Output Format — Docs

The response is the original input JSON with a result object added, plus an updated memory if fitting was performed.

Core result fields

Field	Type	Description
`E`	float	Endurance (0–1). Higher = greater aerobic capacity
`Vmax`	float	Maximum speed (m/s). The absolute top speed this person can achieve
`V5`	float	Internal speed parameter (m/s). Use Vmax for display and interpretation
`P`	float	HRmax (bpm). Model-predicted maximum heart rate
`confidence`	float	Overall confidence (0–1). Product of criteria below
`dataQualityCriterion`	float	Data quality score (0–1)
`fitQualityCriterion`	float	Model fit quality (0–1)
`intensityCriterion`	float	Exercise intensity appropriateness (0.9–1.0)

Bayesian estimates (from memory — use these, not the raw values above):

Field	Type	Description
`trueE`	float	Bayesian-refined endurance
`trueVmax`	float	Bayesian-refined maximum speed (m/s)
`trueV5`	float	Bayesian-refined internal speed parameter
`trueP`	float	Bayesian-refined HRmax
`trueConfidence`	float	Accumulated confidence (0–10)

Conditional fields

Field	Condition	Type	Description
`prediction`	`returnPrediction: true`	float[]	Predicted HR array, one per data sample
`V`	`fixedSpeed: true`	float	V-scale intensity at the test speed
`propOfMax`	`fixedSpeed: true`	float	Recovery start / max tolerable duration
`stats`	`returnStats: true`	object	Full derived metrics (see Derived Metrics)
`errors`	`returnErrors: true`	object	Memory error surface for Bayesian updates

Example response (fit mode)

{
  "result": {
    "E": 0.65,
    "Vmax": 4.71,
    "V5": 4.2,
    "P": 185.0,
    "confidence": 0.82,
    "trueE": 0.64,
    "trueVmax": 4.68,
    "trueV5": 4.18,
    "trueP": 184.0,
    "trueConfidence": 3.5,
    "prediction": [80.0, 82.3, 85.1, ...]
  },
  "memory": {
    "secsSinceEpoch": 1712000000,
    "confidence": 0.82,
    "minKey": "18504265",
    "minValue": 12.5,
    "cutoffValue": 15.0,
    "keys": [18504265, 18504266, ...],
    "errorsX1000": [12500, 13200, ...]
  }
}

Bayesian memory fields

The memory object is returned when fit: true. Pass it back as input for the next session to enable multi-session parameter refinement.

Field	Type	Description
`secsSinceEpoch`	int	Timestamp of this analysis
`confidence`	float	Accumulated confidence (0–10 scale)
`minKey`	string	Best parameter combination key
`minValue`	float	Lowest error value
`cutoffValue`	float	Error cutoff for pruning
`keys`	int[]	Parameter combination keys
`errorsX1000`	int[]	Errors × 1000 (integer-encoded)

See Bayesian Memory for key encoding and decay mechanics.

Speed unit note

Vmax is returned in m/s. To convert:

km/h: multiply by 3.6
mph: multiply by 2.237