Data analysis system for early diagnosis of digestive tract diseases

Abstract

The invention discloses a data analysis system for early diagnosis of digestive tract diseases, relates to the field of medical health, and constructs a multi-level digestive tract disease risk assessment model through machine learning and deep learning technologies. In the preliminary evaluation stage, oral cavity gas data in the standard data set C is trained by using a support vector machine (SVM), a preliminary digestive tract risk score D1 is generated, and the preliminary digestive tract risk score D1 is compared with a preset preliminary risk threshold value T1 to judge whether a patient has a digestive tract disease risk or not. And when the score D1 exceeds the initial threshold value T1, the system automatically starts a deep learning model such as an LSTM network to carry out more accurate secondary evaluation, and calculates and outputs a deep digestive tract risk score D2. By extracting big data and analyzing medical record data, the model is further optimized, a secondary risk threshold T2 is set, and finally, real-time examination prompts such as gastroscope or blood examination are sent to a doctor or an automatic system through linkage of the system and a diagnosis and treatment server.

Description

Technical Field

[0001] The present invention relates to the field of medical health technology, and in particular to a data analysis system for early diagnosis of digestive tract diseases. Background Art

[0002] Early diagnosis of digestive tract diseases is an important branch in the field of medical health. Early detection and timely intervention are particularly important in the prevention and treatment of digestive system diseases. With the development of science and technology, especially the advancement of sensor technology and big data analysis, disease screening based on biomarker detection has gradually become an emerging trend. In the early diagnosis of digestive tract diseases, oral gas detection, as a non-invasive system, has become a hot topic of research and application because it can reflect the metabolic and microbial activities in the digestive system. This system can provide early disease signals by real-time monitoring of oral gas concentration and analyzing changes in gas composition, thereby providing a new technical approach for early screening of digestive tract diseases.

[0003] At present, the current early diagnosis technology of digestive tract diseases still faces many challenges. Existing diagnostic systems usually rely on invasive tests, such as gastroscopy, endoscopy or blood tests, which not only impose a certain physical burden on patients, but are also expensive and can only be implemented when the disease has already manifested itself. In addition, although the traditional gas detection system has certain advantages, it still has problems such as long data collection cycle, insufficient data processing and analysis, and low accuracy of diagnostic results. This limits the practical application of gas detection technology, especially in the accuracy of early diagnosis. Therefore, it is urgent to develop a new type of data analysis system to make up for the shortcomings of the traditional system by improving detection efficiency and accuracy. Summary of the invention

[0004] In view of the deficiencies of the prior art, the present invention provides a data analysis system for early diagnosis of digestive tract diseases, which solves the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a data analysis system for early diagnosis of digestive tract diseases, including a gas detection module, a data processing module, a preliminary analysis module and a deep analysis module; The gas detection module uses an oral gas detection device to set a detection cycle, collects the patient's oral gas data in real time, and builds a diagnosis and treatment server, and transmits the oral gas data to the diagnosis and treatment server through a wireless transmission network; The data processing module receives the oral gas data in real time in the diagnosis and treatment server, pre-processes the oral gas data, obtains a standard data set C, extracts features from the standard data set C, obtains a feature data set, and constructs a time series database, sets a transmission port, and stores the standard data set C and the feature data set in the time series database through the transmission port; The preliminary analysis module constructs a machine learning model, extracts the standard data set C through the transmission port, inputs it into the machine learning model, outputs a preliminary digestive tract risk score D1, and presets a preliminary risk threshold T1, performs a preliminary comparative evaluation on the preliminary risk threshold T1 and the preliminary digestive tract risk score D1, and analyzes the normal range of gas concentration and the limit of abnormal fluctuation; When the result of the preliminary comparative evaluation is that the oral gas is abnormal, the deep analysis module constructs a deep learning model, extracts the feature data set through the transmission port, inputs it into the deep learning model, calculates and outputs a deep gastrointestinal risk score D2, and trains based on the deep learning model to obtain a secondary risk threshold T2, then performs a secondary comparative evaluation on the deep gastrointestinal risk score D2 and the secondary risk threshold T2, and deeply analyzes the deep correlation between the patient's gastrointestinal symptoms and gas concentration.

[0006] Preferably, the gas detection module includes a gas data acquisition unit and a data transmission unit; The gas data acquisition unit constructs a gas sensor array through an oral gas detection device. Each gas sensor responds to a specific oral gas, collects oral gas from a patient monitored by the digestive tract, obtains the patient's oral gas data, and sets the acquisition frequency to once every 5 minutes. The gas sensor array includes electrochemical gas sensors, semiconductor gas sensors and optical sensors; The oral gas data includes ammonia concentration C1, methane concentration C2, hydrogen sulfide concentration C3, acetic acid concentration C4, ..., the i-th oral gas C i ; The data transmission unit constructs a diagnosis and treatment server and adopts a wireless communication network WIFI to integrate and connect the diagnosis and treatment server with the sensor transmission interface of the gas sensor array, and records the data collection time of the gas sensor array to transmit the oral gas data to the diagnosis and treatment server.

[0007] Preferably, the data processing module includes a standardization processing unit, a data storage unit and a feature extraction unit; The standardization processing unit receives the oral gas data in real time in the diagnosis and treatment server, pre-processes the oral gas data, and integrates the pre-processed oral gas data to obtain a standard data set C; The preprocessing includes timestamp labeling, denoising, data standardization and time series smoothing; The standard data set C = {C1(t), C2(t), C3(t), C4(t), ..., C i (t)}; The timestamp tagging is to perform real-time timestamp on the collected oral gas data by recording the data collection time of the gas sensor array; The denoising is performed by removing random noise in the collected oral gas data using a filtering algorithm; The data standardization is performed by standardizing the oral gas data and making the units and magnitude differences of different oral gas data dimensionless; The time series smoothing optimizes the volatility of oral gas data in the short term by using exponential smoothing; The data storage unit constructs a time series database and sets a transmission port, wherein the transmission port includes a write port and a write port, and constructs a data storage table, wherein the data storage table includes a real-time data table and a historical data table. The standard data set C obtained after preprocessing is stored in the real-time data table in the time series database through the write port. At the same time, the original stored data in the real-time data table is automatically transferred to the historical data table and stored in order according to timestamps.

[0008] Preferably, the feature extraction unit extracts the standard data set C in the real-time data table and the historical data table through the write port, and performs feature extraction to obtain a feature data set, wherein the feature extraction includes extraction of the rate of change of gas concentration, extraction of gas concentration volatility, and extraction of the correlation between gas concentration and in vivo health indicators; Then, the feature data set is written into the time series database through the write port for data storage; The characteristic data set includes the gas concentration change rate R, the gas concentration fluctuation value Vc and the gas concentration correlation index p (C i , C j ), where C j represents the j-th oral gas data in the standard data set C; The feature extraction unit includes a gas concentration extraction unit, a gas concentration fluctuation extraction unit and a correlation extraction unit; The gas concentration extraction unit extracts all processed oral gas data in the standard data set C through the write port, performs concentration change extraction, obtains the gas concentration change rate R, and analyzes the change of each oral gas data; The rate of change R of the gas concentration is calculated and outputted by the following algorithm formula; ; In the formula, R i represents the rate of change of the gas concentration of the i-th oral gas data in the standard data set C, C i (t) represents the i-th oral gas data in the standard data set C at time t, C i (t-1) represents the i-th oral gas data in the standard data set C at time t-1, and △t represents the sampling time interval.

[0009] Preferably, the gas concentration fluctuation extraction unit extracts all processed oral gas data in the standard data set C through a write port, and calculates the standard deviation at a certain time to obtain the gas concentration fluctuation value Vc, and analyzes the stability of the digestive tract; The gas concentration fluctuation value Vc is calculated and outputted by the following algorithm formula; ; Where N represents the total number of oral gas data in the time window, Vc i represents the gas concentration fluctuation value of the i-th oral gas data in the standard data set C, Represents the mean gas concentration of the i-th oral gas data in the standard data set C.

[0010] Preferably, the correlation extraction unit extracts different oral gas data in the standard data set C through the write port through correlation extraction, and calculates the output gas concentration correlation index p (C i , C j ), to analyze the relationship between different gases; The gas concentration correlation index p (C i , C j ) The output is calculated by the following algorithm formula; ; In the formula, represents the mean value of the i-th oral gas data in the standard data set C, C j represents the jth oral gas data in the standard data set C, Represents the mean of the j-th oral gas data in the standard data set C.

[0011] Preferably, the preliminary analysis module includes a preliminary detection calculation unit and a preliminary evaluation unit; The preliminary detection calculation unit constructs a machine learning model by using a support vector machine SVM, and extracts a standard data set C through the write port and inputs it into the machine learning model to calculate and output a preliminary digestive tract risk score D1, and perform preliminary detection of the gastrointestinal health of the digestive tract; The preliminary digestive tract risk score D1 is calculated and output by the following machine learning model; ; In the formula, C1(t) represents the value of ammonia concentration at time t, C2(t) represents the value of methane concentration at time t, C3(t) represents the value of hydrogen sulfide concentration at time t, and C4(t) represents the value of acetic acid concentration at time t. represents the Sigmoid function, which is used to map the output to the range of 0 to 1 to represent the risk score, , , and They respectively represent the coefficients to be learned of ammonia concentration C1, methane concentration C2, hydrogen sulfide concentration C3 and acetic acid concentration C4, which are obtained through machine learning training.

[0012] Preferably, the preliminary evaluation unit extracts the standard data set C in the historical data table through the write port, sets the preliminary risk threshold T1 according to the normal ranges of ammonia concentration C1, methane concentration C2, hydrogen sulfide concentration C3 and acetic acid concentration C4, and then performs a preliminary comparative evaluation of the preliminary digestive tract risk score D1 and the preliminary risk threshold T1 to perform a preliminary analysis of the digestive tract disease risk in the oral gas. The specific evaluation content is as follows; When the initial gastrointestinal risk score D1 is less than the initial risk threshold T1, it means that there is no abnormality in oral gas, and no intervention and regular monitoring are required; When the preliminary gastrointestinal risk score D1 ≥ preliminary risk threshold T1, it indicates that there is an abnormality in oral gas and a disease risk in the digestive tract, and a deep assessment mechanism is executed.

[0013] Preferably, the depth analysis module includes a depth analysis unit and a depth evaluation unit; The deep analysis unit executes a deep assessment mechanism when the preliminary comparison and assessment shows that there is a disease risk in the digestive tract. The deep assessment mechanism constructs a deep learning model by using an LSTM network and a neural network, extracts a feature data set and a standard data set C through a write port, inputs them into the deep learning model, and calculates and outputs a deep digestive tract risk score D2; The deep digestive tract risk score D2 is calculated and output by the following algorithm formula; ; In the formula, represents the deep learning adjustment coefficient, f represents the nonlinear function, L LSTM Represents a long short-term memory network.

[0014] Preferably, the depth assessment unit accesses the digestive tract case system through an API application interface, extracts real case results, sets a secondary risk threshold T2 through big data analysis and machine learning models, and then performs a secondary comparative assessment on the deep digestive tract risk score D2 and the secondary risk threshold T2, deeply analyzing the deep correlation between the patient's digestive tract symptoms and gas concentration. The specific assessment content is as follows; When the deep digestive tract risk score D2 ≥ the secondary risk threshold T2, it means that the patient's oral gas is abnormal and there is a risk of digestive tract disease. At this time, a medical examination prompt is sent to the client through the diagnosis and treatment server; When the deep gastrointestinal risk score D2 is less than the secondary risk threshold T2, it indicates that the patient's oral gas risk is within an acceptable range. At this time, a regular detection reminder is sent to the client through the diagnosis and treatment server.

[0015] Beneficial Effects

[0016] The present invention provides a data analysis system for early diagnosis of digestive tract diseases. It has the following beneficial effects: (1) The system realizes real-time collection of oral gas data of patients through the gas sensor array of the oral gas detection equipment. The collection cycle is set to a frequency of once every 5 minutes to ensure efficient monitoring of changes in gas concentration. Through the connection between the diagnosis and treatment server and the wireless transmission network, the gas data is transmitted and stored in the time series database, supporting real-time updates and queries. On this basis, the system pre-processes the collected oral gas data, effectively eliminating random noise and optimizing data quality. This technical solution can accurately capture the early signals of potential gastrointestinal lesions, provide early diagnosis support for gastrointestinal diseases through changes in oral gas concentration, and significantly improve the sensitivity and diagnostic efficiency of disease warning.

[0017] (2) The system uses the concentration data of various types of oral gases in the standard data set C, combined with a multi-dimensional feature extraction algorithm, to deeply analyze the volatility, rate of change of gas concentration, and correlation with health indicators in the body. Specifically, the rate of change of gas concentration, volatility, and correlation indicators between different gas concentrations provide an effective basis for judging the volatility of digestive tract diseases. By calculating the rate of change of gas concentration R, it is possible to detect the rapid fluctuation of the patient's oral gas, indicating potential pathological changes. The combination of volatility extraction and correlation analysis can further reveal the stability of gastrointestinal health, and provide doctors with in-depth insights into the nature of lesions, progression trends, and their relationship with other physiological indicators, facilitating more accurate risk assessment and intervention decisions.

[0018] (3) The system uses machine learning and deep learning technologies to build a multi-level gastrointestinal disease risk assessment model. In the initial assessment stage, the support vector machine (SVM) is used to train the oral gas data in the standard data set C to generate a preliminary gastrointestinal risk score D1, which is compared with the preset preliminary risk threshold T1 to determine whether the patient has a risk of gastrointestinal disease. When the score D1 exceeds the preliminary threshold T1, the system automatically starts a deep learning model such as an LSTM network to perform a more accurate secondary assessment and calculate and output a deep gastrointestinal risk score D2. By extracting big data analysis and medical record data, the model is further optimized and the secondary risk threshold T2 is set. Finally, the system is linked to the diagnosis and treatment server to send real-time examination prompts such as gastroscopy or blood tests to doctors or automated systems. This intelligent multi-layer risk assessment mechanism not only improves the accuracy of early diagnosis, but also provides patients with more personalized diagnosis and treatment plans, reduces the risk of misdiagnosis and missed diagnosis, and improves the efficiency of clinical decision-making and the success rate of patient treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 The present invention is a flow chart of a data analysis system for early diagnosis of digestive tract diseases. DETAILED DESCRIPTION

[0020] 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. Example 1

[0021] The present invention provides a data analysis system for early diagnosis of digestive tract diseases. Figure 1 , including the following steps: The gas detection module uses the oral gas detection device to set the detection cycle, collect the patient's oral gas data in real time, and build a diagnosis and treatment server to transmit the oral gas data to the diagnosis and treatment server through a wireless transmission network; The data processing module receives the oral gas data in real time in the diagnosis and treatment server, pre-processes the oral gas data, obtains the standard data set C, extracts features from the standard data set C, obtains the feature data set, and constructs a time series database, sets a transmission port, and stores the standard data set C and the feature data set in the time series database through the transmission port; The preliminary analysis module constructs a machine learning model, extracts the standard data set C through the transmission port, inputs it into the machine learning model, outputs the preliminary gastrointestinal risk score D1, and presets the preliminary risk threshold T1. The preliminary risk threshold T1 is compared and evaluated with the preliminary gastrointestinal risk score D1 to analyze the normal range of gas concentration and the limit of abnormal fluctuation; When the result of the preliminary comparative evaluation is abnormal oral gas, the deep analysis module constructs a deep learning model, extracts the feature data set through the transmission port, inputs it into the deep learning model, calculates and outputs the deep gastrointestinal risk score D2, and obtains the secondary risk threshold T2 through training based on the deep learning model. The deep gastrointestinal risk score D2 is then compared and evaluated with the secondary risk threshold T2 for a secondary comparison, so as to deeply analyze the deep correlation between the patient's gastrointestinal symptoms and gas concentration.

[0022] In this embodiment, the system effectively improves the accuracy and timeliness of early detection of digestive tract diseases through multi-level real-time data collection, preprocessing and analysis. The specific implementation method is to first collect the patient's gas data in real time through the oral gas detection device, and transmit the data to the constructed diagnosis and treatment server. The diagnosis and treatment server preprocesses the collected data to form a standard data set C, and further constructs a feature data set through feature extraction. These data are then stored in a time series database to provide stable data support for subsequent deep learning and machine learning analysis. In the machine learning stage, the constructed machine learning model analyzes the standard data set C, outputs a preliminary digestive tract risk score D1, and compares and evaluates it with the set preliminary risk threshold T1 to preliminarily determine whether the patient has a potential risk of digestive tract disease. On this basis, if an abnormality is found, a deep analysis is performed through a deep learning model to further optimize the risk assessment, and a secondary risk threshold T2 is obtained through training, and a secondary comparison is performed with the deep digestive tract risk score D2 to accurately analyze the deep correlation between the patient's digestive tract symptoms and oral gas concentration. Through the above system, the potential risk of digestive tract diseases can be quickly identified at an early stage by real-time monitoring of changes and fluctuations in gas concentration and its relationship with digestive tract health. Compared with traditional diagnostic methods, the present invention has significant advantages in accuracy, timeliness and sensitivity, and can provide more personalized diagnosis and intervention plans. Specific beneficial effects include: improving the accuracy of early diagnosis: by collecting oral gas data in real time and combining it with the analysis of machine learning models, it is possible to accurately capture potential lesion signals in the early stages of gastrointestinal diseases, avoiding the omission of early symptoms in traditional examinations. Enhanced sensitivity and accuracy of risk assessment: combined with deep learning models for in-depth analysis of gas concentrations, it is possible to identify subtle changes in gas concentration fluctuations, accurately predict disease development trends, and optimize patient management and treatment plans. Intelligent multi-level analysis mechanism: through a two-level risk assessment system that combines machine learning and deep learning, the assessment mechanism can be automatically adjusted at different risk levels, which improves the automation and intelligence level of the system, provides more accurate decision support for clinicians, and reduces the occurrence of misdiagnosis and missed diagnosis. Example 2

[0023] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the gas detection module includes a gas data acquisition unit and a data transmission unit; The gas data acquisition unit constructs a gas sensor array through oral gas detection equipment. Each gas sensor responds to specific oral gases and collects oral gases from patients undergoing digestive tract monitoring to obtain oral gas data of the patients. The acquisition frequency is set to once every 5 minutes. Gas sensor arrays include electrochemical gas sensors, semiconductor gas sensors, and optical sensors; Oral gas data include ammonia concentration C1, methane concentration C2, hydrogen sulfide concentration C3, acetic acid concentration C4, ..., the i-th oral gas C i ; The data transmission unit constructs a diagnosis and treatment server and uses the wireless communication network WIFI to integrate the diagnosis and treatment server with the sensor transmission interface of the gas sensor array, records the data collection time of the gas sensor array, and transmits the oral gas data to the diagnosis and treatment server.

[0024] In this embodiment, early diagnosis and monitoring of digestive tract diseases are achieved through precise oral gas detection technology. The specific implementation method is to first construct a gas sensor array composed of an electrochemical gas sensor, a semiconductor gas sensor and an optical sensor through an oral gas detection device to perform real-time monitoring of different oral gases such as ammonia, methane, hydrogen sulfide, acetic acid, etc. Each sensor can respond to specific oral gases to ensure efficient and comprehensive collection of oral gas data of patients, and set the collection frequency to once every 5 minutes to ensure the continuity and timeliness of data collection. At the same time, a diagnosis and treatment server is designed, which is connected to the gas sensor array through a wireless communication network WIFI to transmit oral gas data to the diagnosis and treatment server in real time. The diagnosis and treatment server can not only receive data, but also record the collection time of the sensor array, and perform data storage and subsequent processing. In this process, through the automatic transmission and centralized management of data, the efficient collection and processing of gas data is ensured, and errors or omissions caused by human factors in traditional detection systems are avoided. Example 3

[0025] This embodiment is explained in Example 2. Please refer to Figure 1 ,Specifically: the data processing module includes a standardization processing unit, a data storage unit and a feature extraction unit; The standardization processing unit receives the oral gas data in real time in the diagnosis and treatment server, pre-processes the oral gas data, and integrates the pre-processed oral gas data to obtain a standard data set C; Preprocessing includes timestamp labeling, denoising, data normalization, and time series smoothing; Standard dataset C = {C1(t), C2(t), C3(t), C4(t), ..., C i (t)}; The timestamp tagging is used to time-stamp the collected oral gas data in real time by recording the data collection time of the gas sensor array; Denoising removes random noise from the collected oral gas data by using a filtering algorithm; Data standardization: Oral gas data are standardized to make the differences in units and magnitudes of different oral gas data dimensionless; Time series smoothing optimizes the volatility of oral gas data in the short term by using exponential smoothing; The data storage unit constructs a time series database and sets transmission ports, which include a write port and a write port. At the same time, it constructs a data storage table, which includes a real-time data table and a historical data table. The standard data set C obtained after preprocessing is stored in the real-time data table in the time series database through the write port. At the same time, the original storage data in the real-time data table is automatically transferred to the historical data table and stored in order according to timestamps.

[0026] In this embodiment, the system significantly improves the quality of data and the accuracy of subsequent analysis by real-time reception, preprocessing and standardized storage of oral gas data. First, the diagnosis and treatment server receives oral gas data in real time and performs preprocessing, mainly including timestamp marking, denoising, data standardization and time series smoothing. Timestamp marking records the data acquisition time of the gas sensor array to mark the oral gas data in real time to ensure the timeliness of the data. The denoising process uses a filtering algorithm to remove random noise to ensure the stability of the data. Data standardization performs dimensionless processing on data of different gas concentrations to eliminate unit and magnitude differences, so that data of different gases can be compared under the same standard. Time series smoothing uses an exponential smoothing algorithm to optimize the volatility of oral gas data in the short term, thereby smoothing changes in gas concentration and facilitating the identification of long-term trends and potential anomalies. The preprocessed oral gas data constitutes a standard data set C={C1(t),C2(t),C3(t),C4(t),...,C i (t)} and store it by building a time series database. The data is stored in the real-time data table through the write port, and then automatically transferred to the historical data table and stored in order by timestamp. This implementation optimizes the quality, storage and management of the data, making the data more suitable for subsequent feature extraction, machine learning model training and disease risk assessment, thereby improving the accuracy and timeliness of early diagnosis of digestive tract diseases. Example 4

[0027] This embodiment is explained in Example 3, please refer to Figure 1 ,Specifically: the feature extraction unit extracts the standard data set C in the real-time data table and the historical data table through the ,write port, and performs feature extraction to obtain the feature data set ,wherein the feature extraction includes the extraction of the ,change rate of gas concentration, the extraction of gas concentration volatility, and the ,correlation between gas concentration and the in-vivo health indicators; Then, the feature data set is written into the time series database through the write port for data storage; The characteristic data set includes the gas concentration change rate R, gas concentration fluctuation value Vc and gas concentration correlation index p (C i , C j ), where C j represents the j-th oral gas data in the standard data set C; The feature extraction unit includes a gas concentration extraction unit, a gas concentration fluctuation extraction unit and a correlation extraction unit; The gas concentration extraction unit extracts all processed oral gas data in the standard data set C through the write port, and extracts the concentration change, obtains the gas concentration change rate R, and analyzes the change of each oral gas data; The rate of change R of gas concentration is calculated and output by the following algorithm formula; ; In the formula, R i represents the rate of change of the gas concentration of the i-th oral gas data in the standard data set C, C i (t) represents the i-th oral gas data in the standard data set C at time t, C i (t-1) represents the i-th oral gas data in the standard data set C at time t-1, and △t represents the sampling time interval.

[0028] The gas concentration fluctuation extraction unit extracts all processed oral gas data in the standard data set C through the write port, calculates the standard deviation at a certain time, obtains the gas concentration fluctuation value Vc, and analyzes the stability of the digestive tract; The gas concentration fluctuation value Vc is calculated and output by the following algorithm formula; ; Where N represents the total number of oral gas data in the time window, Vc i represents the gas concentration fluctuation value of the i-th oral gas data in the standard data set C, Represents the mean gas concentration of the i-th oral gas data in the standard data set C.

[0029] The correlation extraction unit extracts different oral gas data in the standard data set C through the write port through correlation extraction, and calculates the output gas concentration correlation index p (C i , C j ), to analyze the relationship between different gases; Gas concentration correlation index p (C i , C j ) The output is calculated by the following algorithm formula; ; In the formula, represents the mean value of the i-th oral gas data in the standard data set C, C j represents the jth oral gas data in the standard data set C, Represents the mean of the j-th oral gas data in the standard data set C.

[0030] In this embodiment, the system further improves the application effect of oral gas data in disease diagnosis through feature extraction. This process extracts the standard data set C in the real-time data table and the historical data table through the write port, conducts in-depth gas concentration analysis, generates a feature data set, and provides key data support for subsequent risk assessment and model prediction. Feature extraction includes three main aspects: extraction of the rate of change of gas concentration, extraction of gas concentration volatility, and extraction of the correlation between gas concentration and health indicators in the body. First, the rate of change of gas concentration R is calculated by calculating the change of oral gas data to obtain the rate of change of each gas concentration over time. When the rate of change of gas concentration is large, it may indicate a sharp pathological change in the gastrointestinal tract, such as fluctuations in gastric acid secretion or abnormal gastrointestinal microbial activity, providing a potential early diagnosis signal for digestive tract diseases. Secondly, the extraction of gas concentration volatility Vc reflects the degree of fluctuation of gas concentration by calculating the standard deviation of gas concentration within a certain period of time. Larger volatility is closely related to the fluctuating symptoms of gastrointestinal diseases, such as excessive gastric acid and abnormal gastrointestinal motility, providing a reliable basis for judging the health status of the gastrointestinal tract. Finally, the gas concentration correlation index p (C i , C j ) By calculating the correlation index between different gases, the relationship between different oral gases is analyzed. This helps to discover the interaction of gas concentrations and their potential association with digestive tract health, providing support for a comprehensive assessment of the functional status of the digestive tract. Through this series of feature extraction steps, the acquired feature data set not only provides accurate indicators for the risk assessment of digestive tract diseases, but also provides a rich data source for subsequent machine learning model training and disease diagnosis. Overall, the present invention effectively enhances the application value of oral gas data in the early screening of digestive tract diseases, and improves the accuracy, timeliness and reliability of diagnosis. Example 5

[0031] This embodiment is explained in Example 4. Please refer to Figure 1 ,Specifically: the preliminary analysis module includes a preliminary detection ,computation unit and a preliminary evaluation unit; The preliminary detection calculation unit constructs a machine learning model by using a support vector machine SVM, and extracts a standard data set C through a write port and inputs it into the machine learning model to calculate and output a preliminary digestive tract risk score D1, and conducts a preliminary detection of the gastrointestinal health of the digestive tract; The preliminary gastrointestinal risk score D1 is calculated and output by the following machine learning model; ; In the formula, C1(t) represents the value of ammonia concentration at time t, C2(t) represents the value of methane concentration at time t, C3(t) represents the value of hydrogen sulfide concentration at time t, and C4(t) represents the value of acetic acid concentration at time t. represents the Sigmoid function, which is used to map the output to the range of 0 to 1 to represent the risk score, , , and They respectively represent the coefficients to be learned of ammonia concentration C1, methane concentration C2, hydrogen sulfide concentration C3 and acetic acid concentration C4, which are obtained through machine learning training.

[0032] The preliminary assessment unit extracts the standard data set C in the historical data table through the write port, sets the preliminary risk threshold T1 according to the normal range of ammonia concentration C1, methane concentration C2, hydrogen sulfide concentration C3 and acetic acid concentration C4, and then makes a preliminary comparative assessment of the preliminary gastrointestinal risk score D1 and the preliminary risk threshold T1, and makes a preliminary analysis of the gastrointestinal disease risk in oral gas. The specific assessment contents are as follows; When the initial gastrointestinal risk score D1 is less than the initial risk threshold T1, it means that there is no abnormality in oral gas, and no intervention and regular monitoring are required; When the preliminary gastrointestinal risk score D1 ≥ preliminary risk threshold T1, it indicates that there is an abnormality in oral gas and a disease risk in the digestive tract, and a deep assessment mechanism is executed.

[0033] In this embodiment, the system constructs a machine learning model by using a support vector machine (SVM) algorithm to perform a preliminary digestive tract risk assessment on oral gas data, providing effective support for early diagnosis. Part of the oral gas concentration data such as ammonia C1, methane C2, hydrogen sulfide C3 and acetic acid C4 in the standard data set C are extracted and input into the support vector machine model for calculation, and the preliminary digestive tract risk score D1 is output. It is obtained through SVM training. The Sigmoid function maps the output value to between 0 and 1, reflecting the size of the digestive tract risk. The training of the machine learning model can not only perform weighted processing on different gas concentrations, but also fuse multiple gas concentration features to generate a comprehensive risk score D1, which provides a basis for the preliminary judgment of the digestive tract health status. Part of it is based on the standard data set C in the historical data table, and the preliminary risk threshold T1 is set according to the normal range of ammonia, methane, hydrogen sulfide and acetic acid concentrations. On this basis, the preliminary digestive tract risk score D1 is compared with the threshold T1 to determine whether the oral gas is abnormal. When the preliminary risk score D1 is less than the threshold T1, it indicates that there is no gastrointestinal abnormality and the patient can continue to be monitored regularly; when the preliminary risk score D1 is greater than or equal to the threshold T1, it indicates that the patient may be at risk of gastrointestinal disease and needs further in-depth evaluation. Through this process, combined with the machine learning model and risk threshold setting, the present invention realizes early screening and rapid evaluation of gastrointestinal diseases, improving the accuracy and real-time nature of diagnosis. Example 6

[0034] This embodiment is explained in Example 5. Please refer to Figure 1 ,Specifically: the depth analysis module includes a depth analysis unit and a depth evaluation unit; The deep analysis unit executes the deep evaluation mechanism when the preliminary comparison and evaluation shows that there is a disease risk in the digestive tract. The deep evaluation mechanism builds a deep learning model by using the LSTM network and the neural network, extracts the feature data set and the standard data set C through the write port, inputs them into the deep learning model, and calculates and outputs the deep digestive tract risk score D2; The deep digestive tract risk score D2 is calculated and output by the following algorithm formula; ; In the formula, represents the deep learning adjustment coefficient, f represents the nonlinear function, L LSTM Indicates long and short-term memory; LSTM receives a standard data set C = {C1(t), C2(t), C3(t), C4(t), ..., C i(t)} as input, learn the time dependency of these data, use historical data for training, and update the model parameters through the back-propagation algorithm, so that LSTM can accurately capture the pattern of gas concentration changes over time; The LSTM network helps deep learning models capture long-term trends in gas concentrations over time, which is particularly effective in detecting chronic or intermittent diseases. For example, LSTM can identify recurrent gastric ulcers and the long-term effects of chronic gastrointestinal diseases, which is of great significance for the timely detection of diseases.

[0035] The deep assessment unit accesses the digestive tract case system through the API application interface, extracts real case results, and analyzes the deep digestive tract risk scores D2 of a large number of patients and the impact of their actual health problems, such as the diagnosis results of gastrointestinal diseases, through big data analysis and machine learning models. The scores are highly correlated, and the threshold is automatically optimized through the training data set, and the secondary risk threshold T2 is preset. The deep digestive tract risk score D2 is then compared with the secondary risk threshold T2 for a secondary comparative assessment, and the deep correlation between the patient's digestive tract symptoms and gas concentration is deeply analyzed. The specific evaluation content is as follows; When the deep digestive tract risk score D2 ≥ the secondary risk threshold T2, it means that the patient's oral gas is abnormal and there is a risk of digestive tract disease. At this time, the diagnosis and treatment server sends a medical examination reminder to the client, prompting gastroscopy and blood tests, etc.; When the deep gastrointestinal risk score D2 is less than the secondary risk threshold T2, it indicates that the patient's oral gas risk is within an acceptable range. At this time, a regular detection reminder is sent to the client through the diagnosis and treatment server.

[0036] In this embodiment, the system constructs a deep learning model through the LSTM long short-term memory network and the neural network to further accurately evaluate the patient's digestive tract health status. The feature data set and the standard data set C are extracted through the write port, and these data are input into the deep learning model to calculate the deep digestive tract risk score D2. The score is calculated by the LSTM network model. The LSTM network captures the pattern of gas concentration changes over time by learning the long-term dependence of time series data, so that it can effectively identify the potential risks of chronic or intermittent diseases, such as repeated gastric ulcers and the long-term effects of chronic gastrointestinal diseases. The LSTM model can optimize its parameters through the back propagation algorithm, so that it can accurately predict the digestive tract health trend in long-term observation, and provide a scientific basis for the early detection and early warning of the disease. The digestive tract case system is accessed through the API interface, and the deep digestive tract risk score D2 of a large number of patients is subjected to regression analysis or classification analysis in combination with big data analysis and machine learning models to determine the high correlation with actual health problems (such as the diagnosis results of gastrointestinal diseases). Based on this analysis, the secondary risk threshold T2 is set, and the deep digestive tract risk score D2 is compared with the secondary risk threshold T2 to further evaluate the deep association between the patient's digestive tract symptoms and gas concentration. When the deep gastrointestinal risk score D2 ≥ the secondary risk threshold T2, it means that the patient's oral gas is abnormal and there is a risk of gastrointestinal disease. At this time, the system will send a medical examination reminder to the client through the diagnosis and treatment server, and recommend the patient to undergo gastroscopy, blood tests and other further diagnosis; and when the deep gastrointestinal risk score D2 < the secondary risk threshold T2, it means that the patient's gastrointestinal risk is still within an acceptable range. The system will send a regular inspection reminder to the client and continue monitoring. Through this deep evaluation mechanism, the present invention significantly improves the early diagnosis ability of gastrointestinal diseases. The LSTM network model provides an accurate tool for the prediction and monitoring of chronic diseases by capturing the long-term dependence of time series data, while big data analysis and regression analysis further enhance the adaptability and accuracy of the system to a large number of patient cases. The beneficial effect of the system is that through the accurate deep gastrointestinal risk score D2 and its secondary evaluation mechanism, it can timely identify potential gastrointestinal disease risks, provide personalized medical examination recommendations, greatly improve the accuracy and timeliness of disease warning, and effectively support early intervention and treatment of patients.

[0037] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations may be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A data analysis system for early diagnosis of digestive tract diseases, characterized by: It includes gas detection module, data processing module, preliminary analysis module and in-depth analysis module; The gas detection module uses an oral gas detection device to set a detection cycle, collects the patient's oral gas data in real time, and builds a diagnosis and treatment server, and transmits the oral gas data to the diagnosis and treatment server through a wireless transmission network; The data processing module receives the oral gas data in real time in the diagnosis and treatment server, pre-processes the oral gas data, obtains a standard data set C, extracts features from the standard data set C, obtains a feature data set, and constructs a time series database, sets a transmission port, and stores the standard data set C and the feature data set in the time series database through the transmission port; The preliminary analysis module constructs a machine learning model, extracts the standard data set C through the transmission port, inputs it into the machine learning model, outputs a preliminary digestive tract risk score D1, and presets a preliminary risk threshold T1, performs a preliminary comparative evaluation on the preliminary risk threshold T1 and the preliminary digestive tract risk score D1, and analyzes the normal range of gas concentration and the limit of abnormal fluctuation; When the result of the preliminary comparative evaluation is that the oral gas is abnormal, the deep analysis module constructs a deep learning model, extracts the feature data set through the transmission port, inputs it into the deep learning model, calculates and outputs a deep gastrointestinal risk score D2, and trains based on the deep learning model to obtain a secondary risk threshold T2, then performs a secondary comparative evaluation on the deep gastrointestinal risk score D2 and the secondary risk threshold T2, and deeply analyzes the deep correlation between the patient's gastrointestinal symptoms and gas concentration.

2. A data analysis system for early diagnosis of digestive tract diseases according to claim 1, characterized in that: The gas detection module includes a gas data acquisition unit and a data transmission unit; The gas data acquisition unit constructs a gas sensor array through an oral gas detection device. Each gas sensor responds to a specific oral gas, collects oral gas from a patient monitored by the digestive tract, obtains the patient's oral gas data, and sets the acquisition frequency to once every 5 minutes. The gas sensor array includes electrochemical gas sensors, semiconductor gas sensors and optical sensors; The oral gas data includes ammonia concentration C1, methane concentration C2, hydrogen sulfide concentration C3, acetic acid concentration C4, ..., the i-th oral gas C i ; The data transmission unit constructs a diagnosis and treatment server and adopts a wireless communication network WIFI to integrate and connect the diagnosis and treatment server with the sensor transmission interface of the gas sensor array, and records the data collection time of the gas sensor array to transmit the oral gas data to the diagnosis and treatment server.

3. A data analysis system for early diagnosis of digestive tract diseases according to claim 2, characterized in that: The data processing module includes a standardization processing unit, a data storage unit and a feature extraction unit; The standardization processing unit receives the oral gas data in real time in the diagnosis and treatment server, pre-processes the oral gas data, and integrates the pre-processed oral gas data to obtain a standard data set C; The preprocessing includes timestamp labeling, denoising, data standardization and time series smoothing; The standard data set C = {C1(t), C2(t), C3(t), C4(t), ..., C i (t)}; The timestamp tagging is to perform real-time timestamp on the collected oral gas data by recording the data collection time of the gas sensor array; The denoising is performed by removing random noise in the collected oral gas data using a filtering algorithm; The data standardization is performed by standardizing the oral gas data and making the units and magnitude differences of different oral gas data dimensionless; The time series smoothing optimizes the volatility of oral gas data in the short term by using exponential smoothing; The data storage unit constructs a time series database and sets a transmission port, wherein the transmission port includes a write port and a write port, and constructs a data storage table, wherein the data storage table includes a real-time data table and a historical data table. The standard data set C obtained after preprocessing is stored in the real-time data table in the time series database through the write port. At the same time, the original stored data in the real-time data table is automatically transferred to the historical data table and stored in order according to timestamps.

4. A data analysis system for early diagnosis of digestive tract diseases according to claim 3, characterized in that: The feature extraction unit extracts the standard data set C in the real-time data table and the historical data table through the write port, and performs feature extraction to obtain a feature data set, wherein the feature extraction includes extraction of the rate of change of gas concentration, extraction of gas concentration volatility, and extraction of the correlation between gas concentration and in vivo health indicators; Then, the feature data set is written into the time series database through the write port for data storage; The characteristic data set includes the gas concentration change rate R, the gas concentration fluctuation value Vc and the gas concentration correlation index p (C i , C j ), where C j represents the j-th oral gas data in the standard data set C; The feature extraction unit includes a gas concentration extraction unit, a gas concentration fluctuation extraction unit and a correlation extraction unit; The gas concentration extraction unit extracts all processed oral gas data in the standard data set C through the write port, performs concentration change extraction, obtains the gas concentration change rate R, and analyzes the change of each oral gas data; The rate of change R of the gas concentration is calculated and outputted by the following algorithm formula; ； In the formula, R i represents the rate of change of the gas concentration of the i-th oral gas data in the standard data set C, C i (t) represents the i-th oral gas data in the standard data set C at time t, C i (t-1) represents the i-th oral gas data in the standard data set C at time t-1, and △t represents the sampling time interval.

5. A data analysis system for early diagnosis of digestive tract diseases according to claim 4, characterized in that: The gas concentration fluctuation extraction unit extracts all processed oral gas data in the standard data set C through the write port, calculates the standard deviation at a certain time, obtains the gas concentration fluctuation value Vc, and analyzes the stability of the digestive tract; The gas concentration fluctuation value Vc is calculated and outputted by the following algorithm formula; ； Where N represents the total number of oral gas data in the time window, Vc i represents the gas concentration fluctuation value of the i-th oral gas data in the standard data set C, Represents the mean gas concentration of the i-th oral gas data in the standard data set C.

6. A data analysis system for early diagnosis of digestive tract diseases according to claim 5, characterized in that: The correlation extraction unit extracts different oral gas data in the standard data set C through the write port through correlation extraction, and calculates the output gas concentration correlation index p (C i , C j ), to analyze the relationship between different gases; The gas concentration correlation index p (C i , C j ) The output is calculated by the following algorithm formula; ； In the formula, represents the mean value of the i-th oral gas data in the standard data set C, C j represents the jth oral gas data in the standard data set C, Represents the mean of the j-th oral gas data in the standard data set C.

7. A data analysis system for early diagnosis of digestive tract diseases according to claim 6, characterized in that: The preliminary analysis module includes a preliminary detection calculation unit and a preliminary evaluation unit; The preliminary detection calculation unit constructs a machine learning model by using a support vector machine SVM, and extracts a standard data set C through the write port and inputs it into the machine learning model to calculate and output a preliminary digestive tract risk score D1, and perform preliminary detection of the gastrointestinal health of the digestive tract; The preliminary digestive tract risk score D1 is calculated and output by the following machine learning model; ； In the formula, C1(t) represents the value of ammonia concentration at time t, C2(t) represents the value of methane concentration at time t, C3(t) represents the value of hydrogen sulfide concentration at time t, and C4(t) represents the value of acetic acid concentration at time t. represents the Sigmoid function, which is used to map the output to the range of 0 to 1 to represent the risk score, , , and They respectively represent the coefficients to be learned of ammonia concentration C1, methane concentration C2, hydrogen sulfide concentration C3 and acetic acid concentration C4, which are obtained through machine learning training.

8. A data analysis system for early diagnosis of digestive tract diseases according to claim 7, characterized in that: The preliminary evaluation unit extracts the standard data set C in the historical data table through the write port, sets the preliminary risk threshold T1 according to the normal range of ammonia concentration C1, methane concentration C2, hydrogen sulfide concentration C3 and acetic acid concentration C4, and then performs a preliminary comparative evaluation of the preliminary digestive tract risk score D1 and the preliminary risk threshold T1 to perform a preliminary analysis of the digestive tract disease risk in oral gas. The specific evaluation content is as follows; When the initial gastrointestinal risk score D1 is less than the initial risk threshold T1, it means that there is no abnormality in oral gas, and no intervention and regular monitoring are required; When the preliminary gastrointestinal risk score D1 ≥ preliminary risk threshold T1, it indicates that there is an abnormality in oral gas and a disease risk in the digestive tract, and a deep assessment mechanism is executed.

9. A data analysis system for early diagnosis of digestive tract diseases according to claim 8, characterized in that: The depth analysis module includes a depth analysis unit and a depth evaluation unit; The deep analysis unit executes a deep assessment mechanism when the preliminary comparison and assessment shows that there is a disease risk in the digestive tract. The deep assessment mechanism constructs a deep learning model by using an LSTM network and a neural network, extracts a feature data set and a standard data set C through a write port, inputs them into the deep learning model, and calculates and outputs a deep digestive tract risk score D2; The deep digestive tract risk score D2 is calculated and output by the following algorithm formula; ； In the formula, represents the deep learning adjustment coefficient, f represents the nonlinear function, L LSTM Represents a long short-term memory network.

10. A data analysis system for early diagnosis of digestive tract diseases according to claim 9, characterized in that: The depth assessment unit accesses the digestive tract case system through an API application interface, extracts real case results, sets a secondary risk threshold T2 through big data analysis and machine learning models, and then conducts a secondary comparative assessment of the deep digestive tract risk score D2 and the secondary risk threshold T2, deeply analyzing the deep correlation between the patient's digestive tract symptoms and gas concentration. The specific assessment content is as follows; When the deep digestive tract risk score D2 ≥ the secondary risk threshold T2, it means that the patient's oral gas is abnormal and there is a risk of digestive tract disease. At this time, a medical examination prompt is sent to the client through the diagnosis and treatment server; When the deep gastrointestinal risk score D2 is less than the secondary risk threshold T2, it indicates that the patient's oral gas risk is within an acceptable range. At this time, a regular detection reminder is sent to the client through the diagnosis and treatment server.

Images