A PILOT STUDY OF WEARABLE DEVICES TO DETECT SEPSIS

J YO1, C KARSCHIMKUS2, S CHRISTOV3, F VOLPATO4, P CHAMPION DE CRESPIGNY5, SG HOLT6
1Melbourne Health, Parkville, Australia, 2Melbourne Health, Parkville, Australia, 3Melbourne Health, Parkville, Australia, 4Freelance Signal Analyst, Melbourne, Australia, 5Melbourne Health, Parkville, Australia, 6Melbourne Health, Parkville, Australia

Aim: Health monitoring devices are popular, accepted and purchased by many patients. We assessed the feasibility of a watch device to detect infection.
Background: Infection is an important cause of morbidity and mortality among patients with kidney disease, and early identification and treatment reduces mortality. Diagnosis remains a challenge, but fever is a relatively reliable sign.
Methods: We conducted a single-center prospective cross-sectional pilot study of in-patients to demonstrate proof of concept. Patients randomly admitted under the nephrology unit between August 2017 and April 2018 were consented to wear a monitoring device. Participants wore a clinical grade device that measured peripheral temperature at the wrist (Empatica E4) along with other physiological variables, and we report early analysis of temperature (t) alone. We classified patients as having sepsis, no sepsis and/or treated sepsis and examined how well t correlated with the presence of sepsis.
Results: 104 patients underwent data recording and 82 patients had complete data. 58 patients had no sepsis,15 patients had sepsis and 9 patients had treated sepsis. None of the 41 patients who had at least 84% of readings <35°C had active sepsis. Of the 41 patients with greater than 16% of readings ≥35°C, 15 had sepsis (PPV=21%, NPV=100%). 9/10 patients who had >0.5% of readings ≥37°C had sepsis (PPV=52%, NPV=92%) (1 false positive was admitted with limb ischaemia after occlusion of a vascular graft).
Conclusion: Peripheral temperature measurement may add value to sepsis detection. Further data analysis is underway to refine the detection algorithm. We hope to be able to incorporate such algorithms into a consumer device that can continuously monitor and signal the user of impending infection.


Biography:
Dr. Jennifer Yo is currently a final year Advanced Nephrology Trainee at Melbourne Health.

Categories