SINEW 15 Biomarker Dataset

The SINEW 15 Biomarker dataset was extracted from the sensor data collected by a longitudinal study called Sensors IN-home for Elder Wellbeing (SINEW).

As the name suggested, the SINEW 15 Biomarker dataset consists of 15 high-level biomarker features, derived from raw sensor readings. They cover four basic aspects of daily living, namely physical, activity, cognitive, and sleep, of each participant over a specific period. The digital biomarkers are of significance, as they can be used for early detection of mild cognitive impairment (MCI), providing an opportunity for timely intervention before it progresses to Alzheimer's disease.

For the purpose of training and evaluating machine-learning-based predictive models for MCI detection, each of the biomarker records is tagged with the cognitive status - either mild cognitive impairment (MCI) or normal cognition (NC) according to the participant's closest clinical assessments within a six month period from the date the sensor data are captured.

Altogether, three data sets are provided in this release, namely

• SINEW 15 Daily Biomarker data set, consisting of records of daily biomarker feature values tagged with the corresponding cognitive status labels;
• SINEW 15 Weekly Biomarker data set, consisting of records of weekly averaged biomarker values tagged with the corresponding cognitive status labels; and
• SINEW 15 Monthly Biomarker data set, consisting of records of monthly averaged biomarker values tagged with the corresponding cognitive status labels.

Each record/row in the three biomarker data sets contains 15 biomarker feature values as inputs and a cognitive class label (MCI vs NC) as outputs.