Health management plan generation method based on intelligent voice interaction

Abstract

The present invention relates to the field of medical big data technology, and in particular to a method for generating a health management plan based on intelligent voice interaction, comprising: extracting medical diagnosis and rehabilitation feature information based on historical medical big data, preprocessing the medical diagnosis and rehabilitation feature information, determining a preset period based on a data continuity evaluation value of the medical diagnosis and rehabilitation feature information, filling in missing values ​​of first structured feature information, dividing a data set based on a time series according to a preset ratio, training a time series prediction model, and determining whether the time series prediction model is qualified based on the accuracy of the time series prediction model, using voice interaction to obtain a user's historical basic medical information, and generating a health management plan for the user based on the time series prediction model. The present invention overcomes the problem in the prior art of lacking an efficient and accurate preprocessing method for medical diagnosis and rehabilitation feature information, resulting in poor accuracy of the time series prediction model in generating a health management plan.

Description

Technical Field

[0001] The present invention relates to the field of medical big data technology, and in particular to a method for generating a health management plan based on intelligent voice interaction. Background Art

[0002] Health management is an important issue in daily life and medical care. Health management is especially important for people with chronic diseases, elderly diseases or those who need long-term rehabilitation treatment. Traditional health management methods mainly rely on medical staff to provide individualized guidance and management for patients, but due to limited medical staff resources, it is often impossible to provide comprehensive and detailed management for each patient. Moreover, traditional health management methods usually cannot obtain patients' health status information in real time, and the tracking and management of patients' health status is not timely enough.

[0003] In order to solve the above problems, a health management method based on big data and artificial intelligence technology has emerged in recent years. This method collects and analyzes the patient's health data to generate personalized health management plans, thereby achieving real-time and accurate health management of patients. However, existing methods usually require a large amount of medical data, and in practical applications, patients' medical data are often incomplete, which requires preprocessing of incomplete data to improve the effect of health management. How to efficiently preprocess incomplete medical data and how to generate personalized health management plans based on the preprocessed data are important issues currently faced.

[0004] Chinese Patent Publication No.: CN114864038A discloses a method and device for generating a single disease health management plan based on big data, including: extracting multi-dimensional data features from medical big data to obtain first feature information; constructing a medical behavior time series model based on the first feature information; extracting multi-dimensional data features from big data for implementation of health management tasks to obtain second feature information; constructing a health management task time series model based on the second feature information; fitting the medical behavior time series model and the health management task time series model to obtain an optimized health management implementation time series model; inputting relevant information of the patient into the optimized health management implementation time series model to obtain a health management plan for the patient;

[0005] It can be seen that the existing technology has the following problems: due to the lack of efficient and accurate preprocessing methods for the medical diagnosis and rehabilitation feature information obtained after data feature extraction of historical medical big data, the accuracy of the health management plan generated by the time series prediction model is poor. Summary of the invention

[0006] To this end, the present invention provides a health management plan generation method based on intelligent voice interaction, which is used to overcome the problem that the prior art lacks efficient and accurate preprocessing methods for medical diagnosis and rehabilitation feature information, resulting in poor accuracy of health management plans generated by time series prediction models.

[0007] To achieve the above object, the present invention provides a method for generating a health management plan based on intelligent voice interaction, comprising:

[0008] Step S1, extracting multi-dimensional data features based on historical medical big data to obtain medical diagnosis and rehabilitation feature information;

[0009] Step S2, determining a preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information according to the data type of the medical diagnosis and rehabilitation feature information;

[0010] Step S3, when it is determined that the medical diagnosis and rehabilitation feature information is preprocessed in the first preprocessing manner, determining a preset period according to the data continuity evaluation value of the medical diagnosis and rehabilitation feature information;

[0011] Step S4, when it is determined that the preprocessing of the medical diagnosis and rehabilitation feature information in the first preprocessing manner is completed, determining a filling method for filling the missing values ​​of the first structured feature information according to the time interval;

[0012] Step S5, combining the first structured feature information and the second structured feature information to obtain a data set based on a time series, dividing the data set into a training set and a test set according to a preset ratio, and training the time series prediction model on the training set;

[0013] Step S6, determining whether the time series prediction model is qualified according to the accuracy of the trained time series prediction model on the test set;

[0014] Step S7, when the user's historical basic medical information is obtained through voice interaction, a qualified time series prediction model generates a health management plan for the user within a preset time period in the future according to the historical basic medical information;

[0015] Among them, in step S6, when the time series prediction model is unqualified, it is determined to adjust the preset data continuity evaluation value, and the adjustment method of the preset data continuity evaluation value is determined according to the relative difference between the accuracy rate and the preset accuracy rate.

[0016] Further, determining a preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information according to the data type of the medical diagnosis and rehabilitation feature information;

[0017] When the data type is numerical data, determining that the preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information is the first preprocessing method;

[0018] When the data type is non-numeric data, determining that the preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information is the second preprocessing method;

[0019] Among them, the first preprocessing method is to sample the medical diagnosis and rehabilitation feature information at a preset period to obtain first structured feature information, and the second preprocessing method is to perform one-hot encoding on the medical diagnosis and rehabilitation feature information to obtain second structured feature information.

[0020] Further, when it is determined that the medical diagnosis and rehabilitation feature information is preprocessed in the first preprocessing manner, the preset period is determined according to the data continuity evaluation value of the medical diagnosis and rehabilitation feature information;

[0021] The data continuity evaluation value of the medical diagnosis and rehabilitation feature information is calculated according to the following formula:

[0022]

[0023] Among them, P represents the data continuity evaluation value, Ai represents the time interval between the medical diagnosis and rehabilitation feature information at the i-th moment and the medical diagnosis and rehabilitation feature information at the i+1 moment, the unit is day, i≥1, n represents the total number of medical diagnosis and rehabilitation feature information, and A0 represents the average time interval.

[0024] Further, when it is determined that the preprocessing of the medical diagnosis and rehabilitation feature information in the first preprocessing manner is completed, several filling methods for filling the missing values ​​of the first structured feature information are determined according to the time interval, and the several filling methods include a first filling method of using the medical diagnosis and rehabilitation feature information closest to the time when the missing value is located as the missing value and a second filling method of taking the average of the first structured feature information closest to the time when the missing value is located as the missing value;

[0025] The time interval is the time interval between the moment when the missing value of the first structured feature information is located and the nearest moment in the medical diagnosis and rehabilitation feature information.

[0026] Furthermore, after the training of the time series prediction model is completed, the accuracy of the trained time series prediction model on the test set is obtained, so that when the accuracy is greater than a preset accuracy rate, the time series prediction model is determined to be qualified; when the accuracy is less than or equal to the preset accuracy rate, the time series prediction model is determined to be unqualified, and the preset data continuity evaluation value is determined to be adjusted.

[0027] Further, when determining to adjust the preset data continuity evaluation value, several adjustment methods for adjusting the preset data continuity evaluation value are determined according to the relative difference between the accuracy rate and the preset accuracy rate, and the several adjustment methods include a first adjustment method for adjusting the second preset data continuity evaluation value with a first adjustment coefficient and a second adjustment method for adjusting the first preset data continuity evaluation value with a second adjustment coefficient.

[0028] Further, when it is determined to adjust the preset data continuity evaluation value in the first adjustment manner, the second preset data continuity evaluation value is adjusted according to the following first adjustment coefficient, setting

[0029]

[0030] Wherein, X1 represents the first adjustment coefficient, and ΔL represents the relative difference.

[0031] Further, when it is determined to adjust the preset data continuity evaluation value in the second adjustment manner, the first preset data continuity evaluation value is adjusted according to the following second adjustment coefficient, setting

[0032]

[0033] Wherein, X2 represents the second adjustment coefficient.

[0034] Further, after the preset data continuity evaluation value is adjusted, the percentage of the number of missing values ​​in the first structured feature information to the total number of the first structured feature information is obtained, so as to determine the correction of the preset time interval when the percentage is greater than the preset percentage.

[0035] Further, when determining to modify the preset time interval, the preset time interval is modified according to the following correction coefficient:

[0036]

[0037] Wherein, Z represents the correction coefficient, C represents the percentage, and C0 represents the preset percentage.

[0038] Compared with the prior art, the beneficial effect of the present invention lies in that the present invention adopts different preprocessing methods for different types of data, which can make the preprocessing more targeted and better retain the characteristics and information of the original data, thereby improving the accuracy of subsequent model training and prediction, sampling or one-hot encoding the medical diagnosis and rehabilitation feature information to convert the originally irregular or difficult to process data into structured data, which is convenient for subsequent data processing and analysis, sampling the numerical data to enhance the continuity and consistency of the data, thereby improving the prediction effect of the time series prediction model, and at the same time, one-hot encoding is performed on the non-numeric data to convert the discrete categorical variables into continuous numerical variables, making the model easier to process and learn.

[0039] Furthermore, the present invention selects the sampling period based on the data continuity evaluation value of the medical diagnosis and rehabilitation characteristic information, which is more flexible than a fixed sampling period, thereby improving the accuracy of sampling, using a shorter sampling period to obtain more information, and using a longer sampling period to reduce the volatility and instability of the data.

[0040] Furthermore, the present invention fills in missing values ​​to increase the integrity of the data, thereby providing a more accurate and complete data basis for subsequent model training, and uses the closest medical diagnosis and rehabilitation feature information as the missing value to better reflect the patient's actual health status in the time period. When the time interval is long, the mean is used as the missing value to avoid errors caused by individual deviations, thereby improving the generalization ability and accuracy of the model.

[0041] Furthermore, the present invention adjusts the preset data continuity evaluation value according to the accuracy of the time series prediction model, which not only improves the flexibility of model training, but also enables the time series prediction model to be optimized and updated in real time to improve the accuracy of the model.

[0042] Furthermore, the present invention determines the adjustment method of the preset data continuity evaluation value according to the relative difference between the accuracy rate and the preset accuracy rate, thereby improving the flexibility of data preprocessing and model training, and improving the accuracy of the model on the test set.

[0043] Furthermore, the present invention improves the accuracy of the model on the test set by dynamically adjusting the preset data continuity evaluation value, and dynamically adjusts the preset data continuity evaluation value according to the difference between the model's accuracy and the preset accuracy, so as to reduce unnecessary consumption of computing and storage resources while ensuring the model performance.

[0044] Furthermore, the present invention adjusts the preset time interval in real time according to the percentage of the number of missing values ​​in the data to the total number, so as to reduce data missingness and improve data integrity, thereby making model training and prediction more accurate. When there are fewer missing values ​​in the data, unnecessary corrections to the preset time interval are not performed, which reduces the complexity of data processing and saves computing and storage resources. By dynamically adjusting the preset time interval, the missing values ​​in the data can be processed in a targeted manner, the quality of the data can be improved, and the prediction performance of the model can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A flowchart of a method for generating a health management plan based on intelligent voice interaction according to an embodiment of the present invention;

[0046] Figure 2 A logic block diagram of the preprocessing process in the method for generating a health management plan based on intelligent voice interaction according to an embodiment of the present invention;

[0047] Figure 3 This is a logic block diagram of the adjustment process in the health management plan generation method based on intelligent voice interaction in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0049] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0050] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0051] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0052] See also Figure 1-Figure 3 As shown, Figure 1 A flowchart of a method for generating a health management plan based on intelligent voice interaction according to an embodiment of the present invention; Figure 2 A logic block diagram of the preprocessing process in the method for generating a health management plan based on intelligent voice interaction according to an embodiment of the present invention; Figure 3 This is a logic block diagram of the adjustment process in the health management plan generation method based on intelligent voice interaction in an embodiment of the present invention.

[0053] The method for generating a health management plan based on intelligent voice interaction in an embodiment of the present invention includes:

[0054] Step S1, extracting multi-dimensional data features based on historical medical big data to obtain medical diagnosis and rehabilitation feature information;

[0055] Step S2, determining a preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information according to the data type of the medical diagnosis and rehabilitation feature information;

[0056] Step S3, when it is determined that the medical diagnosis and rehabilitation feature information is preprocessed in the first preprocessing manner, determining a preset period according to the data continuity evaluation value P of the medical diagnosis and rehabilitation feature information;

[0057] Step S4, when it is determined that the preprocessing of the medical diagnosis and rehabilitation feature information in the first preprocessing manner is completed, determining a filling method for filling the missing values ​​of the first structured feature information according to the time interval B;

[0058] Step S5, combining the first structured feature information and the second structured feature information to obtain a data set based on a time series, dividing the data set into a training set and a test set according to a preset ratio, and training the time series prediction model on the training set;

[0059] Step S6, determining whether the time series prediction model is qualified according to the accuracy L of the trained time series prediction model on the test set;

[0060] Step S7, when the user's historical basic medical information is obtained through voice interaction, a qualified time series prediction model generates a health management plan for the user within a future preset time period T based on the historical basic medical information.

[0061] Among them, in the step S6, when the time series prediction model is unqualified, it is determined to adjust the preset data continuity evaluation value, and the adjustment method for adjusting the preset data continuity evaluation value is determined according to the relative difference ΔL between the accuracy L and the preset accuracy L0.

[0062] In an embodiment of the present invention, the preset ratio is preferably 8:2, the time series prediction model is preferably an LSTM model, the preset time period T is preferably 3 days, the medical diagnosis and rehabilitation feature information includes the patient's age, gender, disease diagnosis, medical behavior records, sleep, basic physical parameters, and health management task implementation records, and the historical basic medical information includes the user's age, gender, disease diagnosis, medical behavior records, sleep, and basic physical parameters.

[0063] Specifically, a preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information is determined according to the data type of the medical diagnosis and rehabilitation feature information;

[0064] When the data type is numerical data, determining that the preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information is the first preprocessing method;

[0065] When the data type is non-numeric data, determining that the preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information is the second preprocessing method;

[0066] Among them, the first preprocessing method is to sample the medical diagnosis and rehabilitation feature information at a preset period to obtain first structured feature information, and the second preprocessing method is to perform one-hot encoding on the medical diagnosis and rehabilitation feature information to obtain second structured feature information.

[0067] Specifically, the present invention adopts different preprocessing methods for different types of data, which can make the preprocessing more targeted and better retain the characteristics and information of the original data, thereby improving the accuracy of subsequent model training and prediction, and sampling or one-hot encoding the medical diagnosis and rehabilitation feature information to convert the originally irregular or difficult to process data into structured data, which is convenient for subsequent data processing and analysis. The numerical data is sampled to enhance the continuity and consistency of the data, thereby improving the prediction effect of the time series prediction model. At the same time, the non-numeric data is one-hot encoded to convert discrete categorical variables into continuous numerical variables, making the model easier to process and learn.

[0068] Specifically, when it is determined that the medical diagnosis and rehabilitation feature information is preprocessed in a first preprocessing manner, the preset period is determined according to a comparison result of a data continuity evaluation value P of the medical diagnosis and rehabilitation feature information and a preset data continuity evaluation value, wherein the preset data continuity evaluation value includes a first preset data continuity evaluation value P1 and a second preset data continuity evaluation value P2, wherein P1<P2;

[0069] When P>P2, the preset period is determined to be the first preset period;

[0070] When P2≥P>P1, the preset period is determined to be a second preset period;

[0071] When P1≥P, the preset period is determined to be the third preset period;

[0072] Among them, the first preset period is 1 day, the second preset period is 3 days, and the third preset period is 7 days.

[0073] In the embodiment of the present invention, the first preset data continuity evaluation value P1 is 0.6, and the second preset data continuity evaluation value P2 is 0.9. Those skilled in the art may adjust the first preset data continuity evaluation value P1 and the second preset data continuity evaluation value P2 according to specific circumstances.

[0074] Specifically, the present invention selects the sampling period based on the data continuity evaluation value of the medical diagnosis and rehabilitation characteristic information, which is more flexible than a fixed sampling period, thereby improving the accuracy of sampling, using a shorter sampling period to obtain more information, and using a longer sampling period to reduce the volatility and instability of the data.

[0075] Specifically, the data continuity evaluation value P of the medical diagnosis and rehabilitation feature information is calculated according to the following formula:

[0076]

[0077] Among them, Ai represents the time interval between the medical diagnosis and rehabilitation feature information at the i-th moment and the medical diagnosis and rehabilitation feature information at the i+1 moment, the unit is day, i≥1, n represents the total number of medical diagnosis and rehabilitation feature information, and A0 represents the average time interval.

[0078] Specifically, when it is determined that the preprocessing of the medical diagnosis and rehabilitation feature information is completed in the first preprocessing manner, a filling method for filling the missing values ​​of the first structured feature information is determined according to a comparison result of the time interval B and the preset time interval B0;

[0079] When B≤B0, it is determined to fill in the missing values ​​of the first structural feature information in the first gap-filling manner;

[0080] When B>B0, it is determined to fill in the missing values ​​of the first structural feature information in the second gap-filling manner;

[0081] The time interval B is the time interval between the time when the missing value of the first structured feature information is located and the nearest time in the medical diagnosis and rehabilitation feature information;

[0082] The first method of filling in the gaps is to use the medical diagnosis and rehabilitation feature information closest to the time when the missing value is located as the missing value, and the second method of filling in the gaps is to take the average of the first structured feature information closest to the time when the missing value is located as the missing value.

[0083] In the embodiment of the present invention, the preset time interval B0 is set to 2 days, and those skilled in the art may adjust the preset time interval B0 according to specific circumstances.

[0084] Specifically, the present invention fills in missing values ​​to increase the integrity of the data, thereby providing a more accurate and complete data basis for subsequent model training, and uses the closest medical diagnosis and rehabilitation feature information as the missing value to better reflect the patient's actual health status in the time period. When the time interval is long, the mean is used as the missing value to avoid errors caused by individual deviations, thereby improving the generalization ability and accuracy of the model.

[0085] Specifically, after the training of the time series prediction model is completed, whether the time series prediction model is qualified is determined according to the comparison result of the accuracy L of the trained time series prediction model on the test set and the preset accuracy L0;

[0086] When L≤L0, it is determined that the time series prediction model is unqualified, and the preset data continuity evaluation value is adjusted;

[0087] When L>L0, it is determined that the time series prediction model is qualified;

[0088] In the embodiment of the present invention, the preset accuracy rate L0 is set to 95%, and those skilled in the art may adjust the preset accuracy rate L0 according to specific circumstances.

[0089] Specifically, the present invention adjusts the preset data continuity evaluation value according to the accuracy of the time series prediction model, which not only improves the flexibility of model training, but also enables the time series prediction model to be optimized and updated in real time to improve the accuracy of the model.

[0090] Specifically, when it is determined to adjust the preset data continuity evaluation value, the relative difference △L between the accuracy rate L and the preset accuracy rate L0 is calculated, and the adjustment method for adjusting the preset data continuity evaluation value is determined according to the comparison result of the relative difference △L and the preset relative difference △L0, and △L=(L0-L) / L0 is set;

[0091] When △L≤△L0, it is determined to adjust the preset data continuity evaluation value in a first adjustment manner;

[0092] When △L>△L0, it is determined to adjust the preset data continuity evaluation value in the second adjustment manner;

[0093] The first adjustment method is to adjust the second preset data continuity evaluation value P2 with the first adjustment coefficient X1, and the second adjustment method is to adjust the first preset data continuity evaluation value P1 with the second adjustment coefficient X2.

[0094] In the embodiment of the present invention, the preset relative difference △L0 is set to 0.05. The preset relative difference △L0 is obtained when the accuracy L is 90%. Those skilled in the art can adjust the preset relative difference △L0 according to specific circumstances.

[0095] Specifically, the present invention determines the adjustment method of the preset data continuity evaluation value according to the relative difference between the accuracy rate and the preset accuracy rate, thereby improving the flexibility of data preprocessing and model training, and improving the accuracy of the model on the test set.

[0096] Specifically, when it is determined to adjust the preset data continuity evaluation value in the first adjustment manner, the second preset data continuity evaluation value P2 is adjusted according to the following first adjustment coefficient X1, and the setting

[0097]

[0098] The adjusted second preset data continuity evaluation value is set to P21=P2×X1.

[0099] Specifically, when it is determined to adjust the preset data continuity evaluation value in the second adjustment manner, the first preset data continuity evaluation value P1 is adjusted according to the following second adjustment coefficient X2, setting

[0100]

[0101] The adjusted first preset data continuity evaluation value is set to P12=P1×X2.

[0102] Specifically, the present invention improves the accuracy of the model on the test set by dynamically adjusting the preset data continuity evaluation value, and dynamically adjusts the preset data continuity evaluation value according to the difference between the model's accuracy and the preset accuracy, so as to reduce unnecessary consumption of computing and storage resources while ensuring model performance.

[0103] Specifically, after the preset data continuity evaluation value is adjusted, it is determined whether to correct the preset time interval B0 according to the comparison result of the percentage C of the number of missing values ​​in the first structured feature information to the total number of the first structured feature information and the preset percentage C0;

[0104] When C≤C0, it is determined that the preset time interval B0 is not corrected;

[0105] When C>C0, the preset time interval B0 is determined to be corrected;

[0106] In the embodiment of the present invention, the preset percentage C0 is set to 30%, and those skilled in the art may adjust the preset percentage C0 according to specific circumstances.

[0107] Specifically, the present invention adjusts the preset time interval in real time according to the percentage of the number of missing values ​​in the data to the total number, so as to reduce data missingness and improve data integrity, thereby making model training and prediction more accurate. When there are fewer missing values ​​in the data, unnecessary corrections to the preset time interval are not performed, which reduces the complexity of data processing and saves computing and storage resources. By dynamically adjusting the preset time interval, the missing values ​​in the data can be processed in a targeted manner, the quality of the data can be improved, and the prediction performance of the model can be improved.

[0108] Specifically, when determining to modify the preset time interval B0, the preset time interval B0 is modified according to the following correction coefficient Z, setting

[0109]

[0110] Set the modified preset duration interval to

[0111] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0112] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for generating a health management plan based on intelligent voice interaction, characterized in that: include: Step S1, extracting multi-dimensional data features based on historical medical big data to obtain medical diagnosis and rehabilitation feature information; Step S2, determining a preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information according to the data type of the medical diagnosis and rehabilitation feature information; Step S3, when it is determined that the medical diagnosis and rehabilitation feature information is preprocessed in the first preprocessing manner, sampling the medical diagnosis and rehabilitation feature information at a preset period is determined according to the data continuity evaluation value of the medical diagnosis and rehabilitation feature information to obtain first structured feature information; The data continuity evaluation value of the medical diagnosis and rehabilitation feature information is calculated according to the following formula: Wherein, P represents the data continuity evaluation value, Ai represents the time interval between the medical diagnosis and rehabilitation feature information at the i-th moment and the medical diagnosis and rehabilitation feature information at the i+1 moment, in days, i≥1, n represents the total number of medical diagnosis and rehabilitation feature information, and A0 represents the average time interval; Step S4, when it is determined that the preprocessing of the medical diagnosis and rehabilitation feature information in the first preprocessing manner is completed, a filling method for filling the missing values ​​of the first structured feature information is determined according to the time interval, wherein the time interval is the time interval between the moment when the missing value of the first structured feature information is located and the nearest moment in the medical diagnosis and rehabilitation feature information; Step S5, merging the first structured feature information and the second structured feature information to obtain a data set based on a time series, dividing the data set into a training set and a test set according to a preset ratio, training the time series prediction model on the training set, wherein the medical diagnosis and rehabilitation feature information is uniquely encoded to obtain the second structured feature information; Step S6, determining whether the time series prediction model is qualified according to the accuracy of the trained time series prediction model on the test set; Step S7, when the user's historical basic medical information is obtained through voice interaction, a qualified time series prediction model generates a health management plan for the user within a preset time period in the future according to the historical basic medical information; Among them, in step S6, when the time series prediction model is unqualified, the preset data continuity evaluation value is determined to be adjusted, and the adjustment method of the preset data continuity evaluation value is determined according to the relative difference between the accuracy rate and the preset accuracy rate. The preset data continuity evaluation value includes a first preset continuity evaluation value, whose value is 0.6, and a second preset data continuity evaluation value, whose value is 0.

9.

2. The method for generating a health management plan based on intelligent voice interaction according to claim 1, characterized in that: Determining a preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information according to the data type of the medical diagnosis and rehabilitation feature information; When the data type is numerical data, determining that the preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information is the first preprocessing method; When the data type is non-numeric data, it is determined that the preprocessing method for preprocessing the medical diagnosis and rehabilitation feature information is the second preprocessing method.

3. The method for generating a health management plan based on intelligent voice interaction according to claim 2, characterized in that: When it is determined that the preprocessing of the medical diagnosis and rehabilitation feature information in the first preprocessing method is completed, several filling methods for filling the missing values ​​of the first structured feature information are determined according to the time interval, and the several filling methods include a first filling method of using the medical diagnosis and rehabilitation feature information closest to the time when the missing value is located as the missing value and a second filling method of taking the average of the first structured feature information closest to the time when the missing value is located as the missing value.

4. The method for generating a health management plan based on intelligent voice interaction according to claim 3, characterized in that: When the training of the time series prediction model is completed, the accuracy of the trained time series prediction model on the test set is obtained, so that when the accuracy is greater than the preset accuracy, the time series prediction model is determined to be qualified; when the accuracy is less than or equal to the preset accuracy, the time series prediction model is determined to be unqualified, and the preset data continuity evaluation value is determined to be adjusted.

5. The method for generating a health management plan based on intelligent voice interaction according to claim 4, characterized in that: When determining to adjust the preset data continuity evaluation value, several adjustment methods for adjusting the preset data continuity evaluation value are determined according to the relative difference between the accuracy rate and the preset accuracy rate, and the several adjustment methods include a first adjustment method for adjusting the second preset data continuity evaluation value with a first adjustment coefficient and a second adjustment method for adjusting the first preset data continuity evaluation value with a second adjustment coefficient.

6. The method for generating a health management plan based on intelligent voice interaction according to claim 5, characterized in that: When it is determined to adjust the preset data continuity evaluation value in the first adjustment manner, the second preset data continuity evaluation value is adjusted according to the following first adjustment coefficient: Wherein, X1 represents the first adjustment coefficient, and ΔL represents the relative difference.

7. The method for generating a health management plan based on intelligent voice interaction according to claim 6, characterized in that: When it is determined to adjust the preset data continuity evaluation value in the second adjustment manner, the first preset data continuity evaluation value is adjusted according to the following second adjustment coefficient: Wherein, X2 represents the second adjustment coefficient.

8. The method for generating a health management plan based on intelligent voice interaction according to claim 7, characterized in that: After the preset data continuity evaluation value is adjusted, the percentage of the number of missing values ​​in the first structured feature information to the total number of the first structured feature information is obtained, so as to determine the correction of the preset time interval when the percentage is greater than the preset percentage.

9. The method for generating a health management plan based on intelligent voice interaction according to claim 8, characterized in that: When determining to modify the preset time interval, modify the preset time interval according to the following correction coefficient: Wherein, Z represents the correction coefficient, C represents the percentage, and C0 represents the preset percentage.

Images