Work Description

Date: 05 May, 2025 Dataset Title: Improving the Accuracy of the Wearable Uroflowmeter for Incontinence Monitoring under Dynamic Conditions: Leveraging Machine Learning Methods Dataset Contact: Faezeh Shanehsazzadeh [email protected] Dataset Creators: Name: Faezeh Shanehsazzadeh Email: [email protected] Institution: University of Michigan Department of Mechanical Engineering, Biomechanics Research Laboratory and Pelvic Floor Research Group. ORCID: https://orcid.org/0000-0001-5783-2674 Name: John DeLancey Email: [email protected] Institution: University of Michigan Department of Department of Obstetrics & Gynecology, Pelvic Floor Research Group ORCID: https://orcid.org/0000-0002-1012-7713 Name: James Ashton-Miller Email: [email protected] Institution: University of Michigan Department of Mechanical Engineering, Biomechanics Research Laboratory and Pelvic Floor Research Group. ORCID: https://orcid.org/0000-0003-1528-2787 Funding: This research was funded by the U.S. Public Health Service grants 1 RC2 DK122379-01, P30 AG024824, and Procter & Gamble (through an institutional research contract). The funders had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results. Key Points: - The integration of an optimized 8-vane Etoile flow conditioner with a 0.2D opening significantly improves the stability and accuracy of flow measurements in a wearable uroflowmeter. - Computational fluid dynamics simulations and experimental validations demonstrated a 40% reduction in turbulence intensity and a 67% enhancement in axial velocity profile stability. - Machine learning models, including SVM and XGBoost-PCA, achieved high correlations with true flow rate predictions, supporting robust performance under dynamic conditions. - The refined device effectively addresses the challenges of measuring flow dynamics during daily activities, providing a reliable tool for monitoring urinary incontinence in women in real-world settings. Research Overview: Urinary incontinence affects many women, yet there are no monitoring devices capable of accurately capturing flow dynamics during everyday activities. Building on our initial development of a wearable personal uroflowmeter, this study enhances the device's performance under realistic, dynamic conditions similar to those encountered in daily living. We integrated an optimized 8-vane Etoile flow conditioner with a 0.2D opening into the device. Both computational fluid dynamics simulations and experimental tests demonstrated that this flow conditioner significantly reduced turbulence intensity by 82% and stabilized the axial velocity profile by 67%, increasing the R� of flow rate measurements from 0.44 to 0.92. Furthermore, our machine learning frame-work�utilizing Support Vector Machines (SVM) and Extreme Gradient Boosting (XGBoost) models with Principal Component Analysis (PCA)�accurately predicted the true flow rate with high correlations and robust performance with minimal overfitting. For the test dataset, the SVM achieved a correlation of 0.86, an R� of 0.74, and an MAE of 2.8, whereas the XGBoost-PCA model exhibited slightly stronger performance, with a correlation of 0.88, an R� of 0.76, and an MAE of 2.6. These advances established a solid foundation for developing a reliable, wearable uroflowmeter capable of effectively monitoring urinary incontinence in real-world settings. Methodology: This dataset encompasses both simulation and experimental results. The simulation data originates from computational fluid dynamics (CFD) simulations conducted using COMSOL software to identify an optimized flow conditioner structure designed to reduce turbulence within the measurement channel. Several key metrics were considered during analysis, including normalized turbulence intensity along the sensor for various flow conditioner designs, time-averaged axial velocity profiles at the sensor location, and the standard deviation of the axial velocity across spatial locations. These analyses were instrumental in selecting an optimal blade configuration, which was subsequently validated through benchtop experiments. The experimental data were gathered using the PUF device, which records essential sensor outputs, specifically the flow rate sensor output (FSO) and the flow rate�temperature differential (FTD). True flow rates (TFR) were meticulously controlled using a syringe pump, operated via custom software developed in our laboratory using National Instruments LabVIEW 2021. Additionally, the fluid temperature was measured accurately with a Keysight thermostat, ensuring the precision and reliability of data collection. Instrument and/or Software specifications: Simulation: COMSOL Multiphysics� version 5.6 Control Software: NI LabView 2021 Code Development: Visual Studio 1.97.2 with Python 3.12.4 Python Libraries Used: Matplotlib: 3.9.1, Pandas: 2.2.2, NumPy: 1.26.4, Seaborn: 0.13.2, SciPy: 1.14.1, Scikit-learn (sklearn): 1.5.1, AutoGluon.tabular: 1.1.2b20241120, XGBoost: 2.1.1 Files contained here: 1) COMSOL Folder: - Intensity subfolder: o �Intensity.py�: The Python code used to plot the results of the simulation. o Eight *.txt files exported from COMSOL software represent the turbulence intensity along the sensor of the eight structures compared in the manuscript. Definition of Terms and Variables: -X: Spatial data at the sensor location (mm). -Y: Turbulence intensity along the sensor. - Velocity subfolder: o �Velocity.py�: The Python code used to plot the results of the simulation. o Eight *.txt files exported from COMSOL software represent the time-averaged axial velocity profiles at the sensor location of the eight structures compared in the manuscript. o Eight *_time.txt files exported from COMSOL software represent the standard deviation of the axial velocity across spatial locations of the eight structures compared in the manuscript. Definition of Terms and Variables: *.txt files: -X: Spatial data at the sensor location (mm). -Y: Time-averaged axial velocity (m2/s). *_time.txt files: - Axial velocity at the sensor location during time (m2/s). 2) Laminar Flow - �Compare_Laminar_Turbulence.py�: The Python code used to plot the results of the simulation. - �turbulence.csv�: Experimental results for five measurement cycles while the sensor was located in turbulent flow. - �laminar.cvs�: Experimental results for five measurement cycles while the sensor was located in laminar flow. Definition of Terms and Variables: - x: Fourteen distinct flow rates ranging from 0 to 20?mL/s. - o1-o5: Flow rate sensor outputs (Volts) correspond to five measurement cycles for each flow rate. 3) ML - "data.csv": which includes experimental data. - "Datanalyses.py": the Python code used for data analysis and model implementation. Definition of Terms and Variables: - FSO: Flow Rate Sensor Output (Volt): Represents the voltage output from the flow rate sensor. - FTD: Flow-Temperature Sensor Output Differential (Volt): Indicates the voltage difference from the flow and temp sensors. - TFR: True Flow Rate Applied with Syringe Pump (ml/sec): The actual flow rate administered during experiments, controlled precisely using a syringe pump. - TT: Temperature Measured with Thermostat (Degree Centigrade): The temperature reading obtained from the thermostat. Use and Access: This data set is made available under a Creative Commons Public Domain license (CC0 1.0). Update: This version of the dataset has been refined based on reviewer comments for the article with the same name. Compared to the version uploaded on February 20, 2025, it now includes additional simulation data and the corresponding plotting code added to the "COMSOL" and "Laminar Flow" folders.