SALUDATA Synthetic Remote Monitoring Dataset

Synthetic dataset generated within the SALUDATA project ("Plataforma para el seguimiento integral de pacientes con un sistema de alertas predictivo"), which developed a remote patient monitoring platform with a predictive alert system. It was generated to develop, train and test the platform's alert models and data pipelines without using real patient data. Content 20,000 fictional adult profiles (ages 25–85) 3,680,000 twice-daily multisensor observations (21 variables) over 92 days 120,000 assessment episodes generated from 43 scenario recipes in 10 families Sensor coverage of the SALUDATA platform (ring, in-ear thermometer, blood pressure monitor, smart scale), each field labelled as calibrated or simulated Simulated timestamps from March to May 2026 (not real collection dates) Files: four ZIP archives (common + one per month) that together reconstruct the dataset, plus standalone documentation. See README.md. Calibration: population distributions were calibrated on 923 public adult records from the MIMIC-III Waveform Database (PhysioNet). No source traces are redistributed. Privacy: all profiles are fictional and generated from pooled population statistics of public, de-identified records. No individual source traces are copied or resampled. No formal privacy guarantee (e.g. differential privacy) is claimed. Status: research prototype, not clinically validated. Scenario references were produced with the assistance of large language models and a rule-based simulator, without clinician review. Funding: Ministerio para la Transformación Digital y de la Función Pública (Spain), grant TSI-100130-2024-15.