Intelligent urine monitoring system for newborn urine collection and use method

Abstract

According to the intelligent urine monitoring system for newborn urine collection and the use method, a urine collection device is arranged between two legs of a newborn, the urine of the newborn is collected in an all-weather mode, and after the collected urine flows into a detection area through a pipeline, the urine collection device is used for collecting urine of the newborn; a data acquisition module for a detection area acquires first-level pre-processing data of urine in the detection area, the first-level pre-processing data is combined with urine times to generate multi-level sub-pre-processing data, and then the first-level pre-processing data or the multi-level sub-pre-processing data is received through a data analysis module; a neural network model is adopted to analyze primary preprocessed data or multi-level sub-preprocessed data, and finally a result pushing module is adopted to receive an analysis result of the primary preprocessed data or the multi-level sub-preprocessed data and send the analysis result to a monitoring terminal to solve the problem that existing equipment cannot guarantee timely collection of urine, so that the accuracy of urine collection is improved. The technical problem of monitoring data deviation caused by the fact that urine is infected by the external environment is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of urine analysis, and in particular to an intelligent urine monitoring system for collecting urine from newborns and a method for using the system. Background Art

[0002] Newborns have immature renal function and are prone to serious complications from dehydration, electrolyte imbalances, or infection. Monitoring urine volume, pH, color, and electrolytes can help identify risks such as dehydration, acidosis, and sepsis early, allowing for timely intervention to prevent organ damage or death. Because urine output is low and newborns are less likely to express discomfort, noninvasive, real-time monitoring can mitigate the lags of traditional methods, providing clinically accurate data and ensuring newborn safety.

[0003] However, newborns cannot actively cooperate like adults and cannot collect urine normally, so it is usually necessary to use a collection bag or quickly collect mid-stream urine samples when changing diapers. However, this also cannot guarantee timely collection of urine to prevent urine from being contaminated by the external environment, resulting in deviations in monitoring data. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent urine monitoring system for neonatal urine collection and a method of use, so as to solve the problem that urine cannot be collected in time and urine is not contaminated by the external environment, resulting in deviations in monitoring data.

[0005] The technical solution of the present invention is achieved as follows: A first aspect of the present invention provides an intelligent urine monitoring system for collecting urine from newborns, comprising: A urine collection device for collecting urine produced by the newborn in real time at various time periods and sealing the urine, wherein the urine flows into the detection area through a pipe; a data acquisition module, configured to obtain primary pre-processing data of the urine in the detection area, wherein the primary pre-processing data is combined with the number of urine collection times to generate multi-level sub-pre-processing data; a data analysis module, configured to receive the primary preprocessed data or the multi-level sub-preprocessed data, and analyze the primary preprocessed data or the multi-level sub-preprocessed data using a neural network model; The result push module is used to receive the analysis results of the first-level pre-processed data or the multi-level sub-pre-processed data, and send the analysis results to the monitoring terminal.

[0006] A further technical solution is that the urine collection device includes a collecting head, the top of the collecting head is connected to the pipe, the bottom of the pipe is connected to the detection area, and the detection area is provided with a liquid slide corresponding to the data acquisition module.

[0007] A further technical solution is that a recovery area is provided at the bottom of the collecting head, and the recovery area is connected to the detection area through a soft pressure component.

[0008] A further technical solution is that the soft pressure component includes a recovery box located in the recovery area, the top of the recovery box is connected to the detection area through a pull-out tube, a sealing element is provided on the top of the pull-out tube, and a block is provided at the bottom of the pull-out tube for detachable connection with the top of the recovery box, the sealing element is connected to the bottom surface of the detection area through a spring, and the sealing element can be tightly connected to the output port of the detection area.

[0009] A further technical solution is that the data acquisition module includes: an optical acquisition unit, configured to acquire the primary preprocessing data or the multi-level sub-preprocessing data, wherein the primary preprocessing data or the multi-level sub-preprocessing data include physical indicators and chemical indicators; a first data processing unit for recording a timestamp and a urine collection frequency of the primary preprocessed data to generate the multi-level sub-preprocessed data; The second data processing unit is used to eliminate abnormal values and perform standardization on the first-level preprocessed data or the multi-level sub-preprocessed data.

[0010] A further technical solution is that the specific steps for recording the timestamp and urine collection times of the first-level pre-processed data to generate the multi-level sub-pre-processed data include: Acquiring the first-level preprocessing data of the optical acquisition unit; According to the table characteristics of the first-level preprocessing data and the number of times the sensor receives the data; identifying a timestamp and urine collection times based on the primary pre-processed data; The multi-level sub-preprocessing data is generated according to the timestamp and the number of urine collection times.

[0011] A further technical solution is that the data analysis module includes: A data receiving unit, used for the first-level pre-processed data or the multi-level sub-pre-processed data; An algorithm analysis unit is used to perform model analysis on the first-level preprocessed data or the multi-level sub-preprocessed data through the neural network model.

[0012] A further technical solution is to use a random forest algorithm, wherein the steps of performing model analysis on the primary preprocessed data or the multi-level sub-preprocessed data by using the neural network model include: Initializing a random forest and setting core parameters of the first-level preprocessed data or the multi-level sub-preprocessed data; Constructing multiple decision trees using the first-level preprocessed data or the multi-level sub-preprocessed data, and selecting an optimal split feature from multiple random features for recursive splitting when each decision tree splits; By iterating the decision tree splitting multiple times until a stopping condition is met, a leaf node output category label is generated; After each decision tree outputs a prediction result, the classification task integrates the final result through the majority voting method, and the regression task integrates the result through the prediction value averaging method.

[0013] A further technical solution is to also include a Bayesian network: Construct Bayesian networks based on medical knowledge or statistical methods; The causal relationships between metabolites are explicitly modeled through a directed acyclic graph, and the conditional probabilities of the nodes are calculated as structured features. Merging the structured features with the original first-level preprocessed data or the multi-level sub-preprocessed data to generate a training set; Inputting the training set into the random forest for training; The random forest processes high-dimensional nonlinear relationships by integrating multiple decision trees and outputs disease risk predictions.

[0014] A second aspect of the present invention provides a method for using an intelligent urine monitoring system for collecting urine from a newborn, comprising the following steps: collecting urine produced by the newborn in various time periods in real time and sealing the urine, wherein the urine flows into the detection area through a pipe; Acquiring primary preprocessing data of the urine in the detection area, wherein the primary preprocessing data is combined with the number of urine collection times to generate multi-level sub-preprocessing data; Receiving the primary preprocessed data or the multi-level sub-preprocessed data, and analyzing the primary preprocessed data or the multi-level sub-preprocessed data using a neural network model; Receive the analysis results of the first-level preprocessing data or the multi-level sub-preprocessing data, and send the analysis results to the monitoring terminal.

[0015] The beneficial effects of the present invention are: The urine collection device is installed between the legs of the newborn to collect the newborn's urine around the clock. After the collected urine flows into the detection area through a pipe, the data collection module of the detection area obtains the first-level preprocessing data of the urine in the detection area. The first-level preprocessing data is combined with the number of urine times to generate multi-level sub-preprocessing data. Subsequently, the first-level preprocessing data or multi-level sub-preprocessing data is received by the data analysis module, and the first-level preprocessing data or multi-level sub-preprocessing data is analyzed by a neural network model. Finally, the result push module is used to receive the analysis results of the first-level preprocessing data or multi-level sub-preprocessing data, and the analysis results are sent to the monitoring terminal, so as to solve the technical problem that the existing equipment cannot guarantee the timely collection of urine and avoid the urine being contaminated by the external environment, which leads to deviations in the monitoring data. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A block diagram of an intelligent urine monitoring system for collecting urine from newborns provided in accordance with the first embodiment of the present invention; Figure 2 A schematic structural diagram of a urine collection device provided in Embodiment 2 of the present invention; Figure 3 for Figure 2 Enlarged view of point A in the middle; Figure 4 A block diagram of an intelligent urine monitoring system for collecting urine from newborns provided in embodiment 3 of the present invention; Figure 5 This is a flowchart of the steps of using an intelligent urine monitoring system for collecting urine from newborns provided in Example 4 of the present invention.

[0017] In the figure, 1. Collection head; 2. Pipe; 3. Detection area; 4. Glass slide; 5. Recovery box; 6. Pull-out tube; 7. Sealing element; 8. Spring; 9. Block. DETAILED DESCRIPTION

[0018] In order to better understand the technical content of the present invention, specific embodiments are provided below, and the present invention is further described in conjunction with the accompanying drawings. Example 1

[0019] See also Figure 1In a first aspect, the present invention provides an intelligent urine monitoring system for collecting urine from newborns, comprising: a urine collection device for collecting urine produced by newborns in various time periods in real time and sealing the urine, wherein the urine flows into a detection area through a pipe; a data collection module for acquiring primary preprocessing data of urine in the detection area, wherein the primary preprocessing data is combined with the number of urine collection times to generate multi-level sub-preprocessing data; a data analysis module for receiving the primary preprocessing data or the multi-level sub-preprocessing data, and analyzing the primary preprocessing data or the multi-level sub-preprocessing data using a neural network model; and a result push module for receiving analysis results of the primary preprocessing data or the multi-level sub-preprocessing data, and sending the analysis results to a monitoring terminal.

[0020] The primary pre-processed data refers to the first data collection of urine by the data collection module, and the urine is the first urine monitoring data collected in this monitoring collection.

[0021] Multi-level sub-preprocessing data refers to the monitoring data after the first data collection of the first collected urine, which is then superimposed on the urine output by the subsequent newborn.

[0022] The monitoring terminal is a PC terminal used by the hospital, or a mobile terminal used by family members that is connected to the system used by the hospital PC terminal, such as the mobile terminal used by the family members is bound to the public account generated by the PC terminal.

[0023] It should be noted that the first-level preprocessing data and the multi-level sub-preprocessing data are different data values. Therefore, it is necessary to use the data analysis module to analyze the first-level preprocessing data and the multi-level sub-preprocessing data separately, and further complete the distinction and confirmation of the first-level preprocessing data and the multi-level sub-preprocessing data to ensure that the real data of the urine sample can be accurately analyzed to help doctors understand the physical changes of the newborn.

[0024] The time period used for the first-level preprocessing data and the multi-level sub-preprocessing data refers to 1 natural day or 3 natural days, so that the monitoring data can be compared and analyzed by doctors.

[0025] In one example, after the input end of the urine collection device receives the newborn urine, it flows into the detection area through a pipe. The detection area is the port for urine. When the urine flows into the detection area, it is convenient for the data collection module to collect data.

[0026] In an embodiment of the present invention, a urine collection device is installed between the legs of a newborn to collect urine from the newborn around the clock. After the collected urine flows into the detection area through a pipe, a data collection module for the detection area obtains the primary preprocessing data of the urine in the detection area. The primary preprocessing data is combined with the number of urine times to generate multi-level sub-preprocessing data. Subsequently, the primary preprocessing data or the multi-level sub-preprocessing data is received by the data analysis module, and a neural network model is used to analyze the primary preprocessing data or the multi-level sub-preprocessing data. Finally, a result push module is used to receive the analysis results of the primary preprocessing data or the multi-level sub-preprocessing data, and the analysis results are sent to the monitoring terminal, so as to solve the technical problem that existing equipment cannot guarantee the timely collection of urine and avoid the urine being contaminated by the external environment, resulting in deviations in the monitoring data. Example 2

[0027] Please refer to Figures 2 to 3 The urine collection device includes a collecting head 1, the top of the collecting head 1 is connected to a pipe 2, the bottom of the pipe 2 is connected to a detection area 3, and the detection area 3 is provided with a liquid slide 4 corresponding to a data acquisition module.

[0028] It should be noted that the top of the collection head 1 is designed for use with newborns. It is placed between the newborn's legs, and when the newborn produces urine, it is pumped through the top into the detection area 3. A one-way valve is installed at the top of the pipe 2 to prevent urine from flowing back into the detection area, thereby reducing infection.

[0029] In addition, the collection head 1 is made of a transparent material. During use, the urine collected in the detection area can be viewed at any time. It is also necessary to equip the collection head 1 with an ultraviolet germicidal lamp to sterilize it regularly to avoid infection of the newborn.

[0030] The collection head 1 fits over the newborn's legs, and the top of the head is used to collect urine. After the newborn urinates for the first time or subsequently during the monitoring period, the urine is transferred through a pipe 2 to a detection area 3. This area houses a liquid slide 4. The data acquisition module penetrates the slide 4, rapidly collecting urine data, enabling the subsequent data analysis module to quickly and accurately analyze the data.

[0031] In addition, a silicone fastening belt is provided on the side of the collecting head 1 for stably fitting between the two legs of the newborn and stably and without damage to the newborn's legs.

[0032] Preferably, a recovery area is provided at the bottom of the collecting head 1 , and the recovery area is connected to the detection area 3 through a soft pressure component.

[0033] The recovery area refers to the area where urine needs to be discharged into the detection area 3 in a timely manner after monitoring is completed, and is used to receive the discharged urine.

[0034] In one example, after the soft pressing component is pulled by the recovery area, the urine in the detection area 3 is leaked. After the recovery area receives the urine, the collection head 1 is pulled out through the recovery area and the urine is cleared.

[0035] Preferably, the soft pressing component includes a recovery box 5 located in the recovery area, the top of the recovery box 5 is connected to the detection area 3 through a pull-out tube 6, a sealing element 7 is provided on the top of the pull-out tube 6, the sealing element 7 is connected to the bottom surface of the detection area 3 through a spring 8, and a card block 9 is provided at the bottom of the pull-out tube 6 for detachable connection with the top of the recovery box 5, and the sealing element 7 can be tightly connected to the output port of the detection area 3.

[0036] It should be noted that the sealing element 7 is provided with a sealing rubber ring around it, which seals the bottom surface of the detection area 3 with the sealing rubber ring, making it easier for the detection area 3 to collect urine. When the recovery box 5 is pulled down, it is frictionally connected with the inner wall of the collection head 1. When it is pulled down for the first time, the recovery box is suspended in the collection head 1 by observing the flow of urine in the collection head 1. When the urine is collected, the recovery box 5 is pulled out again, and the recovery box 5 is completely taken out to process the urine inside. The bottom of the pull-out tube 6 is clamped with the top of the recovery box 5. The bottom of the pull-out tube 6 is provided with an elastic block 9. When the recovery box 5 is pulled down with force, the block 9 is separated from the opening of the recovery box 5.

[0037] By pulling down the recovery box 5, the recovery box 5 pulls down the pull-out tube 6, and the pull-out tube 6 separates the sealing element 7 from the bottom outlet of the detection area 3, so that the urine flows out of the detection area 3 through the outlet, and then flows into the recovery box 5 through the pull-out tube 6 through the action of the fluid. After the urine collection is completed, the recovery box 5 is pulled out again, the urine is taken out and proceeds to the next step of processing. Example 3

[0038] Please refer to Figure 4 The data acquisition module includes: an optical acquisition unit, used to collect primary preprocessing data or multi-level sub-preprocessing data, wherein the primary preprocessing data or multi-level sub-preprocessing data include physical indicators and chemical indicators; a first data processing unit, used to record the timestamp and urine collection times of the primary preprocessing data to generate multi-level sub-preprocessing data; a second data processing unit, used to eliminate outliers and perform standardization on the primary preprocessing data or multi-level sub-preprocessing data.

[0039] The optical collection unit refers to the use of spectral detection principles to collect data on urine, and the data collected includes physical indicators, chemical indicators and biological indicators.

[0040] Physical indicators refer to color, turbidity, and refractive index; chemical indicators refer to urine protein, urine sugar, and white blood cells.

[0041] The timestamp refers to the time point when the urine is collected by the urine collection device during the monitoring period when the sensor produces urine in the newborn.

[0042] The number of urine collections refers to the signal triggered by the timestamp, which records the number of urine collections. The recorded number of times facilitates the generation of multi-level sub-preprocessing data by combining the first-level preprocessing data with the timestamp.

[0043] In one example, after collecting primary preprocessed data or multiple sub-preprocessed data via an optical acquisition unit, physical and chemical indicator data are collected. After the optical acquisition unit completes data acquisition, a first data processing unit records the timestamp and urine collection times for the primary preprocessed data to generate multiple sub-preprocessed data. Following data collection, a second data processing unit removes outliers and performs standardization on the primary preprocessed data or multiple sub-preprocessed data.

[0044] It's important to note that removing outliers from data can reduce noise and improve statistical reliability, while standardization can accelerate algorithm convergence, satisfy the normal distribution assumption, and enhance model robustness and interpretability. Furthermore, this facilitates analysis by the data analysis module. The technical principles for removing outliers and standardization are well-known techniques and will not be elaborated on here.

[0045] Optionally, the specific steps for recording the timestamp and urine collection times of the first-level pre-processed data to generate multi-level sub-pre-processed data include: Step A1, obtaining the first-level pre-processing data of the optical acquisition unit; Step A2: based on the table characteristics of the first-level preprocessing data and the number of sensor receptions; Step A3: identifying the timestamp and urine collection times based on the primary preprocessed data; Step A4: Generate multi-level sub-preprocessing data according to the timestamp and the number of urine collections.

[0046] In one example, urine data within the detection zone is first collected using an optical acquisition unit. The primary preprocessed data is then used to identify its surface features, which are physical indicators. After the sensor detects the number of urine collections, the timestamp and urine collection count are identified based on the primary preprocessed data. The surface features, timestamp, and urine collection count are then combined to generate multi-level sub-preprocessed data.

[0047] It is worth noting that this embodiment also uses information such as the newborn's weight, age, urination frequency, body temperature, feeding amount, etc. to identify table features. The table features can facilitate subsequent model data analysis.

[0048] In this embodiment, the data analysis module includes: a data receiving unit for primary preprocessing data or multi-level sub-preprocessing data; an algorithm analysis unit for performing model analysis on the primary preprocessing data or multi-level sub-preprocessing data through a neural network model.

[0049] After the data acquisition module completes urine data collection, the data receiving unit receives the primary preprocessed data or multi-level sub-preprocessed data, which has been standardized and removed of outliers. The algorithm analysis unit then analyzes the primary preprocessed data or multi-level sub-preprocessed data to obtain specific physiological data of the newborn's urine. The algorithm analysis unit uses a random forest algorithm to analyze the dynamic relationship between urine parameters and predict newborn health risks.

[0050] Optionally, the specific steps of performing model analysis on the first-level preprocessed data or the multi-level sub-preprocessed data by using a neural network model include: Step B1, initialize the random forest and set the core parameters of the first-level preprocessed data or multi-level sub-preprocessed data; Step B2: construct multiple decision trees using the first-level preprocessed data or the multi-level sub-preprocessed data, and recursively split each decision tree by selecting the optimal split feature from multiple random features when splitting; Step B3: Iterate the decision tree multiple times until the stopping condition is met, and generate a leaf node output category label; Step B4: After each decision tree outputs a prediction result, the classification task integrates the final result through the majority voting method, and the regression task integrates the result through the prediction value averaging method.

[0051] The majority voting method refers to summarizing the classification results of all trees and selecting the category with the most votes as the final output of the entire random forest.

[0052] The prediction value averaging method refers to taking the average of the prediction results of all models as the final prediction output.

[0053] When initializing a random forest, core parameters need to be set based on the first-level preprocessed data or multi-level sub-preprocessed data, including the number of trees, the size of the feature subset considered when each decision tree splits, the maximum depth of the tree, the minimum number of samples for node splitting, and the minimum number of samples for leaf nodes, and repeatability is ensured by random seeds; then, multiple sub-datasets are generated through sampling with replacement. When each decision tree splits a node, it selects the optimal splitting feature (such as Gini impurity or minimization of information gain) from the randomly selected feature subset, and recursively constructs the tree structure; the splitting process continues to iterate until the stopping condition is met (such as reaching the maximum depth, insufficient number of node samples, or purity threshold), and finally generates a leaf node and outputs the category label or average value; after all decision trees complete the prediction, the classification task integrates the results by majority voting, and the regression task takes the average of the predicted values of each tree, thereby reducing the variance and improving the generalization ability of the model through integrated learning.

[0054] Furthermore, a Bayesian network is also included, and its specific steps include: Step C1: constructing a Bayesian network based on medical knowledge or statistical methods; Step C2: Explicitly model the causal relationship between metabolites through a directed acyclic graph and calculate the conditional probability of the node as a structured feature; Step C3: merging the structured features with the original first-level preprocessed data or the multi-level sub-preprocessed data to generate a training set; Step C4: input the training set into the random forest for training; Step C5: Random forest processes high-dimensional nonlinear relationships by integrating multiple decision trees and outputs disease risk prediction.

[0055] In one example, a Bayesian network is constructed based on medical knowledge or statistical methods, and the causal relationship between metabolites is explicitly modeled through a directed acyclic graph, and the conditional probability distribution of the nodes is calculated as structured features (such as the dependency strength or probability parameter between metabolites); these structured features are then merged with the original first-level preprocessed data or multi-level sub-preprocessed data to generate a comprehensive training set that integrates causal information and original observations; this training set is input into an initialized random forest model, and high-dimensional, nonlinear and complex interactive relationships are processed by integrating multiple decision trees, and finally the prediction results of neonatal disease risk are output. The classification task determines the final category through the majority voting method, and the regression task outputs a continuous risk score through the predicted value averaging method, thereby combining the advantages of causal reasoning and data-driven random forest models to improve the interpretability and accuracy of the prediction. Example 4

[0056] Please refer to Figure 5 The second aspect of the present invention provides a method for using an intelligent urine monitoring system for collecting urine from a newborn, comprising the following steps: Step 101: collecting urine produced by the newborn in each time period in real time and sealing the urine, wherein the urine flows into the detection area through a pipe; Step 102: Obtain primary pre-processing data of urine in the detection area, and generate multi-level sub-pre-processing data by combining the primary pre-processing data with the number of urine collection times; Step 103: receiving the primary preprocessed data or the multi-level sub-preprocessed data, and analyzing the primary preprocessed data or the multi-level sub-preprocessed data using a neural network model; Step 104: Receive analysis results of the first-level pre-processed data or the multi-level sub-pre-processed data, and send the analysis results to the monitoring terminal.

[0057] In an embodiment of the present invention, a urine collection device is installed between the legs of a newborn to collect urine from the newborn around the clock. After the collected urine flows into the detection area through a pipe, a data collection module for the detection area obtains the primary preprocessing data of the urine in the detection area. The primary preprocessing data is combined with the number of urine times to generate multi-level sub-preprocessing data. Subsequently, the primary preprocessing data or the multi-level sub-preprocessing data is received by the data analysis module, and a neural network model is used to analyze the primary preprocessing data or the multi-level sub-preprocessing data. Finally, a result push module is used to receive the analysis results of the primary preprocessing data or the multi-level sub-preprocessing data, and the analysis results are sent to the monitoring terminal, so as to solve the technical problem that existing equipment cannot guarantee the timely collection of urine and avoid the urine being contaminated by the external environment, resulting in deviations in the monitoring data.

[0058] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An intelligent urine monitoring system for collecting urine from newborns, characterized in that: include: A urine collection device is used to collect urine produced by the newborn in real time at various time periods and seal the urine, wherein the urine flows into the detection area through a pipe; a data acquisition module, configured to obtain primary pre-processing data of the urine in the detection area, wherein the primary pre-processing data is combined with the number of urine collection times to generate multi-level sub-pre-processing data; a data analysis module, configured to receive the primary preprocessed data or the multi-level sub-preprocessed data, and analyze the primary preprocessed data or the multi-level sub-preprocessed data using a neural network model; The result push module is used to receive the analysis results of the first-level pre-processed data or the multi-level sub-pre-processed data, and send the analysis results to the monitoring terminal.

2. The intelligent urine monitoring system for neonatal urine collection according to claim 1, characterized in that: The urine collection device includes a collecting head, the top of the collecting head is connected to the pipeline, the bottom of the pipeline is connected to the detection area, and the detection area is provided with a liquid slide corresponding to the data acquisition module.

3. The intelligent urine monitoring system for neonatal urine collection according to claim 2, characterized in that: A recovery area is provided at the bottom inner of the collecting head, and the recovery area is connected to the detection area through a soft pressure component.

4. The intelligent urine monitoring system for neonatal urine collection according to claim 3, characterized in that: The soft pressure assembly includes a recovery box located in the recovery area. The top of the recovery box is connected to the detection area through a pull-out tube. A sealing element is provided on the top of the pull-out tube. A block is provided at the bottom of the pull-out tube to be detachably connected to the top of the recovery box. The sealing element is connected to the bottom surface of the detection area through a spring. The sealing element can be tightly connected to the output port of the detection area.

5. The intelligent urine monitoring system for neonatal urine collection according to claim 1, characterized in that: The data acquisition module includes: an optical acquisition unit, configured to acquire the primary preprocessing data or the multi-level sub-preprocessing data, wherein the primary preprocessing data or the multi-level sub-preprocessing data include physical indicators and chemical indicators; a first data processing unit for recording a timestamp and a urine collection frequency of the primary preprocessed data to generate the multi-level sub-preprocessed data; The second data processing unit is used to eliminate abnormal values and perform standardization on the first-level preprocessed data or the multi-level sub-preprocessed data.

6. The intelligent urine monitoring system for neonatal urine collection according to claim 5, characterized in that: The specific steps for recording the timestamp and urine collection times of the first-level pre-processing data to generate the multi-level sub-pre-processing data include: Acquiring the first-level preprocessing data of the optical acquisition unit; According to the table characteristics of the first-level preprocessing data and the number of times the sensor receives the data; identifying a timestamp and urine collection times based on the primary pre-processed data; The multi-level sub-preprocessing data is generated according to the timestamp and the number of urine collection times.

7. The intelligent urine monitoring system for neonatal urine collection according to claim 1, characterized in that: The data analysis module includes: A data receiving unit, used for the first-level pre-processed data or the multi-level sub-pre-processed data; An algorithm analysis unit is used to perform model analysis on the first-level preprocessed data or the multi-level sub-preprocessed data through the neural network model.

8. The intelligent urine monitoring system for neonatal urine collection according to claim 7, characterized in that: Involving the random forest algorithm, the specific steps of performing model analysis on the first-level preprocessed data or the multi-level sub-preprocessed data by using the neural network model include: Initializing a random forest and setting core parameters of the first-level preprocessed data or the multi-level sub-preprocessed data; Constructing multiple decision trees using the first-level preprocessed data or the multi-level sub-preprocessed data, and selecting an optimal split feature from multiple random features for recursive splitting when each decision tree splits; By iterating the decision tree splitting multiple times until a stopping condition is met, a leaf node output category label is generated; After each decision tree outputs a prediction result, the classification task integrates the final result through the majority voting method, and the regression task integrates the result through the prediction value averaging method.

9. The intelligent urine monitoring system for neonatal urine collection according to claim 8, characterized in that: Also includes Bayesian networks: Construct Bayesian networks based on medical knowledge or statistical methods; The causal relationships between metabolites are explicitly modeled through a directed acyclic graph, and the conditional probabilities of the nodes are calculated as structured features. Merging the structured features with the original first-level preprocessed data or the multi-level sub-preprocessed data to generate a training set; Inputting the training set into the random forest for training; The random forest processes high-dimensional nonlinear relationships by integrating multiple decision trees and outputs disease risk predictions.

10. A method for using the intelligent urine monitoring system for neonatal urine collection according to any one of claims 1 to 9, characterized in that: The following steps are involved: collecting urine produced by the newborn in various time periods in real time and sealing the urine, wherein the urine flows into the detection area through a pipe; Acquiring primary preprocessing data of the urine in the detection area, wherein the primary preprocessing data is combined with the number of urine collection times to generate multi-level sub-preprocessing data; Receiving the primary preprocessed data or the multi-level sub-preprocessed data, and analyzing the primary preprocessed data or the multi-level sub-preprocessed data using a neural network model; Receive the analysis results of the first-level preprocessing data or the multi-level sub-preprocessing data, and send the analysis results to the monitoring terminal.

Images