Postoperative hearing and speech rehabilitation effect prediction method and device for cochlear implant patients

Abstract

The embodiment of the application discloses a kind of postoperative hearing speech rehabilitation effect prediction method and device of cochlear implant patient, the method comprises: obtaining the original information data of target patient to be predicted, the original information data of target patient includes the cochlea type of the target patient, preoperative information and postoperative information;In the case where the cochlea type is normal cochlea, the original information data of the target patient is input into the first prediction model trained in advance, i.e. the first prediction result for normal cochlea can be obtained;In the case where the cochlea type is abnormal cochlea, the original information data of the target patient is input into the second prediction model trained in advance, i.e. the second prediction result for abnormal cochlea can be obtained.The technical problem that the prediction result accuracy is poor caused by the fact that the postoperative hearing speech rehabilitation effect of cochlear implant patient cannot be predicted based on big data in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, specifically to a method and device for predicting the postoperative hearing and speech rehabilitation effects of cochlear implant patients based on machine learning. Background Technology

[0002] Currently, a vast amount of data has accumulated in the field of cochlear implantation, but there are few reports, both domestically and internationally, on machine learning-based predictive models for postoperative outcomes in cochlear implant patients. In clinical practice, a universally applicable machine learning-based predictive model for postoperative outcomes in cochlear implant patients has not yet been established. To fully utilize the data in the field of cochlear implantation and transform it into meaningful resources, constructing a machine learning-based predictive model for postoperative outcomes in cochlear implant patients is crucial.

[0003] However, at present, the recovery effect of cochlear implant patients can only be predicted based on the doctor's personal experience. The prediction results will be subject to the doctor's personal subjectivity and may be wrong due to fatigue. Summary of the Invention

[0004] To address this, embodiments of the present invention provide a machine learning-based method for predicting the postoperative hearing and speech rehabilitation outcomes of cochlear implant patients, thereby at least partially resolving the technical problem in existing technologies where the prediction of postoperative hearing and speech rehabilitation outcomes in cochlear implant patients cannot be based on big data, resulting in poor prediction accuracy. This objective is achieved through the following technical solution:

[0005] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0006] This invention provides a method for predicting the postoperative hearing and speech rehabilitation effect in patients with cochlear implantation, the method comprising:

[0007] Obtain the original information data of the target patient to be predicted, including the cochlear type, preoperative information and postoperative information of the target patient;

[0008] In the case where the cochlear type is normal, the original information data of the target patient is input into a pre-trained first prediction model to obtain a first prediction result for normal cochlear.

[0009] In the case where the cochlear type is malformed cochlea, the original information data of the target patient is input into a pre-trained second prediction model to obtain a second prediction result for malformed cochlea;

[0010] The first prediction model is obtained by training a machine learning model using a pre-built normal cochlear dataset, which is composed of the original information data of normal cochlear patient samples.

[0011] The second prediction model is obtained by training a machine learning model using a pre-built dataset of abnormal cochleas, which consists of raw information data of samples from patients with abnormal cochleas.

[0012] In some embodiments, the training of the first prediction model and the second prediction model includes:

[0013] The raw information data of all patient samples within a previously preset time period are obtained, and the raw information data of the patient samples are integrated based on the cochlear type to obtain the normal cochlear dataset composed of the raw information data of normal cochlear patient samples, and the abnormal cochlear dataset composed of the raw information data of abnormal cochlear patient samples.

[0014] The original information data of the patient sample includes the cochlear type, preoperative information, and postoperative information of the patient sample.

[0015] In some embodiments, the first prediction model, obtained by training a machine learning model using a pre-built normal cochlear dataset, specifically includes:

[0016] The normal cochlear dataset is preprocessed;

[0017] In the preprocessed normal cochlear dataset, the correlation between the preoperative information and the postoperative information of the normal cochlear patient samples is calculated to obtain the first correlation result;

[0018] Sort the first relevance results from high to low according to feature importance;

[0019] The dataset consisting of the first k1 sorted features is input into a pre-built machine learning model to train the first prediction model.

[0020] In some embodiments, preprocessing of the normal cochlear dataset specifically includes:

[0021] The original information data of the patient samples is traversed, data containing missing values ​​and irrelevant variables are deleted, and the remaining continuous variables after deletion are normalized.

[0022] In some embodiments, the correlation between preoperative information and postoperative information is calculated using the chi-square test and the Spearman correlation algorithm.

[0023] In some embodiments, the second prediction model, obtained by training a machine learning model using a pre-built dataset of abnormal cochleas, specifically includes:

[0024] The dataset of deformed cochleas is preprocessed;

[0025] In the preprocessed cochlear deformity dataset, the correlation between the preoperative information and the postoperative information of the cochlear deformity patient samples is calculated to obtain a second correlation result;

[0026] The second relevance results are sorted by feature importance from high to low;

[0027] The dataset consisting of the first k2 sorted features is input into a pre-built machine learning model to train the second prediction model.

[0028] In some embodiments, the preprocessing of the abnormal cochlear dataset specifically includes:

[0029] The original information data of the patient samples is traversed, irrelevant variables are removed, and the remaining discrete variables are one-hot encoded; the original information data of the patients is traversed, and missing values ​​are imputed; the remaining continuous variables after missing value imputed are normalized.

[0030] The present invention also provides a device for predicting the postoperative hearing and speech rehabilitation effect in patients with cochlear implantation, the device comprising:

[0031] The data acquisition unit is used to acquire the original information data of the target patient to be predicted, including the cochlear type, preoperative information and postoperative information of the target patient.

[0032] The first result output unit is used to input the original information data of the target patient into a pre-trained first prediction model when the cochlear type is normal cochlea, so as to obtain a first prediction result for normal cochlea.

[0033] The second result output unit is used to input the original information data of the target patient into a pre-trained second prediction model when the cochlear type is malformed cochlea, so as to obtain a second prediction result for malformed cochlea.

[0034] The first prediction model is obtained by training a machine learning model using a pre-built normal cochlear dataset, which is composed of the original information data of normal cochlear patient samples.

[0035] The second prediction model is obtained by training a machine learning model using a pre-built dataset of abnormal cochleas, which consists of raw information data of samples from patients with abnormal cochleas.

[0036] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method described above.

[0037] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method described above.

[0038] The present invention provides a method for predicting the postoperative hearing and speech rehabilitation effect of cochlear implant patients. This method acquires the original information data of the target patient, including the cochlear type, preoperative information, and postoperative information. If the cochlear type is normal, the original information data is input into a pre-trained first prediction model to obtain a first prediction result for normal cochlea. If the cochlear type is malformed, the original information data is input into a pre-trained second prediction model to obtain a second prediction result for malformed cochlea.

[0039] This approach applies machine learning to predict the postoperative hearing and speech rehabilitation outcomes of cochlear implant patients, using it as a tool to assist doctors in prediction. This avoids predictions being limited by human experience and provides more objective and accurate results. It solves the technical problem in existing technologies where the inability to predict postoperative hearing and speech rehabilitation outcomes of cochlear implant patients based on big data leads to poor prediction accuracy. Attached Figure Description

[0040] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0041] The structures, proportions, sizes, etc. illustrated in this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed herein, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0042] Figure 1 This is one of the flowcharts for the method of predicting the postoperative hearing and speech rehabilitation effect of cochlear implant patients provided by the present invention;

[0043] Figure 2 The second flowchart is a method for predicting the postoperative hearing and speech rehabilitation effect of cochlear implant patients provided by the present invention.

[0044] Figure 3 The third flowchart of the method for predicting the postoperative hearing and speech rehabilitation effect of cochlear implant patients provided by the present invention;

[0045] Figure 4 This is one of the illustrations of the correlation results in one embodiment;

[0046] Figure 5 This is the second illustration of the correlation results in one embodiment;

[0047] Figure 6 This is a structural block diagram of the device for predicting the postoperative hearing and speech rehabilitation effect of cochlear implant patients provided by the present invention.

[0048] Figure 7 This is a schematic diagram of the physical structure of the electronic device provided by the present invention. Detailed Implementation

[0049] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Please refer to Figure 1 , Figure 1 This is one of the flowcharts for the method of predicting the postoperative hearing and speech rehabilitation effect of cochlear implant patients provided by the present invention.

[0051] In one specific embodiment, the method for predicting the postoperative hearing and speech rehabilitation effect in cochlear implant patients provided by the present invention includes the following steps:

[0052] S110: Obtain the original information data of the target patient to be predicted, wherein the original information data of the target patient includes the cochlear type, preoperative information and postoperative information of the target patient;

[0053] S120: When the cochlear type is normal cochlea, the original information data of the target patient is input into a pre-trained first prediction model to obtain a first prediction result for normal cochlea; wherein, the first prediction model is obtained by training a machine learning model using a pre-built normal cochlear dataset, and the normal cochlear dataset is composed of the original information data of normal cochlear patient samples.

[0054] S130: In the case that the cochlear type is malformed cochlea, the original information data of the target patient is input into a pre-trained second prediction model to obtain a second prediction result for malformed cochlea; wherein, the second prediction model is obtained by training a machine learning model using a pre-constructed malformed cochlea dataset, the malformed cochlea dataset being composed of the original information data of malformed cochlea patient samples.

[0055] In some embodiments, the training of the first prediction model and the second prediction model includes:

[0056] The raw information data of all patient samples within a previously preset time period are obtained, and the raw information data of the patient samples are integrated based on the cochlear type to obtain the normal cochlear dataset composed of the raw information data of normal cochlear patient samples, and the abnormal cochlear dataset composed of the raw information data of abnormal cochlear patient samples.

[0057] The original information data of the patient sample includes the cochlear type, preoperative information, and postoperative information of the patient sample.

[0058] The integration of the patient's original information data specifically includes:

[0059] Preprocessing of the original normal cochlear dataset: Iterate through the original patient information data, delete data containing missing values ​​and remove irrelevant variables; normalize the remaining continuous variables after deletion.

[0060] The original cochlear malformation dataset is preprocessed as follows: the original patient information data is traversed, irrelevant variables are removed, and the remaining discrete variables are one-hot encoded; the original patient information data is traversed again, and missing values ​​are imputed; the remaining continuous variables after missing value imputation are normalized.

[0061] In some embodiments, the first prediction model is obtained by training a machine learning model using a pre-built normal cochlear dataset, such as... Figure 2 As shown, the specific steps include:

[0062] S210: Preprocess the normal cochlear dataset; specifically, traverse the original information data of the patient samples, delete data containing missing values ​​and remove irrelevant variables, and normalize the remaining continuous variables after deletion.

[0063] S220: In the preprocessed normal cochlear dataset, calculate the correlation between the preoperative information and the postoperative information of the normal cochlear patient samples to obtain a first correlation result; for example, the chi-square test and Spearman correlation algorithm can be used to calculate the correlation between the preoperative information and the postoperative information.

[0064] S230: Sort the first relevance results from high to low according to feature importance;

[0065] S240: Input the dataset consisting of the first k1 sorted features into a pre-built machine learning model to train the first prediction model.

[0066] In some embodiments, the second prediction model is obtained by training a machine learning model using a pre-built dataset of abnormal cochleas, such as... Figure 3 As shown, the specific steps include:

[0067] S310: Preprocess the abnormal cochlear dataset; specifically, traverse the original information data of the patient samples, remove irrelevant variables and perform one-hot encoding on the remaining discrete variables; traverse the original information data of the patients and impute missing values; normalize the remaining continuous variables after imputation of missing values.

[0068] S320: In the preprocessed cochlear deformity dataset, calculate the correlation between the preoperative information and the postoperative information of the cochlear deformity patient samples to obtain a second correlation result; for example, the chi-square test and Spearman correlation algorithm can also be used to calculate the correlation between the preoperative information and the postoperative information.

[0069] S330: Sort the second relevance results from high to low according to feature importance;

[0070] S340: Input the dataset consisting of the first k2 features after sorting into a pre-built machine learning model to train the second prediction model.

[0071] In other words, during the training of the first and second prediction models, the original patient information data within a preset time period is first acquired and integrated to obtain datasets for normal cochlear implants and abnormal cochlear implants. Each dataset includes preoperative and postoperative information for the corresponding patient. Based on the preoperative and postoperative information in each dataset, the correlation between the preoperative and postoperative information is analyzed, and the features are ranked by importance. Based on the ranked features of each dataset, the corresponding datasets of the top k features are extracted, and prediction models for preoperative information and postoperative information in each dataset are established, obtaining evaluation metrics for the prediction models. Based on the evaluation metrics of the prediction models in each dataset, the best model for each dataset is selected. Specifically, the best prediction model for hearing and speech after cochlear implantation in patients with normal cochlear implants is trained in the normal cochlear implant dataset, and the best prediction model for hearing and speech after cochlear implantation in patients with abnormal cochlear implants is trained in the abnormal cochlear implant dataset. This facilitates doctors in accurately predicting the postoperative hearing and speech rehabilitation effects of patients.

[0072] In some embodiments, when training the first prediction model and the second prediction model, the correlation between the preoperative information and the postoperative information obtained by the analysis can be analyzed using methods such as chi-square test and Spearman correlation to analyze the influence of the preoperative information on the postoperative information in their respective datasets.

[0073] Specifically, the ranking of feature importance includes:

[0074] The influence of the preoperative information on the postoperative information was obtained using the chi-square test and Spearman correlation method. The results were normalized, and the normalized results were summed and sorted to obtain a comprehensive feature ranking with decreasing importance.

[0075] The extraction of the corresponding dataset of the first k features after sorting (it should be understood that, for ease of distinction, these k features are k1 features when training the first prediction model, and k2 features when training the second prediction model) further includes:

[0076] For a dataset containing k features, divide it into a training set and a test set in a 4:1 ratio. The training set is used to train the random forest model, and the test set is used to test the model's performance to obtain the model's evaluation metrics.

[0077] For a dataset containing k features, the details are as follows:

[0078] k = 1, 2, ..., n, where n is the number of features after removing the labels from each dataset (normal cochlea, abnormal cochlea).

[0079] In some embodiments, when training the first and second prediction models, it is necessary to analyze the correlation between the preoperative and postoperative information based on the preoperative and postoperative information of each corresponding patient and rank the features by importance. The specific process can be as follows: using the chi-square test and Spearman correlation method, analyze the influence of the preoperative information on the postoperative information in each dataset. Using the chi-square test and Spearman correlation method, obtain the influence results of the preoperative information on the postoperative information, normalize the respective results, and sum and rank the normalized results to obtain a comprehensive feature ranking in descending order of importance.

[0080] This embodiment describes the chi-square test and the Spearman correlation algorithm.

[0081] Specifically, the chi-square test falls under the category of non-parametric tests, primarily used to compare the correlation between two or more sample rates (proportions) and two categorical variables. Its fundamental idea is to compare the degree of agreement or goodness of fit between theoretical and actual frequencies. A larger chi-square value indicates a greater deviation between the observed and expected values, suggesting a weaker independence between the two events. The specific calculation formula is as follows:

[0082]

[0083] Spearman correlation is also a method for measuring variable similarity. It measures the degree of correlation between continuous variables, with an output range of -1 to +1. 0 represents no correlation, negative values ​​indicate negative correlation, and positive values ​​indicate positive correlation. Variables with a high degree of similarity to the target variable are considered more important. The specific calculation formula is as follows:

[0084]

[0085] Where: d i This represents the difference in rank between the i-th data pairs, where n is the total number of observed samples.

[0086] In this embodiment, influencing factors are analyzed for the two target variables, "SIR" and "CAP". A strong correlation indicates high importance, and vice versa. Specifically, as follows: Figure 4 and Figure 5 As shown.

[0087] Then, based on the features of each dataset sorted, the corresponding datasets of the top k features after sorting are extracted, a prediction model of preoperative information on postoperative information of each dataset is established, and the evaluation index of the prediction model is obtained.

[0088] The steps after extracting the corresponding dataset of the first k features after sorting are as follows:

[0089] For the dataset containing the top k features after extraction and sorting, a training set and a test set are divided in a 4:1 ratio. The training set is used to train the random forest model, and the test set is used to test the model's performance to obtain the model's evaluation metric. Here, k = 1, 2, ..., n, where n is the number of features after removing labels from each dataset (normal cochlea, abnormal cochlea).

[0090] Based on the evaluation metrics of the prediction models for each dataset, the best prediction model for each dataset is selected, namely: the best prediction model for hearing and speech after cochlear implantation in patients with normal cochlear implantation is trained in the normal cochlear implant dataset; and the best prediction model for hearing and speech after cochlear implantation in patients with malformed cochlear implantation is trained in the malformed cochlear implant dataset.

[0091] In the above specific embodiments, the machine learning-based method for predicting postoperative hearing and speech rehabilitation outcomes in cochlear implant patients provided by this invention uses raw data as a foundation and combines artificial intelligence machine learning algorithms to construct a prognostic prediction model, which can assist doctors in predicting the prognosis of cochlear implant patients. Specifically, based on statistical analysis, factors that significantly influence prognostic prediction are identified, and their degree of influence is ranked. The datasets of the k most influential features are selected to construct the model, and the model with the best evaluation index is chosen as the prognostic model, thus maximizing the accuracy of the prognostic model prediction. This solves the technical problem in existing technologies that cannot predict postoperative hearing and speech rehabilitation outcomes in cochlear implant patients based on big data.

[0092] The above method will be applied to a specific example to demonstrate its implementation process and technical effects.

[0093] The original patient information data within a previously preset time period is acquired and integrated to obtain datasets of normal and abnormal cochlea. Each dataset includes the preoperative and postoperative information of the corresponding patient.

[0094] First, the raw data was analyzed as a whole, which included two datasets: a normal cochlea dataset and a cochlea malformation dataset. Each dataset consisted of two parts: "preoperative information" and "postoperative information" of the patient.

[0095] The "preoperative information" features include: age in months; residual hearing (marked as 125 if there is no residual hearing); cochlear foramen width; internal auditory canal width; number of nerve points (e.g., 1, 2, 3, 4...); facial nerve area; vestibulocochlear nerve area; VCN / FN; 40Hz (classified as 1 or 0 based on presence or absence of response); ABR (classified as 1 or 0 based on presence or absence of response); CM (classified as 1 or 0 based on presence or absence of response); DPOAE (classified as 1 or 0 based on presence or absence of response); acoustic impedance representing tympanic membrane status (e.g., A, As, B, C, replaced by 0, 1, 2, 3 respectively); and implantation side (e.g., left side L, right side R).

[0096] The features of "postoperative information" can include the hearing CAP and speech SIR rehabilitation outcomes of cochlear implant patients 2 years after surgery. CAP ranges from 0 to 7; SIR ranges from 1 to 5.

[0097] The features mentioned above are all information features that need to be included in the normal cochlear dataset. Furthermore, the "preoperative information" features of the cochlear malformation dataset are slightly different from those of the normal cochlear dataset, while the "postoperative information" features are the same. The "preoperative information" features in the cochlear malformation dataset include: sex; age in months; cochlear implant manufacturer, such as AB, Med, Nu, Nor; implantation side; average ASSR; 40Hz; ABR; CM; DPOAE; acoustic impedance; PTA; cochlear type, such as CA, CC, CH, IP-I, IP-II, IP-III, RO; and number of nerve points.

[0098] The preprocessing procedure for the normal cochlear dataset is illustrated below:

[0099] Step 1: Since the data integrity is high, data with incomplete feature information is directly deleted.

[0100] Step 2: Delete the features “VCN / FN”, “40Hz”, “ABR”, “CM”, “DPOAE”, “acoustic impedance”, and “implantation side”.

[0101] Step 3: Perform maximum and minimum normalization on the characteristic information of "age in months", "residual hearing", "cochlear foramen width", "internal auditory canal width", "facial nerve area" and "vestibular cochlear nerve area".

[0102] The specific process for preprocessing the cochlear malformation dataset is as follows:

[0103] Step 1: Delete the features “gender”, “implantation side”, “CM”, “DPOAE”, and “PTA”.

[0104] Step 2: Perform one-hot encoding on the feature information of "cochlear implant manufacturer", "40Hz", "ABR", "acoustic impedance" and "cochlear type";

[0105] Step 3: Use the MissForest algorithm to impute missing values ​​in the cochlear malformation dataset;

[0106] Step 4: Perform max-min normalization on the feature information of "age in months" and "average ASSR".

[0107] After preprocessing both datasets, the following steps are the same; here, we take the normal cochlear dataset as an example.

[0108] When ranking important influencing factors, the following steps are included as an example:

[0109] First, the chi-square test and Spearman correlation were used to analyze the degree of influence of the processed feature information on the postoperative hearing CAP rehabilitation effect of patients.

[0110] Then, the results of the chi-square test and Spearman test are normalized. The normalized results are summed and sorted according to the same features to obtain the comprehensive influence feature ranking in descending order of importance.

[0111] When establishing a patient prognostic model, the following steps are included as an example:

[0112] Based on the ranking results of comprehensive influence features, the corresponding datasets for the top k features (k=1, 2, ..., 7) are extracted. For each dataset containing k=1, 2, ..., 7 features, a training set and a test set are divided in a 4:1 ratio. The training set is used to train a random forest model, and the test set is used to test the model's performance. A random forest prediction model for feature-based CAP is built on the training set, and the test set tests the seven prediction models to obtain seven evaluation metrics. Based on the evaluation metrics, the optimal CAP prediction model for hearing in normal cochlear implant patients is selected.

[0113] Based on existing data, a random forest model was trained using centralized feature information and postoperative information (SIR) from normal cochlear datasets to obtain a predictive model for speech rehabilitation after cochlear implantation in patients with normal cochlea. A random forest model was trained using centralized feature information and postoperative information (CAP) from cochlear malformation datasets to obtain a predictive model for hearing rehabilitation after cochlear implantation in patients with cochlear malformation. A random forest model was trained using centralized feature information and postoperative information (SIR) from cochlear malformation datasets to obtain a predictive model for speech rehabilitation after cochlear implantation in patients with cochlear malformation.

[0114] Finally, the above steps are packaged into interactive software to form a platform for predicting the postoperative hearing and speech rehabilitation effects of cochlear implant patients.

[0115] In practical applications, the accuracy rate of the prediction model for hearing rehabilitation after cochlear implantation in patients with normal cochlea can reach 79%; the accuracy rate of the prediction model for speech rehabilitation after cochlear implantation in patients with normal cochlea can reach 93%; the accuracy rate of the prediction model for hearing rehabilitation after cochlear implantation in patients with cochlear malformation can reach 87%; and the accuracy rate of the prediction model for speech rehabilitation after cochlear implantation in patients with cochlear malformation can reach 67%.

[0116] As described above, taking cochlear implant patients as an example, this invention, based on cochlear medicine and combined with artificial intelligence machine learning algorithms, constructs a postoperative hearing and speech rehabilitation model for cochlear implant patients. For preprocessed cochlear implant patient data, through normalization, statistically, the important factors affecting postoperative hearing and speech rehabilitation are summarized and ranked according to their degree of influence. The k most important features are selected sequentially to train prediction models for postoperative hearing and speech rehabilitation outcomes in patients with normal cochlea and patients with cochlear malformations. The optimal number of features is chosen to establish the prognostic model, ensuring both robustness and accuracy.

[0117] In addition to the methods described above, this invention also provides a device for predicting the postoperative hearing and speech rehabilitation effects in patients with cochlear implantation, such as... Figure 6 As shown, the device includes:

[0118] The data acquisition unit 610 is used to acquire the original information data of the target patient to be predicted, the original information data of the target patient including the cochlear type, preoperative information and postoperative information of the target patient;

[0119] The first result output unit 620 is used to input the original information data of the target patient into a pre-trained first prediction model when the cochlear type is normal cochlea, so as to obtain a first prediction result for normal cochlea.

[0120] The second result output unit 630 is used to input the original information data of the target patient into a pre-trained second prediction model when the cochlear type is malformed cochlea, so as to obtain a second prediction result for malformed cochlea.

[0121] The first prediction model is obtained by training a machine learning model using a pre-built normal cochlear dataset, which is composed of the original information data of normal cochlear patient samples.

[0122] The second prediction model is obtained by training a machine learning model using a pre-built dataset of abnormal cochleas, which consists of raw information data of samples from patients with abnormal cochleas.

[0123] In some embodiments, the training of the first prediction model and the second prediction model includes:

[0124] The raw information data of all patient samples within a previously preset time period are obtained, and the raw information data of the patient samples are integrated based on the cochlear type to obtain the normal cochlear dataset composed of the raw information data of normal cochlear patient samples, and the abnormal cochlear dataset composed of the raw information data of abnormal cochlear patient samples.

[0125] The original information data of the patient sample includes the cochlear type, preoperative information, and postoperative information of the patient sample.

[0126] In some embodiments, the first prediction model, obtained by training a machine learning model using a pre-built normal cochlear dataset, specifically includes:

[0127] The normal cochlear dataset is preprocessed;

[0128] In the preprocessed normal cochlear dataset, the correlation between the preoperative information and the postoperative information of the normal cochlear patient samples is calculated to obtain the first correlation result;

[0129] Sort the first relevance results from high to low according to feature importance;

[0130] The dataset consisting of the first k1 sorted features is input into a pre-built machine learning model to train the first prediction model.

[0131] In some embodiments, preprocessing of the normal cochlear dataset specifically includes:

[0132] The original information data of the patient samples is traversed, data containing missing values ​​and irrelevant variables are deleted, and the remaining continuous variables after deletion are normalized.

[0133] In some embodiments, the correlation between preoperative information and postoperative information is calculated using the chi-square test and the Spearman correlation algorithm.

[0134] In some embodiments, the second prediction model, obtained by training a machine learning model using a pre-built dataset of abnormal cochleas, specifically includes:

[0135] The dataset of deformed cochleas is preprocessed;

[0136] In the preprocessed cochlear deformity dataset, the correlation between the preoperative information and the postoperative information of the cochlear deformity patient samples is calculated to obtain a second correlation result;

[0137] The second relevance results are sorted by feature importance from high to low;

[0138] The dataset consisting of the first k2 sorted features is input into a pre-built machine learning model to train the second prediction model.

[0139] In some embodiments, the preprocessing of the abnormal cochlear dataset specifically includes:

[0140] The original information data of the patient samples is traversed, irrelevant variables are removed, and the remaining discrete variables are one-hot encoded; the original information data of the patients is traversed, and missing values ​​are imputed; the remaining continuous variables after missing value imputed are normalized.

[0141] In the above specific embodiments, the cochlear implant patient postoperative hearing and speech rehabilitation effect prediction device provided by the present invention acquires the original information data of the target patient to be predicted. The original information data of the target patient includes the cochlear type, preoperative information, and postoperative information of the target patient. When the cochlear type is normal, the original information data of the target patient is input into a pre-trained first prediction model to obtain a first prediction result for normal cochlea. When the cochlear type is malformed, the original information data of the target patient is input into a pre-trained second prediction model to obtain a second prediction result for malformed cochlea.

[0142] This approach applies machine learning to predict the postoperative hearing and speech rehabilitation outcomes of cochlear implant patients, using it as a tool to assist doctors in prediction. This avoids predictions being limited by human experience and provides more objective and accurate results. It solves the technical problem in existing technologies where the inability to predict postoperative hearing and speech rehabilitation outcomes of cochlear implant patients based on big data leads to poor prediction accuracy.

[0143] Figure 7 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 7As shown, the electronic device may include a processor 710, a communications interface 720, a memory 730, and a communication bus 740, wherein the processor 710, the communications interface 720, and the memory 730 communicate with each other via the communication bus 740. The processor 710 can call logical instructions in the memory 730 to execute the above-described method.

[0144] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0145] On the other hand, the present invention also provides a computer program product, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, and when the program instructions are executed by a computer, the computer is able to perform the methods described above.

[0146] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the methods described above.

[0147] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0148] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using a combination of hardware and software. When applied as software, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0149] The above specific embodiments further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting the postoperative hearing and speech rehabilitation effect in patients with cochlear implantation, characterized in that, The method includes: Obtain the original information data of the target patient to be predicted, including the cochlear type, preoperative information and postoperative information of the target patient; In the case where the cochlear type is normal, the original information data of the target patient is input into a pre-trained first prediction model to obtain a first prediction result for normal cochlear. In the case where the cochlear type is malformed cochlea, the original information data of the target patient is input into a pre-trained second prediction model to obtain a second prediction result for malformed cochlea; The first prediction model is obtained by training a machine learning model using a pre-built normal cochlear dataset, which is composed of the original information data of normal cochlear patient samples. The second prediction model is obtained by training a machine learning model using a pre-built dataset of abnormal cochleas, which consists of raw information data of samples from patients with abnormal cochleas. The process includes, before training the first prediction model and the second prediction model, the following steps: The raw information data of all patient samples within a previously preset time period are obtained, and the raw information data of the patient samples are integrated based on the cochlear type to obtain the normal cochlear dataset composed of the raw information data of normal cochlear patient samples, and the abnormal cochlear dataset composed of the raw information data of abnormal cochlear patient samples. The original information data of the patient sample includes the cochlear type, preoperative information, and postoperative information of the patient sample. The first prediction model, obtained by training a machine learning model using a pre-built normal cochlear dataset, specifically includes: The normal cochlear dataset is preprocessed; In the preprocessed normal cochlear dataset, the correlation between the preoperative information and the postoperative information of the normal cochlear patient samples is calculated to obtain the first correlation result; Sort the first relevance results from high to low according to feature importance; Sort by front A dataset consisting of several features is input into a pre-built machine learning model to train and obtain the first prediction model; The preprocessing of the normal cochlear dataset specifically includes: The original information data of the patient samples is traversed, data containing missing values ​​and irrelevant variables are deleted, and the remaining continuous variables after deletion are normalized.

2. The method for predicting the postoperative hearing and speech rehabilitation effect in cochlear implant patients according to claim 1, characterized in that, The correlation between preoperative information and postoperative information was calculated using the chi-square test and Spearman correlation algorithm.

3. The method for predicting the postoperative hearing and speech rehabilitation effect in patients with cochlear implantation according to claim 1, characterized in that, The second prediction model, trained using a pre-built dataset of abnormal cochleas, is based on a machine learning model and specifically includes: The dataset of deformed cochleas is preprocessed; In the preprocessed cochlear deformity dataset, the correlation between the preoperative information and the postoperative information of the cochlear deformity patient samples is calculated to obtain a second correlation result; The second relevance results are sorted by feature importance from high to low; Sort by front A dataset consisting of several features is input into a pre-built machine learning model to train and obtain the second prediction model.

4. The method for predicting the postoperative hearing and speech rehabilitation effect in patients with cochlear implantation according to claim 2, characterized in that, The preprocessing of the abnormal cochlear dataset specifically includes: The original information data of the patient samples are traversed, irrelevant variables are removed, and the remaining discrete variables are one-hot encoded. Iterate through the original information data of the patient samples and fill in missing values; Normalize the remaining continuous variables after missing values ​​are filled.

5. A device for predicting the postoperative hearing and speech rehabilitation effect in patients with cochlear implantation, used to implement the method as described in any one of claims 1-4, characterized in that, The device includes: The data acquisition unit is used to acquire the original information data of the target patient to be predicted, including the cochlear type, preoperative information and postoperative information of the target patient. The first result output unit is used to input the original information data of the target patient into a pre-trained first prediction model when the cochlear type is normal cochlea, so as to obtain a first prediction result for normal cochlea. The second result output unit is used to input the original information data of the target patient into a pre-trained second prediction model when the cochlear type is malformed cochlea, so as to obtain a second prediction result for malformed cochlea. The first prediction model is obtained by training a machine learning model using a pre-built normal cochlear dataset, which is composed of the original information data of normal cochlear patient samples. The second prediction model is obtained by training a machine learning model using a pre-built dataset of abnormal cochleas, which consists of raw information data of samples from patients with abnormal cochleas.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 4.

Images