A wearable sensor featuring a personalized machine learning platform can accurately predict heart failure (HF) rehospitalization in patients with a history of HF, according to findings from the multicenter LINK-HF study (Multisensor Non-invasive Remote Monitoring for Prediction of Heart Failure Exacerbation) published in Circulation: Heart Failure.
Study participants were adults with a history of HF, New York Heart Association functional class 2 to 4, and were hospitalized for acute HR exacerbation. Additionally, participants had HF with reduced ejection fraction (left ventricular ejection fraction <50%) and HF with preserved ejection fraction (left ventricular ejection fraction ≥50%). During the study, the researchers fitted patients with a wearable sensor, which was placed on the chest.
The sensor collected continuous ECG waveform, continuous 3-axis accelerometry, skin impedance, skin temperature, and activity and posture information. This information provided investigators with data on heart rate and heart rate variability, arrhythmia burden, respiratory rate, gross activity, walking, sleep, body tilt, and body posture. The sensor was connected with an Android phone enabled with Bluetooth. These data were streamed continuously from the sensors to the phone.
Continue Reading
Compliance with sensor use was high, as reflected by the 87 out of 100 participants who completed 30 days of monitoring. Additionally, 74 of the 87 participants completed 90 days of monitoring. Among the 100 participants included in the study (mean age, 68.4±10.2 years), a total of 35 unplanned nontrauma hospitalization events occurred. Of these unplanned events, a total of 24 worsening HF events were observed.
The wearable sensor detected precursors of hospitalization for HF exacerbation with a corresponding 76% to 88% sensitivity rate and an 85% specificity rate. The median time between the initial alert and hospital readmission was 6.5 (range, 4.2–13.7) days. Significant divergences were observed in time-to-HF and time-to unplanned nontrauma hospitalization between participants with an alert (P =.001) and without an alert (P =.008).
Limitations of the study included the lack of formal testing and validation sets, the observational design, and the predominantly male (98%) population.
Due to these limitations, the investigators of this study concluded that the “clinical efficacy and generalizability of this low-cost noninvasive approach to rehospitalization mitigation should be tested” in additional research.
Reference
Stehlik J, Schmalfuss C, Bozkurt B, et al. Continuous wearable monitoring analytics predict heart failure hospitalization: The LINK-HF multicenter study. Circ Heart Fail. 2020;13(3):e006513.
