A medical experiment data processing method and system

Abstract

The application provides a medical experiment data processing method and system, which realizes intelligent integration and efficient processing of multi-modal medical experiment data by introducing an adaptive mixed model, has the characteristics of data preprocessing automation, multi-dimensional feature extraction, adaptive model selection and strong visual interaction. The system can automatically identify and adapt to different types of data processing algorithms, effectively improving the accuracy and efficiency of data processing; through multi-modal feature fusion and time alignment technology, cross-data source correlation analysis is realized; at the same time, user-defined view and real-time feedback optimization are supported, which significantly enhances the practicability and flexibility of the system, provides comprehensive, intuitive and efficient data analysis support for medical researchers, and effectively promotes the deep mining and scientific discovery of medical experiment data.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for processing medical experimental data. Background Technology

[0002] With the rapid development of medical research technology, the types and scale of experimental data have exploded. Modern medical experiments involve multiple data types, including structured numerical data from biochemical analyzers, image data from equipment such as microscopes, CT scanners, and MRI scanners, and text data from clinical observations and experimental records. These data are typically multimodal, high-dimensional, and highly heterogeneous, posing significant challenges to subsequent data processing and analysis.

[0003] In current technologies, the processing of medical experimental data largely relies on traditional statistical software or single-function processing tools. For example, researchers often use tools like SPSS and SAS to process numerical data, and software like ImageJ and MATLAB to analyze image data, while text data may be processed through manual reading or basic natural language processing tools. This fragmented processing approach leads to difficulties in data integration, low efficiency, and challenges in achieving correlation analysis between multimodal data. Furthermore, existing methods often employ fixed algorithm models, lacking the ability to adaptively adjust to data characteristics and failing to meet the processing needs of different types of experimental data. For instance, when processing time series data, traditional methods often ignore the temporal asynchronicity of the data, leading to biased analysis results; when processing high-dimensional image data, insufficient feature extraction often affects model accuracy.

[0004] Another problem with existing technologies is the limitation of visualization. Most systems only support the visualization of single-type data, such as line charts, bar charts, or standalone images, lacking the ability to integrate and compare multimodal data on the same interface. This makes it difficult for researchers to intuitively understand the intrinsic relationships between data and to quickly discover potential patterns. In addition, existing systems usually lack user-interactive operation capabilities, and cannot dynamically adjust data processing strategies and visualization methods according to research needs, limiting the flexibility and depth of research.

[0005] Therefore, there is an urgent need in this field for a comprehensive processing method and system that can integrate multimodal medical experimental data, has adaptive processing capabilities, and supports intelligent analysis and interactive visualization, in order to improve the efficiency and accuracy of medical research. Summary of the Invention

[0006] The purpose of this invention is to provide a medical experimental data processing method and system to solve the problems existing in the prior art.

[0007] To achieve the above objectives, the present invention provides the following solution:

[0008] This invention provides a method for processing medical experimental data, the method comprising:

[0009] S1. Acquire multimodal experimental data from different medical experimental devices, wherein the multimodal experimental data includes structured numerical data, medical image data, and text record data;

[0010] S2. Preprocess the multimodal experimental data, including data cleaning, missing value imputation, outlier removal, and data standardization;

[0011] S3. Extract multi-dimensional features from the preprocessed data, including numerical statistical features, image texture features, and text keyword features;

[0012] S4. Construct and train an adaptive hybrid model based on the extracted features. The adaptive hybrid model includes multiple sub-models and a meta-learner. The meta-learner is used to dynamically select the optimal combination of sub-models.

[0013] S5. Use the trained adaptive hybrid model to perform classification, regression, or predictive analysis on medical experimental data;

[0014] S6. Generate a visualization view based on the analysis results. The visualization view is used to simultaneously display numerical trend charts, medical images, and text summaries.

[0015] S7. Receive data processing instructions input by the user, display the visualization view on a preset interface according to the instructions, and support dynamic configuration of view type, display position and display size.

[0016] Preferably, in step S2, the missing value filling is performed using interpolation, with the following formula:

[0017] ;

[0018] in, For fill value, For adjacent data points, This represents the number of adjacent points.

[0019] The outlier removal method uses the Z-score method to detect outliers, and the formula is as follows:

[0020] ;

[0021] in, For standard scores, if | | > 3 is considered an outlier. For data points, The mean, Standard deviation;

[0022] The data standardization uses the Min-Max method for data normalization, with the following formula:

[0023] ;

[0024] in, and These represent the minimum and maximum values ​​of the data, respectively.

[0025] Preferably, step S2 specifically includes:

[0026] S31. Extract the mean, variance, skewness, and kurtosis from numerical data;

[0027] S32. Use a convolutional neural network (CNN) to extract deep features from medical image data, or use a gray-level co-occurrence matrix (GLCM) to extract texture features;

[0028] S33. Use the TF-IDF method to extract keyword weights from text data;

[0029] S34. Principal Component Analysis (PCA) is used to reduce the dimensionality of the features. The formula is as follows:

[0030] ;

[0031] in, The original feature matrix, The eigenvector matrix, This is the feature matrix after dimensionality reduction.

[0032] Preferably, step S4 specifically includes:

[0033] S41. Train multiple sub-models separately, including Support Vector Machine (SVM), Random Forest, and CNN;

[0034] S42. Use a meta-learner to integrate the prediction results of the sub-models. The meta-learner outputs a weight vector w=[w1,w2,…,wm]. The final prediction result is:

[0035] ;

[0036] in, The total number of sub-models. For the first The weight coefficients of each sub-model For the first The predicted output of each sub-model;

[0037] S43. Optimize the meta-learner using the cross-entropy loss function, the formula is:

[0038] ;

[0039] in, For real labels, To predict probabilities, This represents the number of samples.

[0040] Preferably, the method further includes:

[0041] S8. Based on the alarm threshold range input by the user, highlight parameter values ​​that exceed the range in the visualization view and generate alarm information; the user can drag and drop the visualization view to adjust the display position and scale the display size; in response to the user's input data filtering or sorting commands, dynamically update the data in the visualization view.

[0042] Preferably, the method further includes:

[0043] S9. Perform time alignment processing on data from different devices, including calculating time offsets and calibrating timestamps; generate a multi-source data comparison view based on the aligned time axis to show the data differences under different experimental conditions.

[0044] The present invention also provides a medical experimental data processing system, the system comprising:

[0045] The data input module is used to acquire multimodal medical experimental data;

[0046] The data preprocessing module is used to clean, fill, remove, and standardize the data.

[0047] The feature extraction module is used to extract numerical, image, and text features from the data.

[0048] The model training module is used to build and train adaptive hybrid models;

[0049] The data analysis module is used to perform data classification, regression, or prediction using adaptive hybrid models;

[0050] The output module is used to generate and display visual views;

[0051] The control module is used to receive user commands and coordinate the operation of various modules.

[0052] Preferably, the system further includes:

[0053] The feedback module is used to receive user ratings of the analysis results and update the model parameters using an online learning algorithm based on the ratings.

[0054] Preferably, the data input module automatically acquires data from the electronic medical record system, laboratory information system, and medical imaging equipment, and parses it into a unified format.

[0055] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein when the computer program is executed by the processor, the medical experimental data processing method described above is implemented.

[0056] The present invention achieves the following beneficial technical effects compared to the prior art:

[0057] This invention provides a medical experimental data processing method and system that achieves intelligent integration and efficient processing of multimodal medical experimental data by introducing an adaptive hybrid model. It features automated data preprocessing, multi-dimensional feature extraction, adaptive model selection, and strong visualization interactivity. The system can automatically identify and adapt processing algorithms for different types of data, effectively improving the accuracy and efficiency of data processing. Through multimodal feature fusion and time alignment technology, it achieves cross-data source correlation analysis. Simultaneously, it supports user-defined views and real-time feedback optimization, significantly enhancing the system's practicality and flexibility. This provides medical researchers with comprehensive, intuitive, and efficient data analysis support, powerfully promoting in-depth mining and scientific discovery of medical experimental data. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 The present invention provides a flowchart of a medical experimental data processing method. Detailed Implementation

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0061] The purpose of this invention is to provide a method and system for processing medical experimental data, aiming to solve the problems of difficulty in integrating multimodal medical experimental data, low processing efficiency, insufficient analytical accuracy, and imperfect visualization in existing technologies. By introducing an Adaptive Hybrid Model (AHM) and an intelligent preprocessing workflow, unified processing and in-depth analysis of structured numerical data, medical image data, and text record data are achieved.

[0062] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0063] Example 1:

[0064] The medical experimental data processing method provided in this embodiment, in its specific implementation process, such as... Figure 1 As shown, the system first acquires multimodal experimental data from various medical experimental devices through the data input module. This data may originate from structured numerical data generated by equipment such as biochemical analyzers, microscopes, CT scanners, and MRI machines, as well as medical image data and text data such as clinical observation records. During data acquisition, the system supports automatic data extraction from electronic medical record systems (EMR), laboratory information systems (LIS), and medical image archiving and communication systems (PACS), and parses the data into a standardized format through a unified data interface to ensure data integrity and consistency.

[0065] After acquiring the raw data, the system enters the data preprocessing stage. This stage includes steps such as data cleaning, missing value imputation, outlier removal, and data standardization. For missing value handling, the system uses interpolation to impute missing values; the specific formula is as follows:

[0066] ;

[0067] in, For fill value, For adjacent data points, This represents the number of adjacent points. This method effectively recovers missing values ​​by utilizing local data trends, avoiding analytical biases caused by missing data. For outlier detection, the system uses the Z-score method to identify anomalous data points; the calculation formula is:

[0068] ;

[0069] in, Standard score, For data points, The mean, The standard deviation is | When |>3, the system marks the data point as an outlier and automatically removes it. To further improve data quality, the system uses the Min-Max method for data normalization, mapping the data to the [0,1] interval, as shown in the formula:

[0070] ;

[0071] in, The original data point values ​​to be detected. and These represent the minimum and maximum values ​​of the data, respectively. This process eliminates the influence of different units on the analysis results, laying the foundation for subsequent feature extraction and model training.

[0072] After data preprocessing, the system enters the feature extraction stage. This stage extracts multi-dimensional features from the preprocessed data, including numerical statistical features, image texture features, and text keyword features. For numerical data, the system extracts statistics such as mean, variance, skewness, and kurtosis to characterize the data's distribution characteristics. For medical image data, the system provides two feature extraction methods: one is to use a convolutional neural network (CNN) to extract deep features, capturing the abstract semantic information of the image through multi-layer convolution and pooling operations; the other is to use a gray-level co-occurrence matrix (GLCM) to extract texture features, including parameters such as contrast, correlation, energy, and homogeneity, quantifying local image patterns. For text data, the system uses the TF-IDF (term frequency-inverse document frequency) method to extract keyword weights, highlighting terms with discriminative power. To reduce feature dimensionality and computational complexity, the system further uses principal component analysis (PCA) to perform dimensionality reduction on the features, with the formula:

[0073] ;

[0074] in, This is the original feature matrix, with dimensions n×m (where m is the number of original features). The eigenvector matrix is ​​composed of the eigenvectors of the original data covariance matrix, and has dimensions m×k. This is the feature matrix after dimensionality reduction, with dimensions n×k (n is the number of samples, and k is the number of features after dimensionality reduction). This process significantly improves the efficiency of subsequent model training while retaining most of the original information.

[0075] During the model building and training phase, the system constructs an Adaptive Hybrid Model (AHM) based on the extracted features. This model consists of multiple sub-models (such as Support Vector Machines (SVM), Random Forests, CNNs, etc.) and a meta-learner. First, the system trains each sub-model separately to achieve optimal performance on specific data types or tasks. Then, the meta-learner analyzes the prediction results of each sub-model, dynamically allocates weight coefficients, and forms the final prediction output. Specifically, the meta-learner outputs a weight vector w = [w1, w2, ..., wm], and the final prediction result is:

[0076] ;

[0077] in, The total number of sub-models. For the first The weight coefficients of each sub-model For the first The predicted output of each sub-model. To optimize the meta-learner, the system uses the cross-entropy loss function for training, as shown in the formula:

[0078] ;

[0079] in, The true label (usually 0 or 1). To predict probabilities, the range is [0,1]. Let be the number of samples. By minimizing the loss function, the meta-learner can adaptively select the optimal combination of sub-models, improving the accuracy and robustness of the overall model.

[0080] During the data analysis phase, the system utilizes a pre-trained adaptive hybrid model to perform classification, regression, or predictive analysis on medical experimental data. For example, in disease diagnosis scenarios, the model can classify patient data and output disease type probabilities; in drug efficacy prediction, the model can perform regression analysis to predict the therapeutic effects of specific drugs. The system also supports multi-source data comparison analysis. By performing time alignment processing on data from different devices, calculating time offsets, and calibrating timestamps, it ensures the comparability of data on a unified timeline. Based on the aligned timeline, the system generates a multi-source data comparison view, intuitively displaying the differences in data under different experimental conditions, helping researchers discover potential patterns.

[0081] During the visualization phase, the system generates interactive visualization views based on the analysis results. These views simultaneously display numerical trend charts (such as line charts and bar charts), medical images (such as CT scans and microscopic images), and text summaries (such as key diagnostic conclusions). Users can dynamically configure the view type, display position, and display size through a preset interface, and drag-and-drop adjustments and zooming are supported. Furthermore, the system allows users to set alarm threshold ranges; when parameter values ​​exceed these ranges, the view automatically highlights and generates alarm information to alert researchers to anomalies. The system also responds to data filtering or sorting commands, dynamically updating the visualization content to meet personalized analysis needs.

[0082] To continuously optimize system performance, this invention also integrates a feedback module. This module receives user ratings of the analysis results and updates the model parameters using an online learning algorithm based on those ratings. The parameter update formula is:

[0083] ;

[0084] in, This represents the updated model parameters; This represents the model parameters before the update; This represents the learning rate, which controls the step size for parameter updates. This represents the gradient of the loss function with respect to the parameters, pointing in the direction of the fastest growth of the loss function. This mechanism enables the system to adapt to changes in user needs and the evolution of data distribution, maintaining the timeliness and accuracy of the model.

[0085] Example 2:

[0086] In terms of system architecture, the medical experimental data processing system provided in this embodiment includes a data input module, a data preprocessing module, a feature extraction module, a model training module, a data analysis module, an output module, a control module, and a feedback module. These modules are connected through standardized interfaces to form an efficient data processing pipeline. The control module is responsible for coordinating the operation of each module, receiving user instructions, and scheduling resources. The system can be deployed on a local server or a cloud platform, supports concurrent access by multiple users, and possesses good scalability and stability.

[0087] Example 3:

[0088] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory. When the computer program is executed by the processor, it implements the aforementioned medical experimental data processing method. This electronic device can be a general-purpose computer, a server, or a dedicated medical device to meet the application needs of different scenarios.

[0089] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0090] It should be noted that the components mentioned in the above embodiments are all general standard parts or components known to those skilled in the art. Their structures and principles can be learned by those skilled in the art through technical manuals or conventional experimental methods.

[0091] This invention has illustrated its principles and implementation methods using specific examples. The descriptions of these embodiments are merely illustrative of the method and its core ideas; furthermore, those skilled in the art will recognize that modifications may be made to the specific implementation methods and application scope based on the principles of this invention. Therefore, the content of this specification should not be construed as limiting the invention.

Claims

1. A medical experiment data processing method characterized by comprising: The method includes: S1. Acquire multimodal experimental data from different medical experimental devices, wherein the multimodal experimental data includes structured numerical data, medical image data, and text record data; S2. Preprocess the multimodal experimental data, including data cleaning, missing value imputation, outlier removal and data standardization; S3. Extract multi-dimensional features from the preprocessed data, including numerical statistical features, image texture features, and text keyword features; S4. Construct and train an adaptive hybrid model based on the extracted features. The adaptive hybrid model includes multiple sub-models and a meta-learner. The meta-learner is used to dynamically select the optimal combination of sub-models. S5. Use the trained adaptive hybrid model to perform classification, regression, or predictive analysis on medical experimental data; S6. Generate a visualization view based on the analysis results. The visualization view is used to simultaneously display numerical trend charts, medical images, and text summaries. S7. Receive data processing instructions input by the user, display the visualization view on a preset interface according to the instructions, and support dynamic configuration of view type, display position and display size.

2. The medical experimental data processing method according to claim 1, characterized in that, In step S2, the missing value filling is performed using interpolation, with the following formula: ； wherein, is a fill value, is a neighboring data point, is a number of neighboring points; The outlier removal method uses the Z-score method to detect outliers, and the formula is as follows: ； where, is the standard score, if is considered an outlier, is the data point, is the mean, is the standard deviation; The data standardization uses the Min-Max method for data normalization, with the following formula: ； wherein and are the minimum and maximum data values, respectively.

3. The medical experimental data processing method of claim 1, wherein, Step S2 specifically includes: S31. Extract the mean, variance, skewness and kurtosis from numerical data; S32. Use a convolutional neural network (CNN) to extract deep features from medical image data, or use a gray-level co-occurrence matrix (GLCM) to extract texture features. S33. Use the TF-IDF method to extract keyword weights from text data; S34. Principal Component Analysis (PCA) is used to reduce the dimensionality of the features. The formula is as follows: ； wherein, is the original feature matrix, is the feature vector matrix, is the reduced feature matrix.

4. The medical experimental data processing method of claim 1, wherein, Step S4 specifically includes: S41. Train multiple sub-models, including Support Vector Machine (SVM), Random Forest, and CNN. S42, integrating the sub-model prediction results using the meta-learner, the meta-learner outputs a weight vector w = [w1, w2, …, w m ], and the final prediction result is: ； wherein, is the total number of sub-models, is the weight coefficient of the th sub-model, is the prediction output of the th sub-model; S43. Optimize the meta-learner using the cross-entropy loss function, the formula is: ； wherein, is the true label, is the predicted probability, is the number of samples.

5. The medical experimental data processing method according to claim 1, characterized in that, The method further includes: S8. Based on the alarm threshold range input by the user, highlight parameter values ​​that exceed the range in the visualization view and generate alarm information; the user can drag and drop the visualization view to adjust the display position and scale the display size; in response to the user's input data filtering or sorting commands, dynamically update the data in the visualization view.

6. The medical experimental data processing method according to claim 1, characterized in that, The method further includes: S9. Perform time alignment processing on data from different devices, including calculating time offsets and calibrating timestamps; generate a multi-source data comparison view based on the aligned time axis to show the data differences under different experimental conditions.

7. A medical experimental data processing system, characterized in that, The system includes: The data input module is used to acquire multimodal medical experimental data; The data preprocessing module is used to clean, fill, remove, and standardize the data. The feature extraction module is used to extract numerical, image, and text features from the data. The model training module is used to build and train adaptive hybrid models; The data analysis module is used to perform data classification, regression, or prediction using adaptive hybrid models; The output module is used to generate and display visual views; The control module is used to receive user commands and coordinate the operation of various modules.

8. The medical experimental data processing system according to claim 7, characterized in that, The system also includes: The feedback module is used to receive user ratings of the analysis results and update the model parameters using an online learning algorithm based on the ratings.

9. The medical experimental data processing system according to claim 7, characterized in that, The data input module automatically acquires data from the electronic medical record system, laboratory information system, and medical imaging equipment, and parses it into a unified format.

10. An electronic device, characterized in that, The system includes a memory, a processor, and a computer program stored in the memory, which, when executed by the processor, implements the medical experimental data processing method as described in any one of claims 1 to 6.

Images