Calibrate your own performance equation — calibrate

Fits the relationship between a threshold and race performance on your own athletes, for any threshold, sport and test protocol, so that predict_performance() can use it for new tests. This is the most valid way to predict performance from a step test: the published equations only apply to tests run like those of their studies, while yours applies to your protocol, your analyzer and your athletes.

Usage

calibrate_performance(
  data,
  threshold,
  performance,
  method,
  sport = "running",
  distance = NULL,
  event = NULL,
  protocol = NULL
)

# S3 method for class 'lactate_calibration'
autoplot(object, ...)

Arguments

data: A data frame with one row per athlete.
threshold: The column of data with the threshold, in the intensity unit of the sport.
performance: The column of data with the race time (s) when distance is given, or the mean race intensity otherwise.
method: The lactater method the thresholds come from (e.g. "LTP2"), read from new tests in predict_performance().
sport: "cycling", "running", "rowing" or "swimming", as in lactate_threshold().
distance: The race distance in meters, when performance is a time. Not for cycling.
event: A name for the race. Defaults to the distance (e.g. "10 km"), or "Performance".
protocol: A short description of your test protocol and population, printed with every prediction.
object: The result of calibrate_performance().
...: Not used.

Value

a lactate_calibration tibble with one row: the equation (intercept, slope), its residual SD (sigma), n, the correlation r, the leave-one-out prediction error (loo_error), and what predict_performance() needs. Plot it with autoplot().

Details

Give one row per athlete: the threshold from lactate_threshold() and the result of a race run soon after the test (e.g. within 2 months, as in Roecker et al., 1998). With distance, performance is the race time in seconds and is converted to the mean race intensity of the sport (speed in km/h for running, m/s for swimming, and power in W for rowing, with the Concept2 relation watts = 2.80 / pace^3, pace in s/m); without it, performance is already a mean intensity (e.g. time-trial power in W). The equation is a straight line, intensity = intercept + slope * threshold, as in the published equations. Use the same protocol for the new tests as for the athletes it was fitted on, and more athletes make a narrower, more stable range.

Examples

## thresholds and 10 km times of 12 runners from your lab (simulated here)
set.seed(1)
lab <- data.frame(ltp2 = round(rnorm(12, 15, 1.5), 1))
lab$time_10km <- round(10000 / ((0.5 + 1.02 * lab$ltp2 + rnorm(12, 0, 0.4)) / 3.6))
own <- calibrate_performance(
  lab, "ltp2", "time_10km", method = "LTP2", sport = "running", distance = 10000,
  protocol = "3-min stages, +1 km/h, 1% grade, earlobe"
)
own
#> # Performance equation: 10 km (running): intensity = 1.4 + 0.962 x LTP2, in
#> #   km/h
#> # Fit:                  n = 12, r = 0.94, leave-one-out prediction error 0.481
#> #   km/h
#> # Protocol:             3-min stages, +1 km/h, 1% grade, earlobe
#>   event sport   distance method intercept slope sigma     n  mean    sd   low
#>   <chr> <chr>      <dbl> <chr>      <dbl> <dbl> <dbl> <int> <dbl> <dbl> <dbl>
#> 1 10 km running    10000 LTP2        1.40 0.962 0.465    12  15.4  1.22  13.7
#> # ℹ 10 more variables: high <dbl>, r <dbl>, loo_error <dbl>, step_low <dbl>,
#> #   step_high <dbl>, stage_minutes <dbl>, site <chr>, grade <dbl>,
#> #   source <chr>, protocol <chr>
autoplot(own)