Risk assessment system and method based on gas marker detection
By designing a risk assessment system based on gas marker detection, using the sensor matrix and CNN-TCN model of cross-reaction inhibition strategy, the problem of insufficient accuracy of traditional gastrointestinal disease diagnosis methods is solved, and more accurate risk assessment results are achieved.
Patent Information
- Application Number
- CN202411488318.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-24
- Publication Date
- 2025-02-25
AI Technical Summary
Traditional gastrointestinal disease diagnosis methods are insufficient in accuracy and timeliness, and gas marker detection systems face problems such as data reliability, completeness and sample deviation, which affects the accuracy of the evaluation results.
Design a risk assessment system based on gas marker detection, including gas marker detection components, data preprocessing components and risk assessment components. The system uses a sensor matrix based on cross-reaction suppression strategy for detection, combining data cleaning, correction, denoising, resampling and standardization processing, and finally uses the CNN-TCN model for feature extraction and risk assessment.
By improving the accuracy and reliability of detection data, we provide more accurate results for gastrointestinal disease risk assessment, helping decision makers obtain more reliable references.
Smart Images
Figure CN119517228A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent assessment technology, and more specifically to a risk assessment system and method based on gas marker detection. Background Art
[0002] In the field of modern medicine, early screening and accurate assessment of gastrointestinal diseases have always been the focus of clinical attention. However, traditional diagnostic methods often rely on symptom description, physical examination and imaging, and the accuracy and timeliness of these methods are often limited by many factors, such as the subjective description bias of patients, the hidden nature of the disease and the limitations of imaging technology. In recent years, with the development of biomarker detection technology, especially the detection of gas markers, new ideas have been provided for the diagnosis of gastrointestinal diseases.
[0003] The concentration and type of gas markers, such as volatile organic compounds (VOCs) in exhaled breath, are closely related to human metabolism and disease status. By detecting the changes in these gas markers, non-invasive and rapid screening and evaluation of digestive tract diseases can be achieved. However, in practical applications, gas marker detection and intelligent assessment systems for digestive tract diseases face a series of technical challenges, such as data reliability and integrity and data sample bias in data sampling, and accuracy issues in the evaluation network, which all affect the accuracy of the results. Therefore, how to solve the above problems is an urgent problem for technicians in this field. Summary of the invention
[0004] In view of this, the present invention provides a risk assessment system and method based on gas marker detection to overcome the above problems.
[0005] In order to achieve the above object, the present invention adopts the following technical solution:
[0006] A risk assessment system based on gas marker detection, comprising:
[0007] A gas marker detection component, used to collect test samples according to the evaluation indicators;
[0008] A data preprocessing component, used to preprocess the test sample and obtain preprocessing data;
[0009] The risk assessment component is used to perform feature extraction and risk assessment on the preprocessed data to obtain risk data.
[0010] Optionally, the gas marker detection component adopts a sensor matrix based on a cross-reaction suppression strategy.
[0011] Optionally, the data preprocessing component includes a data cleaning module, a data correction module, a data denoising module, a data resampling module and a data standardization module;
[0012] The data cleaning module is used to identify and process outliers in the test samples;
[0013] The data correction module is used to correct the cleaned data according to the compensation strategy;
[0014] The data denoising module is used to reduce signal noise in the corrected data based on a filtering algorithm;
[0015] The data resampling module is used to adjust the sampling interval or the number of sampling points of the denoised data;
[0016] The data standardization module is used to perform standardization processing on the resampled data to obtain preprocessed data.
[0017] Optionally, the risk assessment component includes:
[0018] A feature extraction module, used for extracting features of the preprocessed data in terms of spatial distribution and time series;
[0019] A risk assessment module, used to assess the risk based on the extracted features;
[0020] A report generation module is used to generate relevant text according to the risk data.
[0021] Optionally, the feature extraction module includes a local feature extraction module, a temporal feature extraction module and a feature fusion module;
[0022] The local feature extraction module is used to extract local features based on the preprocessed data;
[0023] The temporal feature extraction module is used to extract temporal features based on the local features;
[0024] The feature fusion module is used to fuse the multiple time series features to obtain the extracted features.
[0025] A risk assessment method based on gas marker detection, the specific steps are:
[0026] Acquire a test sample according to the evaluation index, and preprocess the test sample to obtain preprocessed data;
[0027] Constructing a risk assessment model based on a neural network, and inputting the preprocessed data into the risk assessment model for risk assessment;
[0028] Used to generate reports based on risk assessment results.
[0029] Optionally, the risk assessment model is a CNN-TCN model, comprising an input layer, a CNN layer, a TCN layer, a fully connected layer and an output layer;
[0030] The input layer is used to obtain the preprocessed data;
[0031] The CNN layer is used to extract local features of the preprocessed data;
[0032] The TCN layer is used to extract temporal features based on the local features;
[0033] The fully connected layer is used to map the time series features to the decision space of risk assessment;
[0034] The output layer is used to output risk assessment data.
[0035] Optionally, an attention mechanism layer and a feature fusion layer are further provided between the TCN layer and the fully connected layer;
[0036] The attention mechanism layer is used for weight matching;
[0037] The feature fusion layer is used to fuse the outputs of the TCN layer and the attention mechanism layer to obtain fused features.
[0038] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a risk assessment system and method based on gas marker detection, which adopts a sensor matrix based on a cross-reaction suppression strategy. Multiple sensors in the sensor matrix cooperate with each other and utilize the cross-reaction suppression effect to accurately distinguish multiple gas components in a complex environment and reduce the interference of non-target gases, thereby greatly improving the accuracy and reliability of the detection data. This improvement in accuracy lays a solid foundation for subsequent risk assessment. The model makes full use of the advantages of convolutional neural networks (CNNs) and temporal convolutional networks (TCNs). The CNN-TCN model is also used in the present invention. The CNN layer can deeply mine the local spatial characteristics of gas marker data, such as concentration distribution, change trends, etc.; while the TCN layer can capture the temporal dependence in the data and analyze the change law of gas markers over time. The combination of the two enables the model to construct a more complex and comprehensive feature representation, thereby improving the accuracy of the assessment results. This improvement in accuracy helps to more accurately predict and evaluate potential risk factors and provide a more reliable reference for decision makers. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0040] Figure 1 A schematic diagram of the system structure provided by the present invention;
[0041] Figure 2 The present invention provides a flow chart of the method. DETAILED DESCRIPTION
[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0043] In one aspect, the present invention includes a risk assessment system based on gas marker detection, such as Figure 1 As shown, including:
[0044] A gas marker detection component, used to collect test samples according to the evaluation indicators;
[0045] A data preprocessing component, used to preprocess the test sample and obtain preprocessing data;
[0046] The risk assessment component is used to perform feature extraction and risk assessment on the preprocessed data to obtain risk data.
[0047] Furthermore, the present embodiment is applied to assessing the risk of gastrointestinal diseases, and the assessment indicators are obtained based on experimental methods, specifically: collecting and analyzing the respiratory gases of healthy patients and patients with different gastrointestinal diseases, obtaining gas markers related to gastrointestinal diseases, such as hydrogen, methane, nitric oxide, volatile sulfides, ethanol, volatile organic compounds (such as benzene, toluene, etc.), etc., constructing assessment indicators based on the gas markers, and performing correlation analysis between the obtained assessment indicators and various gastrointestinal diseases, to obtain assessment indicators for various gastrointestinal diseases, such as indicators related to ulcerative colitis including nitric oxide, volatile organic compounds, etc.
[0048] In one embodiment, the gas marker detection component adopts a sensor matrix based on a cross-reaction suppression strategy. Specifically, sensors with higher sensitivity to each indicator are selected according to the evaluation indicators, and in order to reduce cross-reactions, sensors that may produce cross-reactions are separated. If they cannot be separated, physical means such as baffles are used for isolation, or a layer of sensitive material with high selectivity for a specific gas is coated on the surface of the sensor. The material can react specifically with the target gas, but is insensitive or reacts weakly to other gases. The layout method of the sensor matrix is as follows: the detection area is divided based on the number of evaluation indicators and the detection range and sensitivity of each sensor. After the division is completed, the sensor orientation is adjusted based on the diffusion direction and speed of the measured gas to improve the detection accuracy. In this embodiment, the sensor matrix is divided into four layers, the first layer is provided with VOCs sensors; the second layer is provided with hydrogen sensors and methane sensors; the third layer is provided with nitric oxide sensors and volatile sulfide sensors; the fourth layer is provided with ethanol sensors; wherein, hydrogen sensors: select electrochemical sensors or metal oxide semiconductor (MOS) sensors; methane sensors: use thermal conductivity sensors, catalytic combustion sensors or specific MOS sensors; nitric oxide sensors: select electrochemical sensors or chemiluminescence sensors; volatile sulfide sensors: select composite sensors or a combination of multiple sensors to cover different types of sulfides; ethanol sensors: select electrochemical sensors or MOS sensors; VOCs sensors: use PID (photoionization detector) or MOS sensor arrays.
[0049] In one embodiment, the gas marker detection component also includes a primary processing module for performing preliminary processing such as amplification and filtering on the collected signal to ensure the stability and accuracy of the signal.
[0050] In one embodiment, the data preprocessing component includes a data cleaning module, a data correction module, a data denoising module, a data resampling module, and a data standardization module;
[0051] Among them, the data cleaning module is used to identify and process outliers in the test samples, specifically: deleting invalid data, filling missing values, correcting erroneous data, etc.
[0052] A data correction module, used to correct the cleaned data according to a compensation strategy, wherein the compensation strategy includes a temperature compensation strategy and a humidity compensation strategy. The temperature compensation strategy is a mathematical model established based on the relationship between the temperature and output of the sensor, and the humidity compensation strategy is also the same;
[0053] A data denoising module, used to reduce signal noise in the correction data based on a filtering algorithm;
[0054] Data resampling module, used to adjust the sampling interval or number of sampling points of denoised data;
[0055] The data standardization module is used to standardize the resampled data to obtain preprocessed data.
[0056] In one embodiment, the risk assessment component includes:
[0057] A feature extraction module, used for extracting features of the preprocessed data in terms of spatial distribution and time series;
[0058] A risk assessment module, used to assess the risk based on the extracted features;
[0059] A report generation module is used to generate relevant text according to the risk data.
[0060] In one embodiment, the feature extraction module includes a local feature extraction module, a temporal feature extraction module and a feature fusion module;
[0061] Wherein, the local feature extraction module is used to extract local features based on the preprocessed data;
[0062] The temporal feature extraction module is used to extract temporal features based on the local features;
[0063] The feature fusion module is used to fuse the multiple time series features to obtain the extracted features.
[0064] This embodiment also discloses a risk assessment method based on gas marker detection, such as Figure 2 As shown, the specific steps are:
[0065] Step 1: obtaining a test sample according to the evaluation index, and preprocessing the test sample to obtain preprocessing data;
[0066] Step 2: construct a risk assessment model based on a neural network, and input the preprocessed data into the risk assessment model for risk assessment;
[0067] Step 3: Generate a report based on the risk assessment results.
[0068] In step 1, the evaluation index is obtained based on experiments. The specific steps are as follows: exhaled gas samples are collected from healthy people and patients with different digestive tract diseases (including gastritis, gastroesophageal reflux disease, Crohn's disease, ulcerative colitis, etc.) to compare and analyze the differences in gas markers among different populations, thereby constructing the evaluation index.
[0069] In one embodiment, the risk assessment model in step 2 is a CNN-TCN model, including an input layer, a CNN layer, a TCN layer, a fully connected layer and an output layer, wherein the input layer is used to obtain the preprocessed data; the CNN layer is used to extract local features of the preprocessed data; the TCN layer is used to extract time series features based on the local features; the fully connected layer is used to map the time series features to the decision space of risk assessment; the output layer is used to output risk assessment data;
[0070] Furthermore, a combination of multiple convolutional layers, pooling layers and activation functions is adopted in the CNN layer, wherein the convolutional layer adopts a multi-scale convolution block, and local patterns of different lengths are extracted by using convolution kernels of different sizes, thereby increasing the sensitivity of the model to data changes.
[0071] In one embodiment, depthwise separable convolution is introduced into the multi-scale convolution block to reduce computational complexity and the number of parameters.
[0072] Furthermore, the TCN layer includes causal convolution and dilated convolution to capture farther temporal dependencies in a larger receptive field.
[0073] Furthermore, skip connections are introduced in the TCN layer.
[0074] In one embodiment, an attention mechanism layer and a feature fusion layer are further provided between the TCN layer and the fully connected layer.
[0075] The attention mechanism layer is used for weight matching;
[0076] The feature fusion layer is used to fuse the outputs of the TCN layer and the attention mechanism layer to obtain fusion features to improve the accuracy of the prediction results.
[0077] In one embodiment, a Dropout layer is introduced in the fully connected layer to reduce overfitting of the model.
[0078] In one embodiment, after the risk assessment model is constructed, it is necessary to iteratively train the model until the preset conditions are met, and then stop the training. The specific steps are:
[0079] Construct a data set, identify the target groups for sample collection (such as healthy people, patients with different gastrointestinal diseases, etc.), collect exhaled gas samples from the subjects, and record relevant clinical information (such as age, gender, disease diagnosis, etc.).
[0080] Data preprocessing: cleaning, correction, denoising, resampling and standardization of collected data to obtain preprocessed data;
[0081] Feature extraction: Extract features that have an impact on risk assessment from preprocessed data, such as the concentration, ratio or combination of specific gas markers.
[0082] Data annotation: Define clear labels for each sample (such as healthy, gastritis, gastroesophageal reflux disease, etc.).
[0083] Dataset division: The labeled dataset is randomly divided into training set, validation set and test set. The training set is used for model training, the validation set is used to adjust model parameters (such as hyperparameter tuning), and the test set is used to evaluate model performance.
[0084] In order to ensure that the training set, validation set, and test set have similar distributions in terms of disease type, age, gender, etc., stratified sampling can be used for division.
[0085] Model training:
[0086] Initialize the model: Initialize the parameters of the constructed CNN-TCN model, and the initialized model parameters are randomly generated.
[0087] Training process: Input the training set data into the model, and obtain the prediction results through each layer of the model (such as CNN layer, TCN layer, fully connected layer, etc.); calculate the loss function (such as cross entropy loss, mean square error, etc.) based on the prediction results and the true label, and update the model parameters according to the loss function.
[0088] Iterative optimization: Repeat the training process until the preset conditions are met (such as loss function convergence, accuracy stability, reaching the maximum number of iterations, etc.).
[0089] Hyperparameter tuning: During training, model performance can be optimized by adjusting hyperparameters such as learning rate, batch size, optimizer type, etc. This usually requires multiple experiments on the validation set to find the best hyperparameter combination.
[0090] Visual report: In the report, risk levels are presented in an intuitive manner, such as using color coding (green for low risk, yellow for medium risk, and red for high risk) or radar charts. The report lists in detail the key factors that affect the risk assessment results, such as changes in the concentration of specific gas markers, disease type, patient age, gender, etc. For each key factor, its changing trend and correlation analysis can be displayed through charts (such as bar charts and line charts).
[0091] In another embodiment, a decision library is constructed to automatically match and recommend corresponding measures based on the risk assessment results, which must cover multiple aspects such as medical intervention, lifestyle adjustment, and further examination to ensure that patients receive comprehensive and effective treatment and management.
[0092] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.
[0093] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A risk assessment system based on gas marker detection, characterized in that: include: A gas marker detection component, used to collect test samples according to the evaluation indicators; A data preprocessing component, used to preprocess the test sample and obtain preprocessing data; The risk assessment component is used to perform feature extraction and risk assessment on the preprocessed data to obtain risk data.
2. A risk assessment system based on gas marker detection according to claim 1, characterized in that: The gas marker detection component adopts a sensor matrix based on a cross-reaction suppression strategy.
3. A risk assessment system based on gas marker detection according to claim 1, characterized in that: The data preprocessing component includes a data cleaning module, a data correction module, a data denoising module, a data resampling module and a data standardization module; The data cleaning module is used to identify and process outliers in the test samples; The data correction module is used to correct the cleaned data according to the compensation strategy; The data denoising module is used to reduce signal noise in the corrected data based on a filtering algorithm; The data resampling module is used to adjust the sampling interval or the number of sampling points of the denoised data; The data standardization module is used to perform standardization processing on the resampled data to obtain preprocessed data.
4. A risk assessment system based on gas marker detection according to claim 1, characterized in that: The risk assessment components include: A feature extraction module, used for extracting features of the preprocessed data in terms of spatial distribution and time series; A risk assessment module, used to assess the risk based on the extracted features; A report generation module is used to generate relevant text according to the risk data.
5. A risk assessment system based on gas marker detection according to claim 4, characterized in that: The feature extraction module includes a local feature extraction module, a temporal feature extraction module and a feature fusion module; The local feature extraction module is used to extract local features based on the preprocessed data; The temporal feature extraction module is used to extract temporal features based on the local features; The feature fusion module is used to fuse the multiple time series features to obtain the extracted features.
6. A risk assessment method based on gas marker detection, characterized in that: The specific steps are: Acquire a test sample according to the evaluation index, and preprocess the test sample to obtain preprocessed data; Constructing a risk assessment model based on a neural network, and inputting the preprocessed data into the risk assessment model for risk assessment; Used to generate reports based on risk assessment results.
7. A risk assessment method based on gas marker detection according to claim 6, characterized in that: The risk assessment model is a CNN-TCN model, including an input layer, a CNN layer, a TCN layer, a fully connected layer and an output layer; The input layer is used to obtain the preprocessed data; The CNN layer is used to extract local features of the preprocessed data; The TCN layer is used to extract temporal features based on the local features; The fully connected layer is used to map the time series features to the decision space of risk assessment; The output layer is used to output risk assessment data.
8. A risk assessment method based on gas marker detection according to claim 7, characterized in that: An attention mechanism layer and a feature fusion layer are also provided between the TCN layer and the fully connected layer; The attention mechanism layer is used for weight matching; The feature fusion layer is used to fuse the outputs of the TCN layer and the attention mechanism layer to obtain fused features.