Skip to main content

Session Data

A SessionData object describes a single user's biometric state. It is used both for the current_session and each entry in past_sessions. To explore how to structure the input data, see Input Data.

Biometric time series

These four fields are arrays of numeric samples, representing the biometric data over time. The number of samples depends on your wearable's sampling rate. Please be sure to pass all samples you have for the relevant window, not the averages.

FieldWindowUnit
HR_day_1h1 hour before predictionbpm
RR_day_1h1 hour before predictionbreaths/min
HR_nightFull sleep windowbpm
RR_nightFull sleep windowbreaths/min
tip

Denser time series (e.g., 1-minute resolution) generally yield better predictions than sparse ones (e.g., 10-minute resolution). If your wearable provides high-resolution data, we recommend passing it all.

Sleep architecture

Sleep fields describe the structure of the previous night's sleep. These typically come directly from a wearable or by a sleep-tracking API.

Duration fields

sleepTimeSeconds Total time in the sleep window (including awake time)
deepSleepSeconds Time in slow-wave (N3) sleep
remSleepSeconds Time in REM sleep
lightSleepSeconds Time in light (N1 + N2) sleep
awakeSleepSeconds Time awake during the sleep window

Demographics

FieldTypeNotes
ageintegerYears
height_cmintegerCentimetres
weight_kgfloatKilograms
sexstring or null"M" or "F". Pass null if unknown.

Demographics are used to population-normalize the output scores. Omitting or misreporting them will degrade the quality of normalization.