Personalized Urination Activity Management Based on an Intelligent System Using a Wearable Device

Area of Science:

• Biomedical Engineering
• Urology
• Wearable Technology

Background:

• Urinary pattern analysis is crucial for diagnosing and managing urological conditions.
• Current monitoring methods can be invasive or inconvenient for patients.
• Smart wearable devices offer a non-invasive approach to physiological data collection.

Purpose of the Study:

• To develop and evaluate a urinary management system using smart bands for collecting and analyzing urinary time and interval data.
• To provide web-based visualization for monitoring and feedback to urological patients.
• To assess the system's accuracy and clinical applicability.

Main Methods:

• Designed a device recognizing urination via patient-specific posture changes.
• Utilized sequential data analysis with recurrent neural networks (RNN) and long short-term memory (LSTM).
• Implemented a web service (HTML5) for visual data support and clinical diagnostic assistance.

Main Results:

• The system achieved a high average accuracy of 95.8% in recognizing urinary activity.
• Evaluation conducted on 30 men without prior urination issues over 3 days.
• Algorithm soundness demonstrated based on urological clinical guidelines.

Conclusions:

• The urinary activity management system demonstrates high accuracy and clinical applicability.
• Wearable devices and pattern-recognition algorithms offer a new methodology for studying physiological behaviors.
• This system has the potential to significantly impact diagnostic assistance for clinicians.