← Entering CPET Data · Home · Normative Predictions →

Breath-by-Breath (BxB) Tab

The BxB tab is a dedicated workspace for breath-by-breath CPET data. It hosts the Wasserman 9-Panel Plot and a time-down data table. The tab appears only when the loaded test contains BxB data; otherwise it shows an empty-state message.

BxB data arrives in Systole one of three ways:

• Imported from a Breeze .sum file (BxB section)
• Imported from a Systole .pedal v5.0 file
• Loaded from the MedGraphics SQL Server database during direct import

Wasserman 9-Panel Plot

Nine scatter panels arranged in a 3×3 grid:

Phase markers (Exercise / Recovery) appear as vertical reference lines, derived from the source file's testTiming or auto-detected from the BxB stream when missing. Recovery data is trimmed from the V-slope and VO₂-vs-Work panels for clean regression visualization. The horizontal predicted-peak-VO₂ reference line tracks the equation currently selected on the Reference tab.

The VO₂ vs Work Rate panel is drawn to a fixed scale

Panel 3 is the one panel whose angle is the finding, so it does not auto-fit to the data:

• The axes are locked to a 10:1 span ratio (VO₂ span = 10 × the watts span, both starting at zero), so a ΔVO₂/ΔWR of 10 mL/min/W has the same slope on screen for a 60 W child and a 300 W adult. Two tests can be compared by eye.
• A dashed 10 mL/min/W reference line runs from the test's resting VO₂ (the median of the pre-exercise breaths). A shallow work-to-VO₂ response visibly diverges from it instead of having to be read off the axis numbers.

On a treadmill there is no measured work rate, so the x-axis is the ACSM estimate back-computed from speed and grade — it is labelled "Work (watts, est.)" and the legend chip's tooltip repeats the caveat. Treat the treadmill relationship as noisier, and partly a reflection of that arithmetic rather than of the patient.

Data Quality: Outlier Handling

Breath-by-breath data occasionally contains corrupted breaths — a swallow, cough, sigh, partial breath, or sensor dropout produces a single cycle with physically impossible gas exchange (e.g. VO₂ or VCO₂ at or below zero). Because the ventilatory equivalents are ratios (VE/VO₂ and VE/VCO₂), a near-zero denominator makes the value explode — a real example reaches VE/VCO₂ ≈ 27,000 against normal values of 20–50. Left unhandled, one such breath flattens the VE/VO₂ & VE/VCO₂ panel and skews the averaged trace and the reported VE/VCO₂ slope.

No stored breath is ever changed or deleted — the filtering is applied at view/compute time only.

• Physiologically invalid breaths are excluded (from the plots and from all computed/reported values: Rest/AT/Peak means and the VE/VCO₂ slope). A breath is invalid when VO₂ ≤ 0, VCO₂ ≤ 0, or a ventilatory equivalent exceeds 80 — values only reachable with a corrupted near-zero denominator. This is deliberately conservative: on real noisy tests it removes well under a few percent of breaths, and it keeps high-but-real values (a VE/VCO₂ of 73 is preserved). Because the rule is purely per-breath physiology, it never removes real coherent patterns such as exertional oscillatory ventilation (EOV), whose breaths all have valid gas exchange.
• Excluded breaths are drawn faint, not hidden, on the VE/VO₂ & VE/VCO₂ panel, so you can always see what was removed. The y-axis is scaled to the valid data; if an excluded point is too extreme to fit, a small "↑ N artifacts off-scale" note appears at the top of that panel.
• A data-quality badge (e.g. "⚠ 12 invalid breaths excluded (3.4%)") appears in the toolbar when a test has invalid breaths, so a noisy acquisition is visible rather than silently cleaned.

Developer reference for thresholds and rationale: docs/BXB_ARTIFACT_HANDLING.md.

Editing Workflow (Working Draft)

All edits on the BxB tab are previews held in a working draft until you commit them. The big right-justified Apply Changes to Test button on the title row writes the draft to the test's summary data (Rest / AT / Peak values, VE/VCO₂ slopes, AT timestamp, applied averaging method, and exercise window). Until you click Apply, nothing changes in the test record.

Switching tabs with unapplied changes opens a prompt with three choices: Stay on tab (cancel the switch), Discard & leave (drop the draft and switch), or Apply changes & leave (commit the draft, exactly as the Apply button does, then switch). Clicking Verify with unapplied changes also prompts you to apply or discard.

Averaging Method

Seven buttons under "Averaging Method:" select how breaths are smoothed for chart display and summary computation:

• 10s rolling average
• 15s rolling average
• 20s rolling average
• 30s rolling average (default)
• 8-breath mean
• Mid5/7 — average of middle 5 of 7 breaths
• Raw — no averaging

Button styling shows the state:

• Filled — currently previewing this method
• Green outline — the method already applied to the test
• Clicking a different method previews it; clicking Apply commits.

The site-wide default lives in Settings → Default BxB Averaging Method.

Anaerobic Threshold (AT)

Set or adjust AT on the V-slope panel:

• Drag the AT line on V-slope. The line snaps to the nearest plotted exercise breath. All 9 panels update live to reflect the new AT.
• Set AT button (visible when no AT is present) — places a marker at the exercise midpoint as a starting point.
• Clear AT — removes the AT marker entirely.
• Discard AT Changes — reverts the working draft AT to the saved value.

AT is captured as both a VO₂ value and a timestamp, so the time-down table highlights the AT row by elapsed time rather than by VO₂ — avoiding spurious matches from transiently high early-exercise values.

Exercise Window Markers

Click Adjust Exercise Window to enter draggable mode. Two thick red vertical lines mark exercise start and recovery start; drag either to redefine the exercise phase. Discard Window Changes reverts to the saved markers.

Adjusting the window affects which breaths count as Rest, Exercise, and Recovery for summary computation, AT phase membership, and VE/VCO₂ slope regressions.

Time-Down Data Table

Below the 9-panel plot, a tabular view of the BxB stream resampled to fixed intervals (10 s for 20s/30s rolling, 5 s for 10s/15s rolling). Features:

• Sticky column headers as you scroll
• Phase separator rows ("START EXERCISE", "START RECOVERY") with timestamps
• AT (purple) and Peak (red) row labels
• Rest rows highlighted blue
• Cycle / treadmill column auto-detection (work rate vs speed+grade)

Important: The time-down table values are interpolated for readability and may differ slightly from the Exercise tab summary, which is computed from the dense per-breath rolling average. Peak VO₂ and AT in the 9-panel plot and Exercise tab summary are always exactly consistent.

Download CSV

Download CSV (top right of the Breath-by-Breath Data heading) exports exactly the rows shown in the time-down table — same resampling, same columns, same cycle/treadmill layout — to a CSV file. Two additions make it analysis-ready:

• A Phase column labels every row rest, exercise, at, peak, or recovery (the AT and Peak rows take precedence on their specific row, mirroring the table's highlighting).
• Time is written as both Time (mm:ss, as shown) and Time_sec (raw elapsed seconds).

The file is named from the patient and date of service (e.g. Parker_Riley_2026-07-02_bxb-timedown.csv).

Apply Changes to Test

Clicking Apply Changes to Test on the title row commits the working draft. After Apply:

• If you changed only AT (moved or cleared it), only the AT values change: every AT field is taken from the plotted data at the new AT point (VO₂ at AT is the value you set), or cleared, along with the VE/VCO₂ slope to AT and the AT time. Rest, Peak, Recovery and the slope to peak are untouched — so fixing an AT that an import got wrong never disturbs the cart's reported peak.
• If you changed the averaging method or the exercise window, the whole summary is recomputed: Rest / AT / Peak values, both VE/VCO₂ slopes, the applied averaging method (recorded in test metadata and carried through .pedal round-trip) and the window markers.

If Breeze already averaged this test's breath data, the averaging buttons are locked and a note says so — the rows are shown and summarised as exported.

The button is disabled when no draft changes are pending.

← Entering CPET Data · Home · Normative Predictions →