The dataset is a repository of cleaned data from a long-term, personalized screening and annotation campaign at a nursing home in Rosarito, Mexico.
General Description and Participants Subjects: The data corresponds to 45 institutionalized older adults.
Duration: The total collection period spans 39 months.
Collection Campaigns: It was divided into two phases: the first lasted 6 months with 15 participants (19,715 records), and the second lasted 33 months with 40 participants (184,806 records).
Types of Data Included The dataset combines information from Internet of Things (IoT) infrastructure and manual annotations by staff.
Specific data types include: Location: Resident records in places such as the bedroom, bathroom, dining room, or garden.
Posture: Information on whether the resident is walking, sitting, standing, or has fallen.
Medical Data (Vital Signs): Measurements of temperature, blood pressure, pulse, glucose, weight, and oxygen levels.
Panic and Pain: Records triggered by panic buttons or pain reports.
Caregiver Notes: Unstructured information about mood (e.g., "calm," "happy") and details of activities such as family visits or meals.
Processing and Standardization To ensure the dataset's usefulness to the scientific community, the data underwent a rigorous pre-processing process that included:
Cleaning and Duplicate Removal: Correction of grammatical errors and removal of repeated records.
Anonymization: The identities of residents, caregivers, and physicians were protected.
Translation: The original records (mostly in Spanish) were translated into English.
Semantization: The data was materialized in OWL instances to populate a knowledge base.
Each record follows a standardized structure that includes ID, timestamp, anonymous identifiers, location, posture, activity type, and detailed description.