AECOPD prediction through a remote monitoring system

<bold>Rationale:</bold> Early identification of Acute Exacerbation is a crucial issue in COPD management. Promising results from Remote Patient Monitoring (RPM) suggest it could be a game changer in the field. The aim of the study was to assess performances of a predictive scoring model fed by longitudinal data collected through a RPM system. <bold>Methods:</bold> In a six-month prospective study, COPD patients were monitored using Bora band^® wristband (Biosency^TM, France). This device remotely collected SpO₂, Heart Rate (HR), and Breath Rate (BR), while Acute Exacerbation of COPD (AECOPD) identification and severity characterization were done at 3- and 6-month follow-up visits. Each vital sign time serie was stochastically prefiltered, Z-scored and aggregated into a single score called BVS^{3 ®} (BORA Vital Sign Standard Score). <bold>Results:</bold> Over a 6-month monitoring period, we included 158 COPD patients (GOLD grade: 16% I, 48% II, 22% III, 14% IV) with a mean age of 63.3 years (SD 8.1 years) with a M/F gender ratio of 1.5. Patients wore wristbands 66% to 99% of the time (mean 87%, SD 9%). We collected an average of 31.9 SpO₂, 58.8 HR, and 39.9 BR measures per day. A total of 29 AECOPD (25 moderate, 4 severe) were recorded. BVS³ score predicted exacerbations 4.4 days before onset validated by clinicians (SD 4.3), exhibiting sensitivity (86.2%) and specificity (84.6%), with an AUC of 0.93. <bold>Conclusion:</bold> Patients had an excellent acceptance of Bora band^® leading to collection of more than 130 vital signs daily. The specificity, sensitivity, and anticipation of our BVS³ patient specific AECOPD early detection scoring is very promising for enhancing COPD management.