Identification method and device for children with neurodevelopment disorder and electronic equipment

Abstract

The embodiment of the invention discloses a method and a device for identifying children with neurodevelopment disorder and electronic equipment. The method comprises the following steps: acquiring demographic statistics characteristics and intellectual index characteristics of to-be-identified children with normal intelligence; inputting the demographic statistical characteristics and the intellectual index characteristics of the to-be-identified children into a trained convergence identification model; the convergence recognition model is any one of the following machine learning models: a logistic regression model, a random forest model, a decision tree model or an extreme gradient boosting model; obtaining a target recognition result output by the convergence recognition model; the target recognition result indicates whether the to-be-recognized child has the neurodevelopment disorder or not. By implementing the embodiment of the invention, whether the children with normal intelligence have the neurodevelopment disorder problem or not can be automatically identified through the trained machine learning model in combination with the demographic characteristics and the intellectual index characteristics of the children, the accuracy is high, and the dependence on artificial experience is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a method and device for identifying neurodevelopmental disorder children and electronic equipment. BACKGROUND

[0002] Neurodevelopmental disorders (NDDs) refer to behavioral and cognitive disorders that occur during the development stage, which are manifested as significant difficulties in obtaining or performing specific intelligence, movement, language or social function, and affect about 4-10% of children. For neurodevelopmental disorder children with normal intelligence, such as high-functioning autism spectrum disorder (HFASD) and developmental dyslexia (DD), it is usually difficult to be identified and diagnosed by schools and clinics.

[0003] The prevalence of autism spectrum disorder (ASD) is about 1%, of which HFASD accounts for about 70%, which is mainly manifested as social communication disorder, stereotyped repetitive behavior and narrow interests, but with normal intelligence. The prevalence of DD is between 4% and 8%, which is mainly manifested as difficulties in accurately or fluently reading words, decoding and spelling in individuals without intellectual disability, vision or hearing problems. Although the intelligence quotient of children with high-functioning autism spectrum disorder and developmental dyslexia is above 70, the harm is serious, which will seriously affect their studies, life and social interaction, and will last a lifetime.

[0004] At present, Wechsler Intelligence Scale for Children-Fourth Edition (WISC-Ⅳ) is commonly used as an evaluation tool for neurodevelopmental disorder children in clinics. However, the intelligence evaluation of Wechsler scale only pays attention to the analysis of intelligence level, and ignores the influence of unbalanced intelligence structure, which leads to the serious neglect of the importance of intelligence evaluation in the diagnosis of neurodevelopmental disorders. At the same time, since NDDs children often show the characteristics of normal intelligence, only using intelligence evaluation is easy to cause parents or clinicians to ignore NDDs children with normal intelligence, which delays the early identification and intervention of such children. SUMMARY

[0005] The embodiment of the application discloses a method and device for identifying children with neurodevelopmental disorders and an electronic device, which can automatically identify whether children with normal intelligence have neurodevelopmental disorders by combining demographic characteristics and intelligence index characteristics of the children and using a trained machine learning model, has high accuracy and reduces dependence on artificial experience.

[0006] The embodiment of the application discloses a method for identifying children with neurodevelopmental disorders, which comprises the following steps: Obtaining demographic characteristics and intelligence index characteristics of a child to be identified with normal intelligence; Inputting the demographic characteristics and intelligence index characteristics of the child to be identified into a trained convergent identification model; the convergent identification model is any one of the following machine learning models: a logistic regression model, a random forest model, a decision tree model or an extreme gradient boosting model; Obtaining a target identification result output by the convergent identification model; the target identification result indicates whether the child to be identified has a neurodevelopmental disorder.

[0007] As an optional implementation, the convergent identification model is a logistic regression model.

[0008] As an optional implementation, the demographic characteristics at least include age and gender of the child, family income of the child; and the intelligence index characteristics at least include a verbal comprehension index, a processing speed index, a difference between a working memory index and the processing speed index, and a perceptual reasoning index.

[0009] As an optional implementation, the demographic characteristics further include educational level of the mother of the child and educational level of the father of the child; and the intelligence index characteristics further include a working memory index, a difference between a verbal comprehension index and a perceptual reasoning index, a difference between a verbal comprehension index and a working memory index, a difference between a verbal comprehension index and a processing speed index, a difference between a perceptual reasoning index and a working memory index, and a difference between a perceptual reasoning index and a processing speed index.

[0010] As an optional implementation, before the step of inputting the demographic characteristics and intelligence index characteristics of the child to be identified into the trained convergent identification model, the method further comprises the following steps: Training a plurality of initialized sample models one by one using a training data set to obtain a plurality of trained convergent sample models; the plurality of initialized sample models are all binary classification machine learning models; and the training data set comprises demographic characteristics and intelligence index characteristics of a plurality of first sample children; The identification performance of each of the converged sample models for identifying children with neurodevelopmental disorders is tested using a first test data set, and scores of each of the converged sample models under evaluation indicators are obtained; the first test data set includes demographic characteristics and intelligence index characteristics of a plurality of second sample children; the evaluation indicators include one or more of the following indicators: area under the curve, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and Cohen's Kappa value. The scores of each of the converged sample models under the evaluation indicators are compared, and one of the converged sample models is determined as a trained converged identification model.

[0011] As an optional implementation, the method further includes: obtaining a logistic stepwise regression model, wherein the logistic stepwise regression model is constructed by a logistic regression model and a logistic stepwise regression method; determining key identification features from the demographic characteristics and the intelligence index characteristics by using the first test data set and a second test data set in combination with the logistic stepwise regression model; wherein the second test data set includes demographic characteristics and intelligence index characteristics of a plurality of third sample children; the key identification features include: the age and gender of the children in the demographic characteristics, the family income of the children, the verbal comprehension index in the intelligence index characteristics, the processing speed index, the difference between the working memory index and the processing speed index, and the perceptual reasoning index.

[0012] As an optional implementation, the method further includes: calculating the importance of each of the key identification features for identifying children with neurodevelopmental disorders by using a Shapley additive explanation method.

[0013] The application also discloses a device for identifying children with neurodevelopmental disorders, and the device includes: an obtaining unit, configured to obtain demographic characteristics and intelligence index characteristics of a to-be-identified child with normal intelligence; an identification unit, configured to input the demographic characteristics and the intelligence index characteristics of the to-be-identified child into a trained converged identification model; the converged identification model is any one of the following machine learning models: a logistic regression model, a random forest model, a decision tree model, or an extreme gradient boosting model; an output unit, configured to obtain a target identification result output by the converged identification model; the target identification result indicates whether the to-be-identified child has a neurodevelopmental disorder.

[0014] As an optional implementation, the device further includes: The training unit is configured to train a plurality of initialized sample models one by one by using a training data set to obtain a plurality of trained converged sample models; each of the initialized sample models is a machine learning model for binary classification; and the training data set includes demographic characteristics and intelligence index characteristics of a plurality of first sample children. The evaluation unit is configured to test the identification performance of each of the converged sample models in identifying neurodevelopmental disorder children by using a first test data set to obtain a score of each of the converged sample models under an evaluation index; the first test data set includes demographic characteristics and intelligence index characteristics of a plurality of second sample children; and the evaluation index includes one or more of the following indexes: area under the curve, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and Cohen's Kappa value. The first determination unit is configured to determine one of the converged sample models as a trained converged identification model according to the score of each of the converged sample models under the evaluation index.

[0015] The electronic device disclosed in the embodiments of the present application includes a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to implement any one of the identification of neurodevelopmental disorder children disclosed in the embodiments of the present application.

[0016] The computer readable storage medium disclosed in the embodiments of the present application stores a computer program, and the computer program is executed by a processor to implement any one of the identification methods of neurodevelopmental disorder children disclosed in the embodiments of the present application.

[0017] Compared with the related art, the embodiments of the present application have the following beneficial effects: The embodiments of the present application can automatically identify whether a child with normal intelligence has a neurodevelopmental disorder problem by combining the demographic characteristics and intelligence index characteristics of the child and using a trained machine learning model. Since not only intelligence index characteristics are used, but demographic characteristics are also introduced, the neurodevelopmental disorder problem of a child with normal intelligence can be identified more specifically, which is conducive to early identification and clinical screening of high-risk children. Moreover, the embodiments of the present application greatly reduce the dependence on artificial experience and are suitable for all clinical outpatient services that can perform intelligence evaluation. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0019] Figure 1 This is a flowchart illustrating a method for identifying children with neurodevelopmental disorders disclosed in an embodiment of this application; Figure 2 This is a method for training and screening a model disclosed in an embodiment of this application; Figure 3 This is an example diagram showing the relative weights of each key identification feature calculated using the SHAP method; Figure 4 This application discloses an identification device for children with neurodevelopmental disorders. Figure 5 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] It should be noted that the terms "comprising" and "having," and any variations thereof, in the embodiments and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0022] This application discloses a method, device, and electronic device for identifying children with neurodevelopmental disorders. It can automatically identify whether intellectually normal children have neurodevelopmental disorders by combining their demographic and intellectual indicators using a trained machine learning model. This method is highly accurate and reduces reliance on manual experience. Detailed descriptions follow.

[0023] Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for identifying children with neurodevelopmental disorders disclosed in an embodiment of this application. Figure 1 The method shown can be executed by electronic devices with computing capabilities, such as personal computers and industrial computers. Figure 1 As shown, the method may include the following steps: 110. Obtain the demographic and intelligence indicators of the intellectually normal children to be identified.

[0024] In this application embodiment, demographic characteristics are used to reveal differences in the prevalence of neurodevelopmental disorders among different groups, and may include the child's personal information and / or the child's family information. For example, the child's personal information may include, but is not limited to, gender or age. The child's family information may include, but is not limited to, the parents' education level and family income. Variables such as parents' education level and family income may affect the understanding of NDDs.

[0025] Intelligence indicators, used to characterize a child's intelligence level, can be measured using methods such as the Wechsler Intelligence Scale. Intelligence indicators may include, but are not limited to, one or more of the following intelligence indices: Verbal Comprehension Index (VCI), Perceptual Reasoning Index (PRI), Working Memory Index (WMI), Processing Speed ​​Index (PSI), and Full Scale Intelligence Quotient (FSIQ), which can be referred to as intelligence scores; the differences between the Verbal Comprehension Index and Perceptual Reasoning Index (VCI-PRI), the differences between the Verbal Comprehension Index and Working Memory Index (VCI-WMI), the differences between the Verbal Comprehension Index and Processing Speed ​​Index (VCI-PSI), the differences between the Perceptual Reasoning Index and Working Memory Index (PRI-WMI), the differences between the Perceptual Reasoning Index and Processing Speed ​​Index (PRI-PSI), and the differences between the Working Memory Index and Processing Speed ​​Index (WMI-PSI), which can be referred to as intelligence structure. The intelligence indicators of a child to be identified may include intelligence scores and / or intelligence structure.

[0026] 120. Input the demographic and intelligence characteristics of the children to be identified into the trained convergent recognition model.

[0027] In this embodiment, the demographic characteristics obtained for the child to be identified may include one or a combination of the above information (the child's personal information and / or family information), and the obtained intelligence indicator characteristics may include one or a combination of the above indices. The demographic characteristics and intelligence indicator characteristics of the child to be identified can form an input vector and be input into the model.

[0028] In the embodiments of the present application, the convergent recognition model can be any one of the following machine learning models obtained after training: Logistic Regression (LR), Random Forest (RF), Decision Tree (DT), and eXtreme Gradient Boosting (XGBoost). The training method of the machine learning model can use any known training method, such as supervised learning method, unsupervised method, etc., and is not limited in particular. The data set used for training includes the demographic characteristics and intelligence index characteristics of children with normal intelligence and NDDs and typically developing children (TD) of the same age.

[0029] 130. Obtain the target recognition result output by the convergent recognition model.

[0030] In the embodiments of the present application, the convergent recognition model refers to a machine learning model that has been trained to achieve a convergent effect. Since the above-mentioned machine learning models used in the embodiments are all binary classification models, they can output two different results, one indicating the presence of a neurodevelopmental disorder and the other indicating the absence of a neurodevelopmental disorder. Therefore, after inputting the above-mentioned input vector corresponding to the child to be identified into the convergent recognition model, the target recognition result output by the convergent recognition model can indicate whether the child to be identified has a neurodevelopmental disorder.

[0031] That is, the embodiments of the present application disclose an automatic neurodevelopmental disorder child recognition method based on machine learning, which can automatically recognize whether a child with normal intelligence has a neurodevelopmental disorder problem by combining the demographic characteristics and intelligence index characteristics of the child. Since the model training considers the neurodevelopmental disorder problem of children with normal intelligence, not only intelligence index characteristics are used, but also demographic characteristics are introduced, and the sample data collected includes data of children with normal intelligence but with neurodevelopmental disorder, so that the convergent recognition model obtained by training can accurately recognize the neurodevelopmental disorder problem of children with normal intelligence, which is beneficial to early identification and clinical screening of high-risk children. Moreover, the embodiments of the present application greatly reduce the dependence on artificial experience and are suitable for all clinical outpatient services that can perform intelligence assessment.

[0032] In order to better illustrate the convergent recognition model disclosed in the embodiments of the present application, the following describes how to obtain the convergent recognition model. Please refer to Figure 2 , Figure 2 is a model training and screening method disclosed in the embodiments of the present application. Similarly, the method can be executed by an electronic device with computing capability, including the following steps: 210. training each of the initialized sample models using the training dataset to obtain a plurality of trained converged sample models.

[0033] In the embodiments of the present application, each of the initialized sample models is a binary classification machine learning model. For example, each of the four machine learning models described above can be trained.

[0034] In the embodiments of the present application, the sample dataset can be collected from children with NDDs (including HFASD and DD) and TD children through legal channels. The inclusion criteria for the sample children can include: (1) normal hearing and vision (or corrected to normal vision); (2) FSIQ between 80 and 120.

[0035] (3) TD children are clinically evaluated by pediatricians to confirm that they have no NDDs.

[0036] (4) NDDs (HFASD and DD) group received comprehensive neuropsychiatric evaluation and were diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) standard.

[0037] The sample dataset can be randomly divided into two groups: 80% of the data as the training dataset, and 20% of the data as the first test dataset. That is, the training dataset and the first test dataset come from different sample children, hereinafter referred to as first sample children and second sample children. The training dataset includes demographic characteristics and intelligence index characteristics of a plurality of first sample children; the first test dataset includes demographic characteristics and intelligence index characteristics of a plurality of second sample children.

[0038] The training dataset is used to train each of the initialized sample models, so that each of the sample models can have a certain ability to identify NDDs children.

[0039] 220. test the identification performance of each of the converged sample models for identifying children with neurodevelopmental disorders using the first test dataset to obtain the score of each of the converged sample models under the evaluation index.

[0040] In the embodiments of the present application, the evaluation indexes include one or more of the following indexes: Area Under the Curve (AUC), Sensitivity, Specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV), F1 score, Cohen’s Kappa. The above evaluation indexes can be used alone to evaluate the performance of the model, and the specific calculation method is also known technology, which will not be expanded below.

[0041] 230. Determine a converged sample model as a trained converged recognition model according to the scores of the converged sample models under the evaluation indexes.

[0042] In the embodiments of the present application, if only the scores of each converged sample model under a single evaluation index are calculated, the converged sample model with the highest score can be selected as the converged recognition model according to the descending order. If multiple scores of each converged sample model under multiple evaluation indexes are calculated, one converged sample model can be selected as the converged recognition model from the multiple scores according to a preset rule.

[0043] For example, the converged sample model includes a logistic regression model (LR), a random forest model (RF), a decision tree model (DT) or an extreme gradient boosting model (XGBoost). The following table is the evaluation index score of each model under the test data set.

[0044] Table 1. Performance of classification model on the first test data set

[0045] It can be seen that the LR model shows superior sensitivity, NPV, F1 score and Cohen’s Kappa, which can effectively identify NDDs or TD children and maintain high accuracy. Although the RF model achieves a higher AUC, its sensitivity is reduced, which may miss the diagnosis of NDDs (HFASD and DD), resulting in delayed intervention for NDDs children. Therefore, given the balanced performance and lower missed diagnosis rate of the LR model, the LR model can be selected as the converged recognition model as an optional implementation.

[0046] The more features included in the vector input to the machine learning model, the more factors can be incorporated into the identification of NDD children. As mentioned earlier, demographic features may include five demographic information items such as age and gender, while intelligence indicator features may include ten intelligence indicators. In some possible embodiments, due to limitations in the computing power of the computing device, it may be necessary to reduce the size of the input vector. To balance performance and computing power limitations, this application also discloses the following key feature selection methods.

[0047] A logistic stepwise regression model is obtained, which is constructed using a logistic regression model and a stepwise regression method. Then, using the aforementioned first and second test datasets, the stepwise logistic regression model is combined to select key identification features from demographic and intelligence indicators. The second test dataset differs from the first and training datasets and may include demographic and intelligence indicators from several third-sample children.

[0048] Stepwise regression is an iterative algorithm that automatically adds or removes predictor variables to select the combination of variables that contributes most to the model (usually based on statistical significance, such as p-value or AIC criterion). As mentioned earlier, the LR model is a binary classification problem model that uses regression coefficients to measure the impact of each feature on the result.

[0049] For example, the stepwise LR model achieves AUC=0.91, sensitivity=0.81, specificity=0.83, PPV=0.74, and NPV=0.88 on the first test dataset, and AUC=0.83, sensitivity=0.80, specificity=0.71, PPV=0.52, and NPV=0.90 on the second test dataset.

[0050] Based on this, the key identification features output by the stepwise logistic regression model include: the child's age and gender, the child's family income, the verbal comprehension index (VCI), the processing speed index (PSI), the difference between the working memory index and the processing speed index (WMI-PSI), and the perceptual reasoning index (PRI).

[0051] For example, the table below shows the statistical properties of the key features selected by the stepwise logistic regression model.

[0052] Table 2. Performance of the classification model on the first test dataset

[0053] Note: Bold text indicates statistical significance (P < 0.05); OR refers to odds ratios; CI refers to confidence intervals.

[0054] Statistically, a 1 percentage point decrease in VCI, PRI, and PSI increases the risk of NDDs by 56%, 22%, and 55%, respectively. A significant difference in WMI-PSI compared to a balanced intellectual structure increases the risk of NDDs by 32%. Therefore, all of these indicators can serve as key characteristics for identifying the presence of NDDs in children.

[0055] In other words, as an optional implementation method, when acquiring the demographic and intelligence characteristics of the child to be identified, at a minimum, the child's age and gender, the child's family income, verbal comprehension index (VCI), processing speed index (PSI), the difference between working memory index and processing speed index (WMI-PSI), and perceptual reasoning index (PRI) can be acquired, which is beneficial to balancing identification accuracy and computational ability limitations.

[0056] Optionally, to determine the importance of each of the aforementioned key identifying features for the identification of children with NDDs, the Shapley Additive exPlanation (SHAP) method can be used to calculate the degree of importance of each key identifying feature for the identification of children with neurodevelopmental disorders.

[0057] For example, please refer to Figure 3 , Figure 3 This is an example diagram showing the relative weights of the aforementioned key identification features calculated using the SHAP method. For example... Figure 3 As shown, VCI and PSI ranked first and second, contributing 23.83% and 22.87% to the SHAP value, respectively. Factors such as younger age, male sex, lower family income, and significant differences in VCI, PSI, PRI, and WMI-PSI are beneficial for the model to identify NDDs (HFASD and DD).

[0058] Please see Figure 4 , Figure 4 This application discloses an identification device for children with neurodevelopmental disorders. This device can be applied to any of the aforementioned electronic devices. Figure 4 As shown, the device includes: Acquisition unit 410 is used to acquire the demographic characteristics and intelligence index characteristics of the intellectually normal child to be identified; The recognition unit 420 is used to input the demographic characteristics and intelligence indicators of the child to be identified into the trained convergent recognition model; the convergent recognition model is any one of the following machine learning models: logistic regression model, random forest model, decision tree model or extreme gradient boosting model; The output unit 430 is used to obtain the target recognition result output by the convergent recognition model; the target recognition result indicates whether the child to be identified has a neurodevelopmental disorder.

[0059] Optionally, the convergent identification model can be a logistic regression model.

[0060] Optionally, the aforementioned demographic characteristics include at least: the child's age and sex, and the child's family income; the intelligence indicators include at least: verbal comprehension index, processing speed index, the difference between working memory index and processing speed index, and perceptual reasoning index.

[0061] Further optional demographic characteristics may include: the child's mother's education level and the child's father's education level; intelligence indicators may include: working memory index, the difference between verbal comprehension index and perceptual reasoning index, the difference between verbal comprehension index and working memory index, the difference between verbal comprehension index and processing speed index, the difference between perceptual reasoning index and working memory index, and the difference between perceptual reasoning index and processing speed index.

[0062] In some possible embodiments, the identification device for children with neurodevelopmental disorders may further include: Training unit 440 is used to train several initial sample models one by one using the training dataset to obtain several well-trained converged sample models; the several initial sample models are all binary classification machine learning models; the training dataset includes the demographic characteristics and intelligence index characteristics of several first sample children. Evaluation unit 450 is used to test the recognition performance of each converged sample model in identifying children with neurodevelopmental disorders using the first test dataset, and to obtain the score of each converged sample model under the evaluation index; the first test dataset includes the demographic characteristics and intelligence index characteristics of several second sample children; the evaluation index includes one or more of the following indices: area under the curve, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and Cohen's Kappa value; The first determining unit 460 is used to compare the scores of each convergent sample model under the evaluation index and determine a convergent sample model as the trained convergent recognition model.

[0063] Optionally, the identification device for children with neurodevelopmental disorders may also include: The second determining unit 470 is used to obtain a logistic stepwise regression model and, using the first test dataset and the second test dataset, combine the logistic stepwise regression model to determine key identification features from demographic features and intelligence indicator features. Among them, the logistic stepwise regression model is constructed using the logistic regression model and the logistic stepwise regression method; the second test dataset includes the demographic and intelligence indicators of several third sample children; key identification features include: the children's age and gender, and the children's family income in the demographic features; the differences between the verbal comprehension index, processing speed index, working memory index and processing speed index in the intelligence indicators, as well as the perceptual reasoning index.

[0064] Further optionally, the second determining unit 470 can also be used to calculate the importance of each key identifying feature in identifying children with neurodevelopmental disorders using the Shapley additive interpretation method.

[0065] It should be noted that the acquisition unit 410, recognition unit 420, and output unit 430 mentioned above can be referred to as the model application part of the recognition device for children with neurodevelopmental disorders, and the training unit 440, evaluation unit 450, first determination unit 460, and second determination unit 470 can be referred to as the model training part of the recognition device for children with neurodevelopmental disorders. The model application part and the model training part of this device can be physically arranged in different computing devices, or they can be integrated in the same computing device; there is no specific limitation.

[0066] It is understandable that the convergent recognition model determined by the first determining unit 460 is the model used by the recognition unit 420 during recognition; the key recognition features determined by the second determining unit 470 can be used to determine the specific information included in the demographic features and intelligence indicator features acquired by the acquisition unit 510.

[0067] The embodiments of this application disclose an identification device for children with neurodevelopmental disorders. This device can automatically identify whether intellectually normal children have neurodevelopmental problems by combining their demographic and intellectual indicators. This facilitates early identification and clinical screening of high-risk intellectually normal children. Furthermore, the embodiments of this application significantly reduce reliance on human experience and are applicable to all clinical outpatient clinics where intellectual assessment can be performed.

[0068] Please see Figure 5 , Figure 5 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application. For example... Figure 5 As shown, the testing equipment may include: Memory 510 containing computer programs; Processor 520 coupled to memory 510; When the computer program stored in the memory 510 is executed by the processor 520, the processor 520 performs any of the identification methods for children with neurodevelopmental disorders disclosed in the embodiments of this application.

[0069] It should be noted that the electronic device shown in the figure may also include components not shown, such as a power supply, input buttons, screen, RF circuit, Wi-Fi module, and Bluetooth module, which will not be described in detail in this embodiment.

[0070] This application discloses a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements any of the methods for identifying children with neurodevelopmental disorders disclosed in this application.

[0071] This application discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute any of the neurodevelopmental disorder identification methods disclosed in this application.

[0072] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Those skilled in the art should also recognize that the embodiments described in the specification are optional embodiments, and the actions and modules involved are not necessarily essential to this application.

[0073] In the various embodiments of this application, it should be understood that the sequence number of each process does not necessarily imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0074] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they can be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0075] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0076] If the integrated units described above are implemented as software functional units and sold or used as independent products, they can be stored in a computer-accessible memory. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several requests to cause a computer device (which can be a personal computer, server, or network device, specifically a processor in the computer device) to execute some or all of the steps of the methods described in the various embodiments of this application.

[0077] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compactdisc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0078] The foregoing has provided a detailed description of the methods, devices, and electronic devices for identifying children with neurodevelopmental disorders disclosed in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for identifying children with neurodevelopmental disorders, characterized in that, The method includes: Obtain demographic and intelligence indicators of the intellectually normal children to be identified; The demographic and intelligence characteristics of the child to be identified are input into the trained convergent recognition model; the convergent recognition model is any one of the following machine learning models: logistic regression model, random forest model, decision tree model, or extreme gradient boosting model; Obtain the target recognition result output by the convergent recognition model; the target recognition result indicates whether the child to be identified has a neurodevelopmental disorder.

2. The method according to claim 1, characterized in that, The convergence identification model is a logistic regression model.

3. The method according to claim 1, characterized in that, The demographic characteristics include at least the child's age and sex, and the child's family income; the intelligence indicators include at least the verbal comprehension index, processing speed index, the difference between the working memory index and the processing speed index, and the perceptual reasoning index.

4. The method according to claim 3, characterized in that, The demographic characteristics also include: the mother's education level and the father's education level; the intelligence indicators also include: working memory index, the difference between the total IQ verbal comprehension index and the perceptual reasoning index, the difference between the verbal comprehension index and the working memory index, the difference between the verbal comprehension index and the processing speed index, the difference between the perceptual reasoning index and the working memory index, and the difference between the perceptual reasoning index and the processing speed index.

5. The method according to any one of claims 1-4, characterized in that, Before inputting the demographic and intelligence indicators of the child to be identified into the trained convergent recognition model, the method further includes: Several initial sample models are trained one by one using the training dataset to obtain several well-trained convergent sample models; the several initial sample models are all binary classification machine learning models; the training dataset includes the demographic characteristics and intelligence index characteristics of several first sample children; The recognition performance of each convergent sample model in identifying children with neurodevelopmental disorders was tested using the first test dataset, and the scores of each convergent sample model under the evaluation index were obtained. The first test dataset includes the demographic characteristics and intelligence index characteristics of several second sample children. The evaluation index includes one or more of the following indices: area under the curve, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and Cohen's Kappa value. By comparing the scores of each convergent sample model under the evaluation index, one of the convergent sample models is determined as the trained convergent recognition model.

6. The method according to claim 5, characterized in that, The method further includes: Obtain a logistic stepwise regression model, which is constructed using a logistic regression model and a logistic stepwise regression method; Using the first test dataset and the second test dataset, and combining the logistic stepwise regression model, key identification features are determined from the demographic features and the intelligence index features; The second test dataset includes demographic and intelligence indicators of several third sample children; the key identification features include: the children's age and gender, and the children's family income in the demographic features; the differences between the verbal comprehension index, processing speed index, working memory index and processing speed index in the intelligence indicators, and the perceptual reasoning index.

7. The method according to claim 6, characterized in that, The method further includes: The importance of each of the key identifying features in identifying children with neurodevelopmental disorders was calculated using the Shapley interpretation method.

8. A device for identifying children with neurodevelopmental disorders, characterized in that, The device includes: The acquisition unit is used to acquire the demographic characteristics and intelligence indicators of the intellectually normal children to be identified. The identification unit is used to input the demographic characteristics and intelligence index characteristics of the child to be identified into a trained convergent identification model; the convergent identification model is any one of the following machine learning models: logistic regression model, random forest model, decision tree model, or extreme gradient boosting model; The output unit is used to obtain the target recognition result output by the convergent recognition model; the target recognition result indicates whether the child to be identified has a neurodevelopmental disorder.

9. The apparatus according to claim 8, characterized in that, The device includes: The training unit is used to train several initial sample models one by one using the training dataset to obtain several well-trained converged sample models; the several initial sample models are all binary classification machine learning models; the training dataset includes the demographic characteristics and intelligence index characteristics of several first sample children. The evaluation unit is used to test the recognition performance of each converged sample model in identifying children with neurodevelopmental disorders using a first test dataset, and to obtain the score of each converged sample model under the evaluation index; the first test dataset includes the demographic characteristics and intelligence index characteristics of several second sample children; the evaluation index includes one or more of the following indices: area under the curve, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and Cohen's Kappa value. The first determining unit is used to determine a convergent sample model as a trained convergent recognition model based on the scores of each convergent sample model under the evaluation index.

10. An electronic device, characterized in that, The electronic device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to implement the method as described in any one of claims 1 to 7.

Images