Analysis method and device based on dynamic borborygmus data, medium and program product

Abstract

The embodiment of the invention provides an analysis method and device based on dynamic borborygmus data, a medium and a program product, and relates to the field of intelligent medical treatment. The method comprises the following steps: acquiring borborygmus data of a sample collected in continuous time; dividing the borborygmus data into N continuous time period data arranged according to a time sequence based on the acquisition time; generating i accumulated data sets according to the data of the continuous time periods; for each accumulated data set in the i accumulated data sets, calculating at least one borborygmus characteristic value; and predicting whether the sample belongs to a prediction result of MAP according to the dynamic change trend of the borborygmus characteristic value. A series of cumulative data sets containing different historical lengths are constructed, the dynamic evolution trend of data is captured, and the change trend of characteristic values of the data is analyzed, so that the states of patients are classified more accurately and more timely.

Description

Technical Field

[0001] This invention relates to the field of intelligent healthcare, and more specifically, to an analysis method, device, medium, and program product based on dynamic bowel sound data. Background Technology

[0002] Acute pancreatitis (AP) is classified into three categories based on severity: Mild acute pancreatitis (MAP): without organ dysfunction or local or systemic complications, usually recovering within 1-2 weeks, with an extremely low mortality rate. Moderate to severe acute pancreatitis (MSAP): accompanied by transient (≤48 hours) organ dysfunction or local complications, with a low mortality rate. Severe acute pancreatitis (SAP): accompanied by persistent (>48 hours) organ failure, with a high mortality rate. Therefore, MAP is a relatively mild type and is clearly distinct from severe acute pancreatitis (SAP).

[0003] Bowel sound data is primarily collected through manual auscultation using a stethoscope. However, manual auscultation relies on the doctor's experience and may be subject to subjective bias. In recent years, digital technology has been increasingly applied to the collection and analysis of gastrointestinal sounds, converting bowel sounds into electrical signals for objective recording and analysis of their characteristics. However, current remote auscultation and analysis of bowel sounds often employs fixed time windows (e.g., every 5 minutes) for data analysis. This approach ignores the dynamic evolution of data over time, potentially failing to capture early, subtle signs of disease progression, leading to insufficient accuracy and timeliness in classification or early warning. Summary of the Invention

[0004] This invention aims to at least solve one of the technical problems existing in the prior art. To this end, this invention provides an analysis method, device, medium, and program product based on dynamic bowel sound data; the method of this invention constructs a series of cumulative datasets containing different historical lengths, captures the dynamic evolution trend of the data, and analyzes the changing trends of its feature values, thereby classifying the patient's condition more accurately and timely.

[0005] The first aspect of this application discloses an analysis method based on dynamic bowel sound data, the method comprising:

[0006] Acquire bowel sound data of the test sample collected over a continuous period of time;

[0007] Based on the collection time, the bowel sound data is divided into N consecutive time periods arranged in chronological order; where, for any two adjacent time periods, the end time of the previous time period is the start time of the next time period.

[0008] i cumulative datasets are generated based on data from consecutive time periods; wherein, the i-th cumulative dataset contains all data from the first time period up to the i-th time period;

[0009] For each of the i cumulative datasets, calculate at least one bowel sound feature value;

[0010] This is an auxiliary prediction result that uses the dynamic trend of bowel sound characteristic values ​​to predict whether a sample belongs to MAP or non-MAP.

[0011] In some embodiments, bowel sound characteristic values ​​include bowel rate, amplitude, and interval time at fixed intervals;

[0012] i is a natural number greater than or equal to 1, N is a natural number greater than 2, and i is less than or equal to N.

[0013] In some embodiments, bowel sound characteristic values ​​also include: duration and frequency.

[0014] In some embodiments, the dynamic trend includes the slope of the eigenvalue over time, the difference or variance between consecutive eigenvalues.

[0015] In some embodiments, the fixed time includes: every 6 hours, every 12 hours, every 24 hours, and every 48 hours.

[0016] In some embodiments, when the fixed time is 12h, within a preset time, as the accumulated dataset increases, if the growth rate of the intestinal rate value is greater than a first threshold, the decrease rate of the amplitude is greater than a second threshold, the shortening rate of the interval time is greater than a third threshold, and the extension rate of the duration is greater than a fourth threshold, an auxiliary prediction result indicating that the test sample belongs to MAP or non-MAP is better is output.

[0017] In some embodiments, acquiring bowel sound data of a test sample collected over a continuous time period includes:

[0018] Receive bowel sound data transmitted by the bowel sound patch via Bluetooth; the bowel sound data is captured when the bowel sound patch is applied to the lower right abdomen of the sample to be tested.

[0019] A second aspect of this application discloses a computer device, comprising: a memory and a processor; the memory is used to store a computer program; the processor executes the computer program to implement the steps of the above-described method.

[0020] A third aspect of this application discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0021] The fourth aspect of this application discloses a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method.

[0022] This application has the following beneficial effects:

[0023] This application innovatively discloses an analysis method based on dynamic bowel sound data. This method constructs a series of cumulative datasets containing data of varying historical lengths, captures the dynamic evolution trend of the data, and analyzes the changing trends of its characteristic values. This ensures the completeness and temporal rigor of the collected bowel sound data, overcoming the shortcomings of existing technologies that collect data based on fixed time windows. These technologies ignore the dynamic evolution process of data accumulation over time, thus failing to capture early subtle signs of disease changes in a timely manner, leading to insufficient accuracy and timeliness in classification or early warning. This judgment based on "trend" rather than "instantaneous value" has higher robustness and accuracy, effectively solving the problem that clinicians do not know exactly how early bowel sounds weaken. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the method flow provided in the first aspect of the present invention;

[0026] Figure 2 This is a schematic diagram of the system provided in an embodiment of the present invention;

[0027] Figure 3 This is a schematic diagram of a computer device provided in an embodiment of the present invention;

[0028] Figure 4 This is a schematic diagram of the architecture of an exemplary computing device provided in an embodiment of the present invention;

[0029] Figure 5 This is a schematic diagram of the storage medium provided in an embodiment of the present invention;

[0030] Figure 6 This is a dynamic change graph of intestinal rate in three groups of patients at different onset times, provided in an embodiment of the present invention.

[0031] Figure 7 This is a dynamic change graph showing the range of three groups of patients at different onset times, provided in an embodiment of the present invention.

[0032] Figure 8 This is a dynamic change graph of the time interval between different onset times for three groups of patients provided in an embodiment of the present invention;

[0033] Figure 9This is a heatmap of correlation analysis between bowel sound characteristics and disease severity, provided by an embodiment of the present invention, which segments bowel sound data every 12 hours based on the onset time.

[0034] Figure 10 These are ROC curves for model construction using different algorithms, as provided in this embodiment of the invention.

[0035] Figure 11 This is a SHAP plot provided in an embodiment of the present invention, which displays the relative contribution of predictor variables to model results. Detailed Implementation

[0036] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0037] In some of the processes described in the specification, claims, and accompanying drawings of this invention, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0038] 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.

[0039] Figure 1 This is a schematic flowchart of an analysis method based on dynamic bowel sound data provided by an embodiment of the present invention. Specifically, the method includes the following steps:

[0040] S101, acquire bowel sound data of the sample to be tested collected over a continuous time period.

[0041] In some embodiments, the terms “subject” or “test subject” or “sample” as used herein refer to any animal (e.g., a mammal), including but not limited to humans, non-human primates, rodents, etc., which will become the recipient of a particular treatment. Generally, the terms “subject” and “patient” are used interchangeably herein when referring to human subjects. Preferably, the sample to be tested is a human.

[0042] In some embodiments, the method for collecting bowel sound data includes: attaching a bowel sound patch to the lower right abdomen of the sample to be tested to capture and record bowel sound data; transmitting the bowel sound data to a receiver of a terminal via Bluetooth, so that the terminal can acquire the bowel sound data and perform subsequent data processing on the terminal.

[0043] S102, based on the acquisition time, divides the bowel sound data into N consecutive time periods arranged in chronological order.

[0044] For any two adjacent time periods, the end time of the previous time period is the start time of the next time period. For example, when N is 4, the first time period, the second time period, the third time period, and the fourth time period are consecutive. That is, the end time T1_end of the first time period is the same as the start time T2_start of the second time period, the end time T2_end of the second time period is the same as the start time T3_start of the third time period, and so on.

[0045] The key is that these time periods are seamlessly connected; that is, the end of the first time period is the beginning of the second time period, the end of the second time period is the beginning of the third time period, and so on. This continuity ensures the integrity of the data and the rigor of the time sequence.

[0046] S103, generate i cumulative datasets based on data from continuous time periods.

[0047] The i-th cumulative dataset contains all data from the first time period up to the i-th time period. Specifically, when i is 4, the first dataset contains only the first data from the first time period; the second dataset contains the second data from the first and second time periods; the third dataset contains the third data from the first three time periods, and so on, forming a sequence with progressively increasing data volume. This design ensures that each analysis includes all historical information from the beginning to the current moment.

[0048] In some embodiments, when i=1, it is the first data containing monitoring data of the first time period; when i=2, it is the second data containing monitoring data of the first time period and the second time period; when i=3, it is the third data containing monitoring data of the first time period, the second time period and the third time period, and so on.

[0049] S104, for each of the i cumulative datasets, calculate at least one bowel sound feature value.

[0050] In some embodiments, the bowel sound characteristic values ​​include bowel rate, amplitude, and interval time at fixed intervals; in some embodiments, the fixed intervals include: every 6 hours, every 12 hours, every 24 hours, and every 48 hours; preferably every 12 hours.

[0051] In some embodiments, bowel sound characteristic values ​​also include: duration and frequency.

[0052] In some embodiments, i is a natural number greater than or equal to 1, N is a natural number greater than 2, and i is less than or equal to N;

[0053] S105, based on the dynamic trend of bowel sound characteristic values, predicts whether the sample to be tested belongs to MAP or non-MAP.

[0054] Non-MAP refers to AP states other than mild acute pancreatitis (MAP), such as moderate severe acute pancreatitis (MSAP) and severe acute pancreatitis (SAP).

[0055] This judgment, based on "trend" rather than "instantaneous value," has greater robustness and accuracy.

[0056] In some embodiments, the dynamic trend includes the slope of the eigenvalue over time, the difference or variance between consecutive eigenvalues.

[0057] In some embodiments, when the fixed time is 12 hours, within a preset time period, as the accumulated dataset increases, if the growth rate of the intestinal rate value is greater than a first threshold (i.e., if the intestinal rate value shows an upward trend with the increase of the accumulated dataset, especially from the 36th to the 72nd hour), the rate of decrease in amplitude is greater than a second threshold (i.e., the amplitude shows a downward trend), the rate of shortening of the interval time is greater than a third threshold (i.e., the interval time shows a shortening trend), and the rate of prolongation of the duration is greater than a fourth threshold (i.e., the duration shows a prolongation trend), an auxiliary prediction result indicating that the test sample belongs to MAP or non-MAP is better is output. Based on clinical experience, patients with pancreatitis will experience intestinal paralysis in the early stages, and bowel sounds will be significantly reduced. The more severe the patient, the slower the recovery of intestinal function. The recovery of intestinal function is also an important indicator affecting the patient's prognosis.

[0058] Based on clinical experience, we have a preliminary understanding of the patient's bowel sounds, intervals, and duration. It is expected that the bowel sounds and bowel sounds will gradually increase as the patient recovers, with shorter intervals and longer durations.

[0059] The current understanding is that the more severe the intestinal inhibition, the stronger it is. However, our preliminary analysis has also found that the more severe the patient, the slower the recovery of bowel sounds. In fact, many severe patients have not recovered their intestinal function for seven days. Therefore, the weaker the average bowel sounds, the more severe the disease may be and the worse the prognosis may be.

[0060] In some specific embodiments, when i is 4, the bowel sound characteristic values ​​of the first data, the second data, the third data, and the fourth data are calculated respectively, for example:

[0061] Intestinal rate sequence: R1, R2, R3, R4;

[0062] Average duration sequence: D1, D2, D3, D4;

[0063] Average interval time series: I1, I2, I3, I4;

[0064] The processor analyzes the dynamic trends of these sequences. For example, it calculates the slope k_R = (R4 - R1) / 3 of the intestinal rate sequence. If k_R is greater than a preset threshold K_th, and the slope k_I of the average interval time series is less than a preset negative threshold -I_th, then it is determined to be a MAP; otherwise, it is determined to be non-MAP.

[0065] Can bowel sound monitoring devices help predict whether patients will develop severe illness later in life? Can they explore whether early bowel sounds are related to disease progression, predict disease outcome, and assist in clinical intervention? The devices are known, but long-term detection of bowel sounds is crucial in determining the direct relationship between changes in early bowel sounds and the development of severe illness.

[0066] Existing technologies cannot perform long-term dynamic monitoring and have long monitoring and analysis times. This application proposes a technology for remote, long-term, and real-time monitoring, which quantifies bowel sounds, including: bowel rate, frequency, amplitude, duration, and interval.

[0067] In some embodiments, the auxiliary prediction results include, but are not limited to, paper or electronic reports. These results are obtained by intelligent machines based on the relevant data of the subjects and are intended only as a reference for medical personnel, not as the final diagnosis results of the subjects.

[0068] In some embodiments, the threshold is obtained through training on training set samples. It can be a specific threshold or an interval range. The specific form is not specifically limited in this embodiment.

[0069] Figure 3 This is a schematic diagram of a computer device provided in an embodiment of the present invention, such as... Figure 3 As shown, the device 2000 may include: one or more processors 2010 and one or more memories 2020; wherein the memories store computer-readable code that, when run by the one or more processors, can perform the methods described above.

[0070] The processor in this embodiment can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, operations, and logic block diagrams disclosed in this embodiment. The general-purpose processor can be a microprocessor or any conventional processor, and can be based on an x86 or ARM architecture.

[0071] In general, the various exemplary embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, firmware, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of embodiments of this disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0072] For example, the method or apparatus according to embodiments of this disclosure can also be used by means of Figure 4 The architecture of the computing device 3000 shown is used for implementation. For example... Figure 4 As shown, the computing device 3000 may include a bus 3010, one or more CPUs 3020, a read-only memory (ROM) 3030, a random access memory (RAM) 3040, a communication port 3050 connected to a network, an input / output component 3060, a hard disk 3070, etc. The storage devices in the computing device 3000, such as the ROM 3030 or the hard disk 3070, may store various data or files used for processing and / or communication of the methods provided in this disclosure, as well as program instructions executed by the CPU. The computing device 3000 may also include a user interface 3080. Of course, Figure 4 The architecture shown is merely exemplary and can be omitted as needed when implementing different devices. Figure 4 One or more components in the computing device shown.

[0073] This invention also includes a computer-readable storage medium, such as... Figure 5 The diagram illustrates a storage medium 4000 provided in an embodiment of the present invention. The computer storage medium 4020 stores computer-readable instructions 4010. When the computer-readable instructions 4010 are executed by a processor, the method described above according to embodiments of the present disclosure can be performed. The computer-readable storage medium in the embodiments of the present disclosure may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), Synchronous Link Dynamic Random Access Memory (SLDRAM), and Direct Memory Bus Random Access Memory (DR RAM). It should be noted that the memory used in the methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0074] This disclosure also provides a computer program product or system, including a computer program that, when executed by a processor, implements the steps of the above-described method, such as... Figure 2 As shown, the computer program product or computer program includes an analysis system based on dynamic bowel sound data:

[0075] Bowel sound data acquisition module 201 is used or configured to acquire bowel sound data of a sample to be tested collected over a continuous time period;

[0076] The data partitioning module 202 is used or configured to partition the bowel sound data into N consecutive time periods arranged in chronological order based on the acquisition time, wherein for any two adjacent time periods, the end time of the previous time period is the start time of the next time period.

[0077] The cumulative dataset generation module 203 is used or configured to generate i cumulative datasets based on data from consecutive time periods; wherein the i-th cumulative dataset contains all data from the first time period up to the i-th time period;

[0078] Feature extraction module 204 is used or configured to calculate at least one bowel sound feature value for each of the i cumulative datasets;

[0079] The result prediction module 205 is used or configured to predict, based on the dynamic trend of bowel sound characteristic values, whether the sample to be tested belongs to MAP or non-MAP.

[0080] Specifically, the following are the specific research methods:

[0081] This study included 61 patients who visited the outpatient and emergency departments of Peking Union Medical College Hospital, Beijing Sixth Hospital, and Beijing Puren Hospital. Among them, 21 had mild acute bowel sounds (AP), 26 had moderate to severe AP, and 14 had severe AP. All patients were admitted within 48 hours of symptom onset. Bowel sound monitoring was started on the day of admission (Day 0) for patients meeting the inclusion criteria. A bowel sound patch was applied to the lower right abdomen of the patient to capture and record bowel sounds, which were transmitted to a receiver via Bluetooth and processed at the terminal.

[0082] Inclusion criteria:

[0083] Patients must meet the diagnostic criteria for AP; be aged 18-85 years; have had AP within 48 hours of onset; and have or have family members who understand the study protocol and are willing to participate in the study, and have signed written informed consent.

[0084] Exclusion criteria:

[0085] Patients with severe conditions requiring non-medical treatments such as surgery or peritoneal lavage; patients with severe hypertension or diabetes that are poorly controlled with medication; pregnant women; patients with comorbid muscle and nerve disorders, chronic pancreatitis, inflammatory bowel disease, cancer, irritable bowel syndrome, etc.; patients with serious primary diseases of the digestive, respiratory, cardiovascular, or hematologic systems, as well as mental illness or infectious diseases; patients with a history of abdominal surgery (excluding appendectomy or cholecystectomy); patients who have used antibiotics, probiotics, or traditional Chinese medicine preparations within 2 months prior to the onset of illness; patients allergic to silicone, PC materials, or 3M dressings; patients with significantly missing clinical data or who have left the hospital without authorization.

[0086] Data is transmitted to a receiver via Bluetooth, which then uploads the data to a server. The server runs accompanying software to store the received data and record relevant information. For the stored sound, the MFCC (MFCC formula below) is extracted, and a CRNN neural network (consisting of 5 convolutional layers, a bidirectional GRU, and a fully connected layer) is built to identify the bowel sound features in the sound file, thereby obtaining bowel sound segments.

[0087] A = melfilter * X = [a1, a2, ... a 26 ];

[0088] MFCC = DCT(A);

[0089] The MFCC feature extraction algorithm uses 26 Mel filters to obtain Mel spectrum coefficients A, and then performs discrete cosine transform on Mel spectrum coefficients A to obtain MFCC values.

[0090] A continuous auscultation recorder system was used to record the bowel sound level of patients every minute, including multiple dimensions such as bowel rate and interval time. Based on the characteristics of the patients' conditions, the data were integrated into an average bowel sound level every 12 hours or every 24 hours, and correlation analysis was performed with clinical prognostic data.

[0091] Explanation of dimensional metrics:

[0092] ① Intestinal rate: The number of bowel sounds per minute, unit: times / min.

[0093] ② Duration: The duration of a single bowel sound, measured in seconds (s).

[0094] ③ Amplitude: The width of the waveform of bowel sounds oscillating in the frequency domain, measured in decibels (dB).

[0095] ④ Frequency: The characteristic frequency of the bowel sound waveform in the frequency domain, in Hertz (Hz).

[0096] ⑤ Interval: The time interval between two bowel sounds, in seconds (s).

[0097] Taking the onset time as an example, we conducted a detailed analysis of bowel sounds based on the severity of the disease. Table 1 shows the number of patients in each time period, with the left column representing the onset time. 12 means 0-12 hours, and so on.

[0098] Table 1. Statistics on the number of patients with MAP-MSAP-SAP at different onset time based on the onset time.

[0099]

[0100] Tables 2-6, Appendix Figure 6-8 This refers to the dynamic changes in bowel rate, duration, frequency, amplitude, and interval at different onset times in three groups of patients. Figure 9 This is a heatmap showing the correlation between bowel sound characteristics and disease severity, segmented into 12-hour intervals based on the onset time. The left column represents the onset time, with 12 indicating 0-12 hours, and so on.

[0101] Table 2 shows the dynamic changes in bowel rate (Rate) at different onset times in the three groups of patients.

[0102]

[0103] Table 3 shows the dynamic changes in duration of illness at different onset times for the three groups of patients.

[0104]

[0105] Table 4 shows the dynamic changes in amplitude at different onset times for the three groups of patients.

[0106]

[0107] Table 5 shows the dynamic changes in frequency at different onset times for the three groups of patients.

[0108]

[0109] Table 6 shows the dynamic changes in the interval between different onset times for the three groups of patients.

[0110]

[0111] In the baseline description, normally distributed continuous variables are expressed as mean ± standard deviation, and t-tests are used for comparisons between groups; non-normally distributed data are expressed as median (interquartile range), and Wilcoxon rank-sum tests are used for comparisons between groups; categorical variables are expressed as frequency (percentage), and Pearson chi-square tests are used for comparisons between groups. All continuous variables are standardized (mean = 0, standard deviation = 1) to eliminate the influence of dimensions.

[0112] The predictive model was built based on a cohort of 59 acute bowel sound (AP) patients (28 with macrosomia and 31 with non-MAP). The dataset was randomly divided into a training set (model building) and an independent test set (validation set) in an 8:2 ratio. It's important to note that this differs from the previous 61 patients because two patients were only monitored for the first 24 hours after admission, not from 24 to 48 hours. Therefore, the model prediction only included 59 patients. In model building, taking a fixed 24-hour period as an example, the model used the average of continuous time periods (e.g., 0-24h, 24-48h) for prediction (equivalent to 10 dimensions of data), which meets the clinical requirement for continuous changes in bowel sound characteristics.

[0113] Machine learning methods:

[0114] Six algorithms were evaluated for model building: Resilient Network Logistic Regression (ENLR), K-Nearest Neighbors (KNN), Decision Tree, Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost). These methods encompass complementary strengths ranging from linear regularization (ENLR) to ensemble-based nonlinear pattern recognition (XGBoost, RF), ensuring a comprehensive exploration of the boundaries of diagnostic decisions. ENLR combines the advantages of Lasso and Ridge regularization to enhance generalization ability; the KNN algorithm is concise, intuitive, and easy to implement; Decision Tree provides an interpretable decision-making process; Random Forest reduces the risk of overfitting by ensemble multiple trees, enhancing generalization performance and performing well on complex clinical data; SVM excels in high-dimensional spaces and small sample scenarios; and XGBoost, as an efficient gradient boosting algorithm, is robust to outliers and has an anti-overfitting regularization mechanism, making it particularly suitable for complex clinical prediction tasks.

[0115] Model training employed Bayesian optimization for hyperparameter tuning, based on the average performance of 100 five-fold cross-validations. The model parameters were configured as follows: ENLR exploration regularization strength C (0.01-10) and L1 / L2 penalty balance parameter l1_ratio (0-1); KNN optimization of nearest neighbors n_neighbors (3-21) and weight parameters weights (uniform / distance); key parameters for decision trees included max_depth (1-10), min_samples_leaf (1-10), and min_samples_split (2-20); random forest required adjustment of n_estimators. (50-300) and limit max_depth (1-10); SVM test regularization parameter C (0.01-100) and kernel coefficient gamma (0.01-10); XGBoost optimization involves n_estimators (50-300), max_depth (1-10), learning_rate (0.01-0.3), colsample_bytree (0.5-1), reg_lambda (0-1) and min_child_weight (0-10).

[0116] Model training was conducted on the Python 3.12.2 platform, using the scikit-learn 1.4.1 package (ENLR, KNN, decision tree, random forest, SVM), the xgboost 2.1.3 package (XGBoost), and the hyperopt 0.2.7 package (Bayesian optimization).

[0117] Model performance evaluation:

[0118] The optimally configured model was used to quantify performance on the test set using six metrics: accuracy, precision, recall, F1 score, mean precision (AP), and area under the receiver operating characteristic (AUC). Accuracy reflects overall predictive correctness; precision represents the proportion of true positives among predicted positives, which is crucial for reducing unnecessary interventions; recall assesses the model's ability to identify true positives, helping to reduce the risk of missed diagnoses; the F1 score, as the harmonic mean of precision and recall, has particular value in clinical screening scenarios where a balance needs to be struck between false positives (risk of overtreatment) and false negatives (delayed diagnosis); AP integrates the precision-recall tradeoff at different probability thresholds; and AUC assesses discriminative ability by measuring the relationship between the true positive rate (sensitivity) and the false positive rate (1-specificity) at different thresholds.

[0119] like Figure 10For example, using the AUC score as the main performance indicator for model evaluation, the results show that KNN, ENLR, and SVM perform better. To ensure statistical reliability, all performance indicators were evaluated for variability by calculating 95% confidence intervals (CI) through bootstrap resampling (1000 times) of the test set.

[0120] Model interpretation and subgroup analysis:

[0121] The SHapley Additive Explanation (SHAP) framework is used to analyze the decision logic of the optimal algorithm. This game-theoretic feature attribution method can quantify the relative contribution of individual predictor variables to the model results, enabling a refined interpretation of the risk stratification mechanism. SHAP values ​​are calculated using the shap 0.46.0 Python package. Taking 48 hours as an example, the results for the five dimensions are as follows: Figure 11 As shown.

[0122] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0123] In general, the various exemplary embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, firmware, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of embodiments of this disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0124] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0125] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0126] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0127] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0128] The exemplary embodiments of this disclosure described in detail above are merely illustrative and not restrictive. Those skilled in the art will understand that various modifications and combinations can be made to these embodiments or their features without departing from the principles and spirit of this disclosure, and such modifications should fall within the scope of this disclosure.

Claims

1. An analysis method based on dynamic bowel sound data, characterized in that, The method includes: Acquire bowel sound data of the test sample collected over a continuous period of time; The bowel sound data is divided into N consecutive time periods arranged in chronological order based on the collection time; wherein, for any two adjacent time periods, the end time of the previous time period is the start time of the next time period. i cumulative datasets are generated based on the data from the continuous time periods; wherein, the i-th cumulative dataset contains all data from the first time period up to the i-th time period; For each of the i cumulative datasets, calculate at least one bowel sound feature value; Based on the dynamic change trend of the bowel sound characteristic values, an auxiliary prediction result is made to predict whether the sample to be tested belongs to MAP or non-MAP.

2. The analysis method based on dynamic bowel sound data according to claim 1, characterized in that, The bowel sound characteristic values ​​include the bowel rate, amplitude, and interval time at fixed intervals; Where i is a natural number greater than or equal to 1, N is a natural number greater than 2, and i is less than or equal to N.

3. The analysis method based on dynamic bowel sound data according to claim 2, characterized in that, The bowel sound characteristic values ​​also include: duration and frequency.

4. The analysis method based on dynamic bowel sound data according to claim 3, characterized in that, The dynamic trend includes the slope of the feature value changing over time, the difference or variance between consecutive feature values.

5. The analysis method based on dynamic bowel sound data according to claim 2, characterized in that, The fixed times include: every 6 hours, every 12 hours, every 24 hours, and every 48 hours.

6. The analysis method based on dynamic bowel sound data according to claim 5, characterized in that, When the fixed time is 12h, within the preset time, as the accumulated dataset increases, if the growth rate of the intestinal rate value is greater than the first threshold, the decrease rate of the amplitude is greater than the second threshold, the shortening rate of the interval time is greater than the third threshold, and the extension rate of the duration is greater than the fourth threshold, the auxiliary prediction result of whether the sample to be tested belongs to MAP or non-MAP will be output.

7. The analysis method based on dynamic bowel sound data according to claim 1, characterized in that, The acquisition of bowel sound data of the test sample collected over a continuous time period includes: The device receives bowel sound data transmitted via Bluetooth from the bowel sound patch; the bowel sound data is captured when the bowel sound patch is applied to the lower right abdomen of the sample to be tested.

8. A computer device, characterized in that, The device includes: a memory and a processor; the memory is used to store a computer program; the processor executes the computer program to implement the steps of the method according to any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the steps of the method as described in any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the method described in any one of claims 1-7.

Images