Abstract: FR-OR43
Wearable Device for Continuous, Non-Invasive Monitoring of Serum Potassium in Hemodialysis Patients
Session Information
- Fluid, Electrolyte, and Acid-Base Disorders Research
November 04, 2022 | Location: W308, Orange County Convention Center‚ West Building
Abstract Time: 04:48 PM - 04:57 PM
Category: Fluid‚ Electrolyte‚ and Acid-Base Disorders
- 1002 Fluid‚ Electrolyte‚ and Acid-Base Disorders: Clinical
Author
- Miller, Forrest, Alio, San Francisco, California, United States
Background
Maintenance of healthy serum potassium levels is critical for the approximately 500,000 patients undergoing dialysis in the United States. Hypo- and hyperkalemia are serious conditions that can result in extreme adverse outcomes including sudden cardiac death. The SmartPatch system (SP) offers a wearable, remote monitoring system for serum potassium.
Methods
This study evaluated the ability of the SP system to identify hemodialysis patients with hypo- or hyperkalemia. Hemodialysis patients (n=96) with a functional arteriovenous fistula were prospectively enrolled in an IRB-approved study. During one dialysis session per month for four months, a wearable SP was applied over the fistula by a nurse or the patient, neither with prior experience placing the device. Venous blood samples were tested before and after dialysis to measure serum potassium (K+) levels. SP data was remotely collected throughout dialysis and analyzed for hematocrit (Hct) and K+. A classification model was trained and tested, using nested cross-validation, to classify data points according to potassium level. Hypokalemia was defined as K+ < 3.5 mEq/L and hyperkalemia was defined as K+ > 5.2 mEq/L.
Results
A combined set of 1229 data recordings were collected and used in this analysis, and all parameter tuning and performance evaluation was performed using nested k-fold cross validation. Reference serum potassium values ranged from 2.5 to 6.4 mEq/L. The median value was 4.2 mEq/L. Reference Hct values ranged from 18 to 48 percentage points. The median value was 34 percentage points. The SP dyskalemia algorithm identified dyskalemia–defined as either hypokalemia (K+ < 3.5 mEq/L) or hyperkalemia (K+ > 5.2 mEq/L)–with a total weighted recall of 86%. The precision–also known as the positive predictive value–of the model was 86%, indicating that the model achieved both high sensitivity and a low rate of false positives.
Conclusion
The results of this study demonstrate a novel proof-of-concept for a noninvasive PPG-based assessment of K+ status in hemodialysis patients using the Alio system. Future studies will include patients who present with hypokalemia at the start of dialysis, as the safety benefit from at-home surveillance is likely greatest for these patients, compared to those who only present with post-dialysis hypokalemia, as was the case in this study.
Funding
- Commercial Support –