SRDA: Mobile Sensing based Fluid Overload Detection for End Stage Kidney Disease Patients using Sensor Relation Dual...

Area of Science:

• Nephrology
• Biomedical Engineering
• Data Science

Background:

• Chronic kidney disease (CKD) is a prevalent, life-threatening condition.
• End-stage kidney disease (ESKD) patients on hemodialysis struggle with fluid overload due to impaired kidney function.
• Current fluid monitoring methods are invasive, discontinuous, and clinic-bound.

Purpose of the Study:

• To develop a non-invasive system for detecting fluid overload in ESKD patients.
• To leverage passively collected bio-behavioral data from smartwatch sensors for fluid intake management.
• To introduce SRDA (Sensor Relation Dual Autoencoder), a latent graph learning model for fluid overload detection.

Main Methods:

• Utilized a latent graph learning approach with a Sensor Relation Dual Autoencoder (SRDA).
• Employed passively collected bio-behavioral data from smartwatch sensors.
• Validated the system using real-world mobile sensing data from ESKD patients.

Main Results:

• SRDA demonstrated superior performance compared to existing state-of-the-art methods.
• The system achieved high F1 scores and recall in fluid overload detection.
• The study confirmed the efficacy of ubiquitous sensing for ESKD fluid management.

Conclusions:

• SRDA presents a novel, effective, and non-invasive solution for monitoring fluid overload in ESKD patients.
• The findings highlight the potential of wearable sensor technology for improving patient self-management and clinical outcomes.
• Ubiquitous sensing offers a viable alternative for continuous and accessible fluid intake management in hemodialysis patients.