CN-122004868-A - Automatic identification method for start and stop points in urination stage based on flow curve

CN122004868ACN 122004868 ACN122004868 ACN 122004868ACN-122004868-A

Abstract

The invention discloses an automatic identification method of a urination stage start and stop point based on a flow curve, the method comprises the steps of generating a real-time flow curve by collecting flow data in the urination process, and identifying the urination starting point and the urination end point by combining a preset threshold after preprocessing and feature extraction. The method can be used for combining electromyography signals, abdominal pressure/intravesical pressure data and weight data for auxiliary verification, improving the identification reliability through an abnormality correction step, adapting to subjects with different ages and weights, and adjusting parameters according to urination behavior characteristics of different subjects. The method can realize the automation and high-precision identification of the start and stop points in the urination stage, provide data support for diagnosis, health monitoring and the like of the lower urinary tract function, and has simple operation and wide applicability.

Inventors

Jia Zhuomin
WANG XIYOU
WANG YI
CHEN WEIHAO
JIN YIPENG
ZHANG CHENYANG

Assignees

中国人民解放军总医院第三医学中心

Dates

Publication Date: 20260512
Application Date: 20260227

Claims (10)

1. The automatic identification method for the urination stage starting and stopping points based on the flow curve is characterized by comprising the following steps of: S100, collecting fluid flow data in the urination process, and generating a real-time flow curve; S200, preprocessing the flow curve, removing noise interference and standardizing a data format; S300, extracting characteristic parameters of the flow curve, wherein the characteristic parameters comprise a maximum flow value, a flow change rate, a local extremum point of the curve and a slope of the curve; S400, identifying starting point urination starting time and ending point urination ending time of the urination stage based on a preset threshold value and the characteristic parameters, wherein the starting point is the time when the flow value exceeds the preset starting threshold value for the first time and lasts for the preset time, and the ending point is the time when the flow value falls below the preset ending threshold value and keeps the preset duration.
2. The automatic identification method of the start and stop points of the urination stage based on the flow curve according to claim 1, wherein the collection of the flow data in the step S100 is realized by a flow sensor, a weight sensor or a urine flow rate detector, the collection frequency is not lower than 10Hz, and the data precision is not lower than 0.1mL/S.
3. The automatic identification method of the start and stop points of the urination stage based on the flow curve according to claim 1, wherein the characteristic parameter in the step S300 further comprises a curvature characteristic of the flow curve, the curvature characteristic is obtained by dividing the flow curve into a plurality of curve segments, calculating a curvature value of each curve segment, the curve segments comprise a first curve part, a second curve part and a third curve part, and the curvature direction and the curvature of each curve part are used for distinguishing the flow change characteristic of the urination front segment, the urination middle segment and the urination end segment.
4. The automatic identification method for the start and stop points of the urination stage based on the flow curve according to claim 1, further comprising the step of combining electromyography signal auxiliary identification, wherein the method comprises the step of collecting electric activity signals of urethral sphincter muscles or pelvic muscles in the urination process, and verifying the accuracy of the start and stop points of the flow curve identification when the electric activity signals have a preset change pattern, are weakened before the urination starts and are enhanced after the urination ends.
5. The method of claim 1, further comprising the step of combining pressure data to assist in identification, wherein the pressure data is collected during urination and the intravesical pressure data, wherein detrusor pressure is calculated, wherein the intravesical pressure is subtracted from the pressure, wherein the start of urination is confirmed when the detrusor pressure reaches a preset start pressure threshold, and wherein the end of urination is confirmed when the detrusor pressure falls to a preset end pressure threshold.
6. The automatic identification method of the start and stop points of urination stage based on the flow curve according to claim 1, wherein the preset threshold value in the step S400 is determined by establishing a database based on historical urination flow data, and counting the flow characteristic threshold values of different urination stages according to individual classification including age and weight, or training a threshold value model by a machine learning algorithm to dynamically adapt to urination characteristics of different individuals.
7. The automatic identification method for the start and stop points of the urination stage based on the flow curve according to claim 1 is characterized by further comprising the step of subdividing the urination stage into a front section-flow rising section, a middle section-flow stabilizing section and a tail section-flow falling section based on characteristic parameter changes of the flow curve, wherein the start and stop time of each subdivision stage is respectively identified, and the start point and the end point of each subdivision stage are jointly judged through curve slope abrupt change points, curvature change inflection points and local extremum points.
8. The automatic identification method for the start and stop points of the urination stage based on the flow curve according to claim 1 is characterized by further comprising a weight data auxiliary verification step of collecting weight change data of a bearing medium before and after urination, calculating actual urination amount through weight increment, and verifying that the identification result of the start and stop points is valid when the difference value between the actual urination amount and the urination amount calculated by integrating the flow curve is within a preset error range.
9. The automatic identification method of the start and stop points of the urination stage based on the flow curve according to claim 1, wherein the step S400 further comprises an abnormal start and stop point correction step, when abnormal fluctuation of the flow curve segment corresponding to the identified start and stop point comprises curve interruption and mutation, the combination of the historical contemporaneous flow curve characteristic and the behavior pattern data comprises pre-urination preparation behavior and post-urination covering behavior for correction, and the corrected start and stop point moment is output.
10. The automatic identification method of the start and stop points of the urination stage based on the flow curve according to claim 1, wherein the automatic identification method is characterized in that the preset flow threshold, the weight distribution of the relevant characteristic parameters of the flow curve and the data acquisition frequency can be adjusted in a targeted manner according to urination behavior characteristics including urination posture, single urination duration, urine flow fluctuation range and urination interval rule.

Description

Automatic identification method for start and stop points in urination stage based on flow curve Technical Field The invention relates to the technical field of medical detection, in particular to an automatic identification method of a urination stage start-stop point based on a flow curve. Background The accurate identification of the starting and stopping points in the urination stage is a core basic link of medical scenes such as lower urinary tract function assessment, disease diagnosis, rehabilitation monitoring and clinical curative effect evaluation. In clinical practice, diagnosis and disease classification of lower urinary tract dysfunction such as benign prostatic hyperplasia, neurogenic bladder, urinary incontinence, urinary tract obstruction and the like depend on quantitative analysis of the whole urination process, and accurate definition of the starting and stopping points of the urination stage is the premise of calculating key urodynamic indexes such as urination time, maximum uroflow rate, average uroflow rate, residual uroflow and the like, and directly influences the reliability of the diagnosis result and the rationality of a treatment scheme. The method is widely applied to basic urine flow rate detection equipment due to low equipment cost and simple operation, is manually marked by medical staff through observing a flow curve chart and combining subjective feedback of patients, is suitable for high-precision clinical diagnosis scenes, and is based on an indirect identification method of single auxiliary data, such as auxiliary judgment of pressure change trend of an abdominal pressure sensor or an intravesical pressure sensor, but the method is mostly used as a supplementary means and does not form an independent identification system. However, the prior art has a plurality of limitations to be solved in practical application, and is difficult to meet the requirements of clinical accurate diagnosis and large-scale application, and the existing main flow method is mostly dependent on flow sensor data, is easily influenced by factors such as urine splashing, equipment installation errors, environmental noise, patient position fluctuation and the like in the urination process, so that flow curve has artifacts or abnormal fluctuation, and further causes misjudgment of start and stop points. For example, in a low flow urination scenario, a small fluctuation when the flow rate value is close to the threshold value is likely to be misjudged as the end of urination, and a urine splash when the flow rate is high may be misjudged as an invalid signal before the start of urination. In clinical urodynamic examination, manual marking of start and stop points requires medical staff to have abundant experience, consumes long time, is difficult to adapt to scenes such as large-scale physical examination and long-term rehabilitation monitoring, and meanwhile, different medical staff have differences in judgment standards of flow curve characteristics, are high in subjectivity, are easy to cause inconsistent marking results of the same data, and influence objectivity and comparability of diagnosis. The existing threshold judgment method mostly adopts a uniform fixed threshold, and does not consider physiological differences of different crowds, so that the identification accuracy in special crowds is obviously reduced. For example, elderly patients have a slow rise in initial flow due to impaired detrusor function, a fixed threshold tends to delay identifying the onset of urination, and intermittent urination patterns in neurogenic bladder patients tend to be misinterpreted as multiple episodes of urination. In urodynamic test, except flow data, multi-source data such as abdominal pressure, cystocele, urethral sphincter electromyography signals and the like all contain characteristic information related to the urination stage, but the prior art does not fully integrate the data, only relies on single dimension information, so that the identification result lacks cross verification, and the accuracy is insufficient in complex clinical scenes. In clinic, partial patients have abnormal urination modes, such as intermittent urination, low-flow continuous urination, restarting after urination interruption and the like, and the existing method is mostly based on the design of the ascending-stabilizing-descending curve characteristics of normal urination, so that the abnormal modes are difficult to adapt, the judgment of the occurrence of the missing of a starting point or repeated labeling is easy to occur, and the accurate assessment of the severity of the illness is influenced. In addition, with the popularization of artificial intelligence auxiliary diagnosis technology in the medical field, the clinical demands for automation, precision and scale of urination stage identification are increasingly raised. Due to the limitations, the prior art is difficult to meet the application requirements of emerging