Abstract Introduction Wearable devices allow continuous monitoring of physiological sleep variables, but these data remain largely isolated from clinical systems due to the lack of standardized interoperability. This limits diagnostic value, longitudinal follow-up, and their use in clinical decision support. This study proposes and evaluates a model for converting wearable-derived sleep biomarkers into structured HL7 FHIR resources within an electronic health record. Methods A middleware system was developed to receive raw biometric data in JSON format from a wrist-worn wearable device (Samsung Watch). The pipeline included preprocessing of heart rate, oxygen saturation, heart rate variability, movement, total sleep time, estimated sleep stages, habitual bedtime and wake time, sleep onset latency, and total sleep time with oxygen saturation below 90 percent. Semantic mapping was performed using LOINC and SNOMED CT, followed by syntactic conversion into HL7 FHIR resources such as Observation, Device, Patient, and SleepStudy. Validation occurred in a FHIR server connected to an electronic health record. Adults undergoing outpatient sleep evaluation completed seven days of monitoring. Outcomes included data ingestion completeness, FHIR conformity, and feasibility of clinical visualization. Results Preliminary tests demonstrated successful ingestion of most transmitted data, with an average processing latency of a few seconds per dataset. All observations passed structural FHIR validation, and terminology mapping was consistent for core sleep variables. Key parameters, including total sleep time, nocturnal heart rate, oxygen saturation metrics, sleep onset latency, habitual sleep timing, and total sleep time below 90 percent saturation, were displayed in structured longitudinal format within the electronic record. No conflicts or security issues were detected. Clinicians reported clear interpretability of the integrated data. Conclusion The model proved feasible for converting wearable-derived sleep biomarkers into structured HL7 FHIR data suitable for clinical workflows. This approach may support remote monitoring, enhance diagnostic evaluation, and provide a foundation for future predictive algorithms using standardized longitudinal sleep data. Continued development will expand the dataset, integrate additional devices, and assess clinical impact. Support (if any) This project received support from Fundação Maria Emília, Escola Bahiana de Medicina e Saúde Pública, Rede Nacional de Ensino e Pesquisa (RNP), and SENAI CIMATEC - Salvador.
Viana et al. (Fri,) studied this question.