Artificial intelligence (AI) in the initiation of CPAP treatment

Obstructive sleep apnea (OSA) is highly prevalent. Elderly patients may have special difficulty initiating CPAP. This results in poor therapy adaptation and an overload on the healthcare system. AI, in constant development, emerges as a tool to optimize healthcare circuits. <bold>Objective:</bold> To compare the efficency of a natural language processing-based virtual assistant (TUCUVI) to in-person follow-up in the initiation of CPAP for patients aged >=65. <bold>Methods:</bold> Randomized non-inferiority study of OSA patients aged >=65 eligible for CPAP treatment, comparing in person vs telephonic follow-up with TUCUVI over 6 months (Image 1). Data on OSA diagnosis, symptoms, CPAP adherence, adverse effects and number of in-person visits were collected. <fig><object-id>erj;64/suppl_68/PA4364/F1</object-id><object-id>F1</object-id><object-id>F1</object-id><graphic></graphic></fig> <bold>Results:</bold> 51 patients were included, 26 followed with TUCUVI and 25 in person, 56.9% male, with a mean age of 73.2 (SD 5.9) and an AHI at diagnosis of 46.2/h (SD 16.3), no differences between groups. Table 1 shows results from 30 patients at 3-month follow-up. <fig><object-id>erj;64/suppl_68/PA4364/F2</object-id><object-id>F2</object-id><object-id>F2</object-id><graphic></graphic></fig> <bold>Conclusion:</bold> TUCUVI allows a correct adaptation to CPAP similar to in-person control, without significant issues nor additional in-person visits required, based on a partial analysis at 3 months.