the characteristic information of diabetic patients from diabetic patients who have used the preset clinical pathway, and determines the training data set according to the characteristic information; establishes an LSTM neural network, and uses the training data set to train the LSTM neural network to obtain a blood glucose value prediction model; selects the target characteristic data with a preset number of occurrence frequencies higher than the threshold value during the training process of the blood glucose value prediction model; and compares the predicted values ​​obtained by modifying the target characteristic data in the characteristic information and inputting them into the blood glucose value prediction model with the normal predicted values ​​determined by the characteristic information, so as to determine the key features for predicting the blood glucose value of diabetic patients from the target characteristic data, predict the blood glucose value by the key features, and reduce the collection of characteristic information data. The problem that a large amount of characteristic information data of patients is required in the prior art when predicting the blood glucose value of patients, resulting in high resource consumption of hospitals, is solved.

[0075] Embodiment 3

[0076] See also Figure 2 , shown is a prediction device based on a clinical pathway model proposed in a third embodiment of the present invention, the device comprising:

[0077] An acquisition module 100 is used to acquire characteristic information of the diabetic patients from the diabetic patients who have used the preset clinical pathway, and determine a training data set according to the characteristic information;

[0078] A training module 200 is used to establish an LSTM neural network, and train the LSTM neural network using the training data set to obtain a blood sugar value prediction model;

[0079] A selection module 300, used for selecting a preset number of target feature data with an occurrence frequency higher than a threshold value during the training process of the blood glucose level prediction model;

[0080] The prediction module 400 is used to modify the target feature data in the feature information and input the resulting predicted value into the blood glucose value prediction model, and compare it with the normal predicted value determined by the feature information, so as to determine the key features for predicting the blood glucose value of the diabetic patient from the target feature data.

[0081] Furthermore, in the above-mentioned prediction device based on the clinical pathway model, the selection module is specifically used for:

[0082] The LSTM neural network is trained a preset number of times using the training data set, and the training results of each training are obtained. The target number of times that the training results meet the preset conditions is determined from all the training times, and the preset number of target feature data whose occurrence frequency is higher than the threshold value during the target number of training is obtained.

[0083] Furthermore, in the above-mentioned prediction device based on the clinical pathway model, the prediction module is specifically used for:

[0084] A difference between the predicted value and a normal predicted value determined by the feature information is determined, and the target feature data whose difference is within a preset range is determined as the key feature.

[0085] Furthermore, in the above-mentioned prediction device based on the clinical pathway model, the acquisition module includes:

[0086] a cleaning unit, configured to clean the feature information after merging it, and then perform exception processing on the feature information to determine missing features of missing data in the feature information;

[0087] The filling unit is used to fill the missing features with the historical data average value of the missing features, and then normalize the feature information.

[0088] Furthermore, the above prediction device based on the clinical pathway model, wherein the device also includes:

[0089] An input module, used to input the characteristic information and key characteristic data as input data into the blood glucose level prediction model to obtain a characteristic blood glucose value and a key blood glucose value;

[0090] Respectively obtaining the root mean square errors of the characteristic blood sugar value and the key blood sugar value, and comparing the root mean square errors of the characteristic blood sugar value and the key blood sugar value;

[0091] If the root mean square error between the characteristic blood glucose value and the key blood glucose value is within a preset range, it is determined that the key feature is used to accurately predict the blood glucose value of the diabetic patient.

[0092] Furthermore, in the above-mentioned prediction device based on the clinical pathway model, the acquisition module further comprises:

[0093] A search unit, used to first search for a target clinical pathway whose number of pathways is higher than a preset threshold from a preset clinical pathway database;

[0094] A selection unit is used to select a final clinical pathway with a number of person-times higher than a preset threshold from the target clinical pathway as the preset clinical pathway.

[0095] Furthermore, the above prediction device based on the clinical pathway model, wherein the device also includes:

[0096] The obtaining module is used to obtain the key feature data of the diabetic patient, and input the key feature data into the blood glucose level prediction model to obtain the blood glucose level of the diabetic patient.

[0097] The functions or operation steps implemented when the above modules are executed are substantially the same as those in the above method embodiments, and will not be repeated here.

[0098] Embodiment 4

[0099] Another aspect of the present invention further provides a readable storage medium having a computer program stored thereon, and when the program is executed by a processor, the steps of the method described in any one of the above-mentioned embodiments 1 to 2 are implemented.

[0100] Embodiment 5

[0101] Another aspect of the present invention provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in any one of the above-mentioned embodiments 1 to 2 when executing the program.

[0102] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0103] Those skilled in the art will appreciate that the logic and / or steps represented in the flowchart or otherwise described herein, for example, may be considered as an ordered list of executable instructions for implementing logical functions, and may be specifically implemented in any computer-readable storage medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For purposes of this specification, a "computer-readable storage medium" may be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0104] More specific examples (a non-exhaustive list) of computer-readable storage media include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable storage medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0105] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or a combination thereof: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0106] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" 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 present 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 may be combined in any one or more embodiments or examples in a suitable manner.

[0107] The above-described embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the attached claims.

Claims

1. A prediction method based on a clinical pathway model, characterized in that: The method comprises: Acquire characteristic information of the diabetic patients from diabetic patients who have used a preset clinical pathway, and determine a training data set based on the characteristic information; Establishing an LSTM neural network, and using the training data set to train the LSTM neural network to obtain a blood sugar value prediction model; Selecting a preset number of target feature data with a frequency higher than a threshold value during the training process of the blood glucose level prediction model; The step of selecting a preset number of target feature data with a frequency of occurrence higher than a threshold value during the training process of the blood sugar level prediction model comprises: Using the training data set to train the LSTM neural network for a preset number of times, obtaining each training result, determining a target number of times that the training result meets a preset condition from all training times, and obtaining a preset number of target feature data that appear with a frequency higher than a threshold value during the target number of trainings; The target feature data in the feature information are modified and input into the blood glucose level prediction model to obtain a predicted value, which is compared with a normal predicted value determined by the feature information, so as to determine a key feature for predicting the blood glucose level of the diabetic patient from the target feature data; The step of modifying the target feature data in the feature information and inputting the modified target feature data into the blood glucose level prediction model to obtain a predicted value and comparing it with a normal predicted value determined by the feature information to determine a key feature for predicting the blood glucose level of the diabetic patient from the target feature data comprises: Determine a difference between the predicted value and a normal predicted value determined by the feature information, and determine the target feature data whose difference is within a preset range as the key feature; The step of determining a training data set according to the feature information comprises: The characteristic information is merged and cleaned, and then the characteristic information is processed for exceptions to determine missing features of missing data in the characteristic information; Filling the missing features with the historical data average of the missing features, and then normalizing the feature information; After the step of determining the key features for predicting the blood glucose level of the diabetic patient from the target feature data, the following steps are further included: Inputting the characteristic information and key characteristic data as input data into the blood glucose level prediction model to obtain a characteristic blood glucose level and a key blood glucose level; Respectively obtaining root mean square errors of the characteristic blood sugar value and the key blood sugar value, and comparing the root mean square errors of the characteristic blood sugar value and the key blood sugar value; If the root mean square error between the characteristic blood glucose value and the key blood glucose value is within a preset range, it is determined that the key feature is used to accurately predict the blood glucose value of the diabetic patient.

2. The prediction method based on the clinical pathway model according to claim 1, characterized in that: The step of obtaining characteristic information of the diabetic patients from the diabetic patients who have used the preset clinical pathway comprises: First, find the target clinical pathway whose number of pathways is higher than the preset threshold from the preset clinical pathway database; A final clinical pathway with a number of visits higher than a preset threshold is selected from the target clinical pathway as the preset clinical pathway.

3. The prediction method based on the clinical pathway model according to any one of claims 1 to 2, characterized in that: After the step of determining the key features for predicting the blood glucose level of the diabetic patient from the target feature data, the following steps are further included: The key feature data of the diabetic patient are obtained, and the key feature data are input into the blood glucose level prediction model to obtain the blood glucose level of the diabetic patient.

4. A prediction device based on a clinical pathway model, characterized in that: For implementing the prediction method based on the clinical pathway model according to any one of claims 1 to 3, the device comprises: An acquisition module, used to acquire characteristic information of the diabetic patients from the diabetic patients who have used the preset clinical pathway, and determine a training data set according to the characteristic information; A training module, used to establish an LSTM neural network, and train the LSTM neural network using the training data set to obtain a blood sugar value prediction model; A selection module, used for selecting a preset number of target feature data with an occurrence frequency higher than a threshold value during the training process of the blood glucose level prediction model; A prediction module is used to modify the target feature data in the feature information and input the resulting predicted value into the blood glucose value prediction model, and compare it with the normal predicted value determined by the feature information, so as to determine the key features for predicting the blood glucose value of the diabetic patient from the target feature data.

5. A readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

6. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 3 when executing the program.

Images