Skip to contents

Changelog

Source: NEWS.md

lactater 1.0.0

Every method follows its original paper and is solved exactly; new tools check the analysis, build zones, compare tests, predict performance and report the test. What’s new in lactater 1.0 shows how the numbers compare with 0.3.0, and The methods and the evidence how each method is computed.

Before you upgrade

  • Some thresholds differ from 0.3.0: the methods now follow their original definitions and are solved exactly.
  • The native pipe |> replaces %>%, which is no longer re-exported.
  • run_data_input(), the RStudio gadget, is removed: the web form (recording_sheet()) records a test and read_recording() reads it back; seven packages leave Suggests with it.
  • loglog_restrainer is deprecated (Log-log uses all stages, as Beaver et al., 1985, did); a value below 1 warns.
  • Log-log needs at least five stages and LTP six (four with ltp_breakpoints = 1); a curve with more parameters than points is an error.
  • lactate_threshold() returns a lactate_thresholds tibble with method_category first; intensities are whole watts, km/h to 0.1 and m/s to 0.01.
  • R 4.1 and ggplot2 4.0.0 or later are required; broom, dplyr, forcats, lubridate, magrittr, minpack.lm, pracma, segmented, stringr and tidyr are no longer imported.

Methods

  • Log-log: two lines sharing the division stage, the split with the lowest residual sum of squares (Beaver et al., 1985; Jamnick et al., 2018).
  • Dmax family: the point of the curve furthest below the line; the line’s ends sit where each paper put them (dmax_anchor), NA when the curve never dips below the line; ModDmax counts a rise of exactly 0.4 mmol/L.
  • LTP1 and LTP2: the global least-squares optimum of three connected segments by exhaustive search; the segment before LTP1 can be held flat (ltp_fix_first_slope); one turn point for submaximal tests (ltp_breakpoints = 1).
  • New IAT (Dickhuth et al., 1999) and Stegmann (Stegmann et al., 1981, from recovery samples and stage_duration).
  • New Keul and Simon for running (Keul et al., 1979; Simon et al., 1981): the speed where the polynomial rises by 1.26 or 1 mmol/L per km/h.
  • LTratio is the exact minimum of lactate / intensity on its B-spline; the exponential fit needs no starting values.
  • OBLA, Bsln+ and IAT solve the curve exactly for the target (#8, thanks @dierickxsimon; supersedes #10); with several crossings, the last one where lactate rises through the target.
  • Heart rate at a threshold is interpolated between the stages, NA outside them.
  • Log-log and LTP fit the exercise stages only and report the lactate on their own lines.
  • The resting sample is the row with intensity 0, placed one median step before the first stage when fitted.
  • Own targets: fixed (OBLA), baseline_plus (Bsln+), iat_rise (IAT), moddmax_rise (ModDmax), dmax_methods (adds "LTP1-Poly-ModDmax"), max_intensity (drop stages above it).
  • New last_stage_duration: the intensity of a last stage that was not completed is time-weighted (Kuipers et al., 2003); read_recording() does it from the form’s time in the last stage.
  • New sport "rowing"; running, swimming and rowing get a pace column.
  • Column arguments take a bare name, a string or a variable; heart_rate_column is optional; new rpe_column and rpe_scale.
  • .by analyzes many tests in one call; rows with a missing lactate are dropped with a warning.
  • Deterministic results, about 50x faster than 0.3.0 without plots (500 -> 10 ms for every method on demo_data) and nearly twice as fast with them.

Checks and uncertainty

  • ci = TRUE: wild-bootstrap intervals for the noise of the samples (n_boot, ci_level, seed); failed replicates are counted in a warning. Calibration in “Checking the analysis”.
  • New lactate_checks(): stages, resting sample, thresholds not found or at the edge, early dip, no rise in the last stage, test stopped before LTP2, non-blood sample, and whether the LT1 and LT2 methods agree.
  • New lactate_fit_quality() (thanks @Victorivus, #11) and lactate_diagnostics(): residuals, leverage, Cook’s distance, every threshold without each stage, and a six-panel plot.
  • New sample, stage_duration, site and grade record the protocol; predict_performance(), reference_percentiles() and compare_tests() check it against theirs.

Plots and reports

  • Every method’s plot explains itself: the method as title, how it works as subtitle, the construction drawn on the curve, the threshold labeled with its values. autoplot(x) draws an overview of every threshold on one axis, grouped by what the method aims at (the first rise, the second rise, a fixed lactate), with a table of values; autoplot(x, method = ) one method.
  • autoplot(lactate_zones(x), view = "ruler") draws the zones as one bar over the intensity, pace, heart rate and RPE at each boundary, to hand to the athlete.
  • options(lactater.style = "night") and "journal", options(lactater.font = ), and theme_lactater() for your own plots.
  • New lactate_report() (a gt table) and gt_lactater(); options(lactater.web_font = FALSE) leaves the web font out of the tables.
  • New test_report(): the test as a web page and a two-page PDF for the athlete or the coach; report_start() and the report_*() steps build one part by part; two to five tests of one athlete give a progress report.
  • New recording_sheet(): a web form, a printable sheet and a CSV template; read_recording() reads the form’s .xlsx.

Zones, comparisons, performance

  • lactate_threshold(lt1 = , lt2 = ) chooses the athlete’s LT1 and LT2 once: the print, the overview, the zones, the report and the headers of compare_tests() and predict_performance() follow that pair.
  • compare_tests() sets every change against the noise of its method (what two tests differ by when nothing changed), with their ratio and a verdict of “beyond” or “within the noise”, and shows the cv behind it. The default noise of each method is lactater’s own calculation on the open data of Heuberger et al. (2018), in data-raw/reliability_cycling.R; the help page lists the published values beside it. lactate_reliability() gives the bias between tests, the interval of the cv and its degrees of freedom, and compare_tests() widens its intervals for a cv from few athletes; lactate_trend() takes a level.
  • New lactate_zones(): three or five zones from LT1 and LT2 (Bsln + 0.5 and Log-Poly-ModDmax by default) at the intensity domains, with intensity, pace, lactate, heart rate and RPE; split sets the cuts inside the domains. Heart rate is drawn in a panel under the curve, or on a second axis (heart_rate = "axis").
  • New training_load(): time in each zone and the polarization index (Treff et al., 2019).
  • New compare_tests(): each change against the test-retest noise of the method, with a verdict and notes on what differs between the tests; lactate_reliability() measures that noise on your own athletes; curve_shift(), curve_shape() (shift or new shape, by an F test and AICc), lactate_trend() (series of tests), relative_thresholds() (per kg, % of peak).
  • New reference_percentiles(): percentiles of the power at 3 mmol/L and at baseline + 1 mmol/L among German adults of the same sex and age (Fiedler et al., 2025).
  • New predict_performance() (running from Roecker et al., 1998, and Sjodin & Jacobs, 1981; cycling from the open data of Heuberger et al., 2018) and calibrate_performance() for your own equation.
  • New lactate_mlss() (the maximal lactate steady state from constant-load bouts, Beneke’s 10 to 30 min window) and vlamax() (the maximal lactate accumulation rate from a sprint test).

Data and documentation

  • New lactate_tests: 16 step tests across cycling, running, rowing and swimming, from elite to untrained; twelve real (Trinity College Dublin and Heuberger et al. 2018, CC BY 4.0), four simulated. demo_data gains a simulated RPE.
  • New articles on best practices, the methods and the evidence, checking the analysis, zones, training load, comparing tests, predicting performance, the example data and what is new; every citation links to its paper.

lactater 0.3.0

  • Improved lactate_threshold() performance by reducing repeated model preparation and minimizing expensive data wrangling in method pipelines.
  • Optimized plotting performance by removing ggplot_build() from x-axis adjustment and simplifying plot construction paths.
  • Added stronger input/model error handling for edge cases in fitting and segmentation routines.
  • Expanded test coverage with validation and golden-output regression tests for numerical stability.
  • Updated CI quality checks and coverage gates to better protect performance and correctness changes.

lactater 0.2.0

CRAN release: 2023-11-16

  • Added the function lactate_curve() to retrieve data for plotting the lactate curve and heart rate response.

lactater 0.1.4

CRAN release: 2023-04-16

  • Patched a modification to the Dmax method, which was producing negative values in a few edge cases.

  • Make sure that infinite values are not retrieved when converting intensity to log.

lactater 0.1.3

  • Patched a fix to the Dmax method that was producing very high estimations in a few cases.

lactater 0.1.2

CRAN release: 2023-02-22

  • Fixed missing log in intensity in the Log-log method.

lactater 0.1.1

CRAN release: 2022-09-08

  • Removed align attribute in img tag.

lactater 0.1.0

CRAN release: 2021-12-03

  • Added a NEWS.md file to track changes to the package.