Human body breathing gas metabonomics analysis method and system in special environment
By combining GC-MS and machine learning techniques with LSTM and DoubleML methods for respiratory gas metabolomics analysis, the problem of data error in special environments caused by traditional methods has been solved, achieving highly sensitive and real-time metabolomics analysis and providing accurate physiological status assessment.
Patent Information
- Application Number
- CN202511296450.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-11
- Publication Date
- 2025-12-02
AI Technical Summary
Traditional metabolomics analysis is difficult to achieve non-invasive, real-time and highly sensitive respiratory gas analysis under special conditions, and the lack of environmental parameter compensation algorithms leads to large data errors, making it impossible to accurately assess the physiological state of the human body.
GC-MS was used to detect respiratory gas metabolites, DoubleML was used to screen characteristic metabolites, LSTM respiratory recognition model was used for environmental compensation, and a metabolomics analysis model was constructed. Physiological indicators were predicted by combining machine learning.
It enables highly sensitive, non-invasive metabolomics analysis under special conditions, allowing for real-time assessment of human metabolic changes and physiological states, and providing accurate results on physical condition.
Smart Images
Figure CN121040889A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of biomedical detection technology, and more specifically to methods and systems for analyzing human respiratory gas metabolomics under special conditions. Background Technology
[0002] Special environments (such as high altitude and low oxygen, high temperature and high cold) have a significant impact on human metabolism, but traditional metabolomics analysis relies on blood or urine samples, which has problems such as invasive sampling, poor real-time performance, and difficulty in operation in extreme environments.
[0003] Furthermore, existing respiratory gas analysis equipment is bulky, has low sensitivity, and lacks real-time compensation algorithms for environmental parameters (such as temperature and air pressure), resulting in large data errors. The lack of correlation models between respiratory metabolic biomarkers and specific environmental stress responses makes it difficult to achieve accurate dynamic assessment and early warning of physiological states.
[0004] Therefore, how to achieve accurate analysis of human metabolomics under special conditions is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention
[0005] In view of this, the present invention provides a method and system for analyzing human respiratory gas metabolomics under special environments. It is a non-invasive and highly sensitive analysis method that can obtain the correspondence between the complex metabolic regulation mechanisms and changes in metabolic levels and processes caused by different environments during high-intensity training and training activities in special environments such as high altitude, high cold, and high heat, and achieve accurate analysis.
[0006] To achieve the above objectives, the present invention adopts the following technical solution:
[0007] A method for analyzing human respiratory gas metabolomics under special conditions includes the following steps:
[0008] Step 1: Collect the breath gas of workers in special environments, perform GC-MS detection on all metabolites in the collected breath gas, obtain the types and corresponding concentrations of metabolites, and use the DoubleML method to screen the types and corresponding concentrations of metabolites to obtain characteristic metabolites closely related to exercise levels in special environments.
[0009] Step 2: Collect breathing-related parameters of workers in real time under special conditions, use the trained LSTM-based breathing recognition model to perform breathing recognition, and collect gas during the exhalation phase as a detection sample.
[0010] Step 3: Perform GC-MS analysis on the test samples to obtain the concentration of characteristic metabolites in the test samples, and perform environmental compensation;
[0011] Step 4: Input the compensated concentration of characteristic metabolites into the trained metabolomics analysis model to obtain the body status results.
[0012] Preferably, the breathing-related parameters include airflow waveform, pressure change signal, and environmental parameters, including temperature and humidity; a breathing recognition model is constructed based on LSTM to identify the time-varying relationship between environmental parameters and signal distortion, identify the breathing phase, and determine the exhalation and inhalation phases.
[0013] Preferably, the interaction term analysis of the DoubleML method is used to detect effect modification variables, and important metabolites are screened as characteristic metabolites by the significance of the interaction term between metabolites and exercise levels.
[0014] Preferably, the coupling effect of temperature, humidity and pressure is automatically learned based on respiratory-related parameters to establish a basic compensation relationship, and environmental compensation is performed on the concentration of characteristic metabolites to correct the accuracy of respiratory gas detection data under extreme conditions.
[0015] Preferably, a metabolomics analysis model is constructed based on a machine learning model, and characteristic metabolites and their concentrations, as well as corresponding physiological indicators, are collected under special conditions. The metabolomics analysis model is then trained, and the characteristic metabolites, their concentrations, and physiological indicators are matched and learned. The nonlinear mapping relationship between characteristic metabolites, their concentrations, and physiological indicators is trained through hyperparameter optimization and cross-validation.
[0016] Preferably, the concentration of the compensated characteristic metabolites is input into the trained metabolomics analysis model to predict and output physiological indicators, which are then used as the result of the body's state.
[0017] Preferably, a multi-output regression model based on random forest or a multi-task learning model based on neural network is constructed as a metabolomics analysis model.
[0018] Preferably, step 4 further includes determining whether the concentration is within a reasonable range based on the concentration of the characteristic metabolite and the set concentration threshold, and using the determination result as the result of the body status.
[0019] Preferred special environments include high pressure, low pressure, high heat, and extreme cold environments.
[0020] A human respiratory gas metabolomics analysis system under special conditions includes a gas acquisition module, a feature screening module, a detection and analysis module, a compensation module, and a data processing module.
[0021] The gas acquisition module uses an LSTM-based respiratory recognition model deployed under special conditions to identify respiration, sample gas during the exhalation phase, and obtain a test sample.
[0022] The feature screening module uses the DoubleML method to screen for characteristic metabolites and their corresponding concentrations.
[0023] The detection and analysis module performs GC-MS detection on the test sample to achieve qualitative and quantitative analysis of trace metabolites and obtain the concentration of characteristic metabolites in the test sample.
[0024] The compensation module performs environmental compensation for the concentration of characteristic metabolites;
[0025] The data processing module is equipped with a metabolomics analysis model. The concentrations of characteristic metabolites and compensated characteristic metabolites are input into the metabolomics analysis model to obtain the body status results.
[0026] Preferably, the gas acquisition module includes an adaptive breathing mask, equipped with internal sensors, an external environment module, and a sampling controller; the sampling controller deploys a breathing recognition model; the internal sensors monitor breathing gas and collect breathing parameters, including airflow waveforms and pressure change signals; the external environment module monitors and collects environmental parameters in real time; the sampling controller uses an LSTM-based breathing recognition model to extract bidirectional temporal features based on the airflow waveform, pressure change signals, and environmental parameters, accurately distinguishing between the inhalation and exhalation phases, optimizing the sampling timing, improving the accuracy of the sampling data, and collecting gas after the sampling timing is reached.
[0027] Preferably, the detection and analysis module includes a high-precision sensor array, equipped with a MEMS gas sensor and a gas chromatography-mass spectrometry (GC-MS) module. Multiple MEMS gas sensors are used to detect various gases, including CO2, O2, NO, etc.; the GC-MS module performs GC-MS detection, and trace metabolites, including acetone, isoprene, etc., are qualitatively and quantitatively analyzed.
[0028] Preferably, the data processing module further includes a state judgment unit and a concentration judgment unit; the state judgment unit performs physiological index judgment and judges the body state based on the physiological index; the concentration judgment unit judges whether the concentration is within a reasonable range based on the concentration of characteristic metabolites and the set concentration threshold.
[0029] As can be seen from the above technical solution, compared with the prior art, this invention discloses a method and system for analyzing human respiratory gas metabolomics under special environments, which collects respiratory gases through environmental compensation. During detection and analysis, multimodal detection fusion is performed, combining sensor arrays (real-time) and GC-MS (high precision) to balance detection speed and accuracy; a ternary correlation model of "metabolite-environmental parameters-physiological indicators" is constructed to achieve the analysis and prediction of metabolic data and body state results. This invention performs environmental compensation for respiratory gases under special environments and utilizes machine learning for metabolic level assessment, which can more comprehensively analyze and interpret changes in human metabolic state. Machine learning technology can process complex metabolic data, discover hidden patterns and correlations, thereby providing more accurate metabolic level assessment results and allowing for a deeper exploration of the relationship between changes in metabolic state and health and disease. Attached Figure Description
[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0031] Figure 1 This is a schematic diagram of the human respiratory gas metabolomics analysis method under special conditions provided by the present invention;
[0032] Figure 2 This is a schematic diagram illustrating the data and scenario foundation provided by the DoubleML method in this invention;
[0033] Figure 3 This is a schematic diagram of the LSTM network processing procedure provided by the present invention;
[0034] Figure 4 This is a schematic diagram illustrating the evaluation results of the metabolomics analysis model provided by the present invention. Detailed Implementation
[0035] 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.
[0036] This invention discloses a method for analyzing human respiratory gas metabolomics under special conditions, such as... Figure 1 As shown, it includes the following steps:
[0037] S1: Collect respiratory gases from workers operating under special conditions, perform GC-MS detection on all metabolites in the collected respiratory gases, obtain the types and corresponding concentrations of metabolites, and use the DoubleML method to screen the types and corresponding concentrations of metabolites to obtain characteristic metabolites closely related to exercise levels under special conditions.
[0038] S2: Real-time collection of breathing-related parameters of workers in special environments, use of a trained LSTM-based breathing recognition model for breathing recognition, and collection of exhalation phase gas as detection sample;
[0039] S3: Perform GC-MS detection on the test sample to obtain the concentration of characteristic metabolites in the test sample and perform environmental compensation;
[0040] S4: Input the compensated concentration of characteristic metabolites into the trained metabolomics analysis model to obtain the body status results.
[0041] Furthermore, respiratory-related parameters include airflow waveforms, pressure change signals, and environmental parameters, such as temperature and humidity. Based on LSTM, the time-varying relationship between environmental parameters and signal distortion is identified to determine the respiratory phase and distinguish between the expiratory and inspiratory phases. Figure 3 The diagram illustrates the process of an LSTM network processing a sequence.
[0042] Furthermore, DoubleML interaction term analysis was used to detect effect-modifying variables. Important metabolites were screened as characteristic metabolites based on the significance of the interaction term between metabolites and exercise levels, such as... Figure 2The diagram illustrates the data and scenario foundation provided for the DoubleML method to screen feature metabolites. By splitting motion data, it provides feature extraction strategies for machine learning data (divided into intra-motion and inter-motion scenarios, including supervised and unsupervised scenarios, and cases with and without cross-validation). DoubleML provides samples for a "treatment group" (model trained with pre- and post-training data) and a "control group" (model trained with the same training conditions), and then estimates the causal effects under the two strategies using DoubleML. The DoubleML method iterates through multiple rounds to progressively improve a solution set, ultimately finding an optimal or near-optimal solution, which can be represented as a set of feature metabolites. It identifies the feature metabolites most correlated with the subject's physical state under specific conditions, reflecting the subject's physical and motion status. Feature screening is performed using DoubleML or a causal forest model; sample analysis is conducted to obtain the feature metabolites most correlated with the subject's physical state under specific conditions and their thresholds. These metabolites are identified through prior analysis and comparison of exhaled samples, revealing the metabolites most strongly associated with human physiological functions under specific conditions. Feature screening can eliminate metabolites that are only correlated with physiological indicators but have no causal relationship, thus reducing the risk of overfitting.
[0043] Furthermore, based on the automatic learning of the coupled effects of temperature, humidity, and pressure according to respiratory-related parameters, a basic compensation relationship is established to perform environmental compensation for the concentration of characteristic metabolites, thereby correcting the accuracy of respiratory gas detection data under extreme conditions.
[0044] Furthermore, a metabolomics analysis model is constructed based on machine learning. Characteristic metabolites and their concentrations, along with corresponding physiological indicators, are collected under specific environmental conditions. The model is then trained by matching characteristic metabolites, their concentrations, and physiological indicators. Hyperparameter optimization and cross-validation are used to train the nonlinear mapping relationship between characteristic metabolites, their concentrations, and physiological indicators. The trained metabolomics analysis model can predict physiological indicators based on characteristic metabolites and their concentrations. Using this model to predict new metabolic data yields corresponding metabolic level assessment results. Information such as model weights and feature importance can explain which metabolites or metabolic pathways have a greater impact on metabolic levels. Figure 4 The diagram shows the evaluation results of the metabolomics analysis model, which provides a visual representation of the set of characteristic metabolites. The horizontal axis represents different metabolite categories, and the vertical axis shows the distribution of evaluation index values under different data processing groups. * indicates whether there are significant differences in evaluation indexes under different feature extraction strategies.
[0045] Furthermore, the compensated concentrations of characteristic metabolites are input into the trained metabolomics analysis model to predict and output physiological indicators, which are then used as the results of the body's state.
[0046] Furthermore, we construct multi-output regression models based on random forests or multi-task learning models based on neural networks as metabolomics analysis models.
[0047] Furthermore, S4 also includes determining whether the concentration is within a reasonable range based on the concentration of the characteristic metabolite and a set concentration threshold, and using the determination result as the result of the body status.
[0048] Furthermore, the special environments include high-pressure, low-pressure, high-heat, and extremely cold environments. The samples used in the training model are gas chromatographic analysis data of respiratory gases collected under these special environments.
[0049] On the other hand, a human respiratory gas metabolomics analysis system under special conditions includes a gas acquisition module, a feature screening module, a detection and analysis module, a compensation module, and a data processing module.
[0050] The gas acquisition module uses an LSTM-based respiratory recognition model deployed under special conditions to identify respiration, sample gas during the exhalation phase, and obtain a test sample.
[0051] The feature screening module uses the DoubleML method to screen for characteristic metabolites and their corresponding concentrations.
[0052] The detection and analysis module performs GC-MS detection on the test sample to achieve qualitative and quantitative analysis of trace metabolites and obtain the concentration of characteristic metabolites in the test sample.
[0053] The compensation module performs environmental compensation for the concentration of characteristic metabolites;
[0054] The data processing module is equipped with a metabolomics analysis model. The concentrations of characteristic metabolites and compensated characteristic metabolites are input into the metabolomics analysis model to obtain the body status results.
[0055] Furthermore, the gas acquisition module includes an adaptive breathing mask equipped with internal sensors, an external environment module, and a sampling controller. The sampling controller deploys a breathing recognition model. The internal sensors monitor breathing gas and collect breathing parameters, including airflow waveforms and pressure change signals. The external environment module monitors and collects environmental parameters in real time. Based on the airflow waveform, pressure change signals, and environmental parameters, the sampling controller uses an LSTM-based breathing recognition model to perform bidirectional temporal feature extraction, accurately distinguishing between the inhalation and exhalation phases, optimizing the sampling timing, improving the accuracy of the sampling data, and collecting gas when the sampling timing is reached.
[0056] Furthermore, the detection and analysis module includes a high-precision sensor array, equipped with MEMS gas sensors and a gas chromatography-mass spectrometry (GC-MS) module. Multiple MEMS gas sensors are used to detect various gases, including CO2, O2, NO, etc.; the GC-MS module performs GC-MS detection, and trace metabolites, including acetone and isoprene, are qualitatively and quantitatively analyzed.
[0057] Furthermore, the data processing module also includes a state judgment unit and a concentration judgment unit; the state judgment unit performs physiological index judgment and determines the body state based on the physiological index; the concentration judgment unit determines whether the concentration is within a reasonable range based on the concentration of characteristic metabolites and the set concentration threshold.
[0058] In one specific embodiment, under high-altitude hypoxic environment (5200 meters above sea level, 550 hPa), the subject wears an adaptive breathing mask and activates the system. The breath gas is enriched with metabolites through a multi-channel adsorption tube, and environmental parameters are recorded simultaneously. Biomarkers are detected by GC-MS. Since the gas to be detected is affected by environmental parameters during GC-MS gas detection, the concentration of metabolic biomarkers in the breath gas can be obtained more accurately by compensating for the environmental parameters. The metabolomics analysis model integrates metabolic data to provide results on whether the concentration of key metabolites is within a reasonable range and the subject's physical condition.
[0059] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.
[0060] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those 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 invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for analyzing human respiratory gas metabolomics under special conditions, characterized in that, Includes the following steps: Step 1: Collect the breath gas of workers in special environments, perform GC-MS detection on all metabolites in the collected breath gas, obtain the types and corresponding concentrations of metabolites, and use the DoubleML method to screen the types and corresponding concentrations of metabolites to obtain characteristic metabolites closely related to exercise levels in special environments. Step 2: Collect breathing-related parameters of workers in real time under special conditions, use the trained LSTM-based breathing recognition model to perform breathing recognition, and collect gas during the exhalation phase as a detection sample. Step 3: Perform GC-MS analysis on the test samples to obtain the concentration of characteristic metabolites in the test samples, and perform environmental compensation; Step 4: Input the compensated concentration of characteristic metabolites into the trained metabolomics analysis model to obtain the body status results.
2. The method for analyzing human respiratory gas metabolomics under special conditions according to claim 1, characterized in that, Respiratory parameters include airflow waveform, pressure change signal, and environmental parameters, including temperature and humidity. A respiratory recognition model is built based on LSTM to identify the time-varying relationship between environmental parameters and signal distortion, identify the respiratory phase, and determine the expiratory and inspiratory phases.
3. The method for analyzing human respiratory gas metabolomics under special conditions according to claim 2, characterized in that, By studying the coupled effects of temperature, humidity, and pressure based on respiratory-related parameters, a basic compensation relationship is established to compensate for the concentration of characteristic metabolites in the environment.
4. The method for analyzing human respiratory gas metabolomics under special conditions according to claim 1, characterized in that, A metabolomics analysis model was constructed based on a machine learning model. Characteristic metabolites and their concentrations under special conditions, as well as corresponding physiological indicators, were collected to train the metabolomics analysis model.
5. The method for analyzing human respiratory gas metabolomics under special conditions according to claim 1, characterized in that, Step 4 also includes determining whether the concentration is within a reasonable range based on the concentration of the characteristic metabolites and the set concentration threshold, and using the determination result as the result of the body status.
6. The method for analyzing human respiratory gas metabolomics under special conditions according to claim 1, characterized in that, Special environments include high pressure, low pressure, high temperature, and extreme cold environments.
7. A human respiratory gas metabolomics analysis system under special conditions, characterized in that, The method for analyzing human respiratory gas metabolomics under special conditions according to any one of claims 1-6 includes a gas acquisition module, a feature screening module, a detection and analysis module, a compensation module, and a data processing module; The gas acquisition module uses an LSTM-based respiratory recognition model deployed under special conditions to identify respiration, sample gas during the exhalation phase, and obtain a test sample. The feature screening module uses the DoubleML method to screen for characteristic metabolites and their corresponding concentrations. The detection and analysis module performs GC-MS detection on the test sample to achieve qualitative and quantitative analysis of trace metabolites and obtain the concentration of characteristic metabolites in the test sample. The compensation module performs environmental compensation for the concentration of characteristic metabolites; The data processing module is equipped with a metabolomics analysis model. The concentrations of characteristic metabolites and compensated characteristic metabolites are input into the metabolomics analysis model to obtain the body status results.
8. The human respiratory gas metabolomics analysis system under special conditions according to claim 7, characterized in that, The gas acquisition module includes an adaptive breathing mask equipped with internal sensors, an external environment module, and a sampling controller. The sampling controller deploys a breathing recognition model. The internal sensors monitor breathing gas and collect breathing parameters, including airflow waveforms and pressure change signals. The external environment module monitors and collects environmental parameters in real time. Based on the airflow waveform, pressure change signals, and environmental parameters, the sampling controller uses an LSTM-based breathing recognition model to perform bidirectional temporal feature extraction, identify the inhalation and exhalation phases, and collect gas when the sampling time is reached.
9. The human respiratory gas metabolomics analysis system under special conditions according to claim 7, characterized in that, The detection and analysis module includes a high-precision sensor array, equipped with a MEMS gas sensor and a gas chromatography-mass spectrometry module, which uses multiple sets of MEMS gas sensors to detect a variety of gases.