Input Format

The request body is a JSON object with three fields: settings, data, and memory. All are stringified JSON strings.

Settings

Field	Type	Default	Description
`fixedSpeed`	boolean	required	`true` for treadmill/fixed-pace tests, `false` for variable-speed activities
`speed`	float	—	Speed in m/s (required if `fixedSpeed: true`)
`fit`	boolean	`false`	Whether to fit E, Vmax, P from the data
`returnPrediction`	boolean	`false`	Include predicted HR array in result
`returnStats`	boolean	`false`	Include derived metrics (energy, recovery, zones, etc.)
`returnErrors`	boolean	`false`	Include memory error surface

Parameter constraints (when `fit: true`)

Field	Type	Default	Description
`Econstraints`	[min, max, step]	[0.01, 0.99, 0.01]	Endurance search range
`V5constraints`	[min, max, step]	[3.0, 12.0, 0.07]	Internal speed parameter range (m/s). Affects Vmax output.
`Pconstraints`	[min, max, step]	[140, 220, 1.0]	HRmax search range (bpm)

Fixed parameters (when `fit: false`)

Field	Type	Default	Description
`E`	float	0.5	Endurance (0–1)
`Vmax`	float	6.2	Maximum speed (m/s)
`P`	float	180	HRmax (bpm)

Demographics (for derived stats)

Required when returnStats: true.

Field	Type	Default	Description
`weight`	float	75	Body mass (kg)
`height`	float	175	Height (cm)
`age`	int	30	Age (years)
`gender`	string	"male"	"male" or "female"

Sport mode

Field	Type	Default	Description
`sport`	string	"running"	"running", "cycling", or "team_sports"
`squadMeanE`	float	—	Squad average E (for team archetype classification)
`squadMeanVmax`	float	—	Squad average Vmax in m/s
`measuredRHR`	float	—	Measured resting HR in bpm (for RZI calculation)

Data

An array of timestamped objects. Required fields: time and hr.

Field	Type	Required	Description
`time`	float	Yes	Seconds from start
`hr`	float	Yes	Heart rate (bpm)
`speed`	float	Required for variable-speed	Speed (m/s). Required if `fixedSpeed: false`.

Example

[
  {"time": 0, "hr": 80, "speed": 2.5},
  {"time": 1, "hr": 82, "speed": 2.5},
  {"time": 6, "hr": 91, "speed": 2.6}
]

Data requirements

Minimum: ~60 samples for basic fitting, 300+ for reliable results
Sample rate: 1 Hz down to once every 6 seconds. Timestamps don't need to be consecutive.
Session length: no hard limit — marathons, multi-hour rides, full matches are all supported
Speed: in m/s. For cycling, power/100 can be used as a speed proxy.

Memory

The memory field contains the Bayesian prior from previous sessions. Always pass memory between sessions for the same individual — parameter estimates improve significantly with each session, and without prior data the SDK returns single-session fits which may not have fully converged.

For the first session with a new user, omit the memory field. The SDK will initialize a new memory and return it in the response. Store it and pass it back for every subsequent session.

{
  "settings": "...",
  "data": "...",
  "memory": "{\"secsSinceEpoch\":1712000000,\"confidence\":3.5,\"minKey\":\"18504265\",\"keys\":[...],\"errorsX1000\":[...]}"
}

For the first session (no prior data), omit the memory field. The SDK will initialize a new memory and return it in the response.

See Bayesian Memory for details on the memory format, key encoding, and confidence decay.