Skip to content

GLOSSARY.md

Latest commit

 

History

History
161 lines (138 loc) · 10.5 KB

GLOSSARY.md

File metadata and controls

161 lines (138 loc) · 10.5 KB

Glossary

General Information

  • StepCountVersion: Version number of the stepcount program used to run the analysis. This information is useful to ensure reproducibility.
  • StepCountArgs: Full list of command-line arguments supplied to stepcount. This information is useful to ensure reproducibility.
  • Filename: The name of the input file.
  • Filesize(MB): The size of the file in megabytes.
  • Device: Brand, model, or file extension of the recording device (e.g., .cwa, .csv).
  • DeviceID: A unique identifier for the specific device used.
  • ReadErrors: Count of read failures or corrupt blocks encountered during file parsing.
  • SampleRate(Hz): Number of sensor readings captured per second.
  • ReadOK: Boolean flag (1 = successful import, 0 = failure).
  • NumTicks: Total number of raw data samples (ticks) recorded.
  • StartTime: Time when recording began.
  • EndTime: Time when recording ended.
  • WearStartTime: Time marking when valid wear data begins (after non-wear filtering).
  • WearEndTime: Time marking when valid wear data ends.
  • WearTime(days): Total duration of valid wear time, expressed in days.
  • NumInterrupts: Number of interruptions in the data recording.
  • CalibErrorBefore(mg): Calibration error before any correction was applied, measured in milli-g.
  • CalibOK: Boolean flag (1 = calibration succeeded, 0 = failed).
  • CalibErrorAfter(mg): Calibration error after attempting correction, measured in milli-g.
  • NonwearTime(days): Total duration classified as non-wear.
  • NumNonwearEpisodes: Number of distinct non-wear intervals detected.
  • ResampleRate(Hz): Sampling frequency after resampling.
  • NumTicksAfterResample: Number of samples after resampling.
  • Covers24hOK: Flag indicating whether every hour of the 24‑hour cycle has at least one wear epoch on any day (1 = fully covered, 0 = at least one hour never covered across all days).

ENMO

  • ENMO(mg): Euclidean norm minus one and zero-truncated, measured in milli-g.
  • ENMO(mg)_Hour{XX}: ENMO in the hour bin {XX} (00–23).
  • ENMO(mg)_Weekend: Overall average ENMO in weekend days.
  • ENMO(mg)_Hour{XX}_Weekend: ENMO in the hour bin {XX} (00-23) on weekend days.
  • ENMO(mg)_Weekday: Overall average ENMO in weekdays (Monday–Friday).
  • ENMO(mg)_Hour{XX}_Weekday: ENMO in the hour bin {XX} (00-23) on weekdays.

Steps

  • TotalSteps: Total step count.
  • StepsDayAvg: Mean daily steps.
  • StepsDayMed: Median daily steps.
  • StepsDayMin: Minimum steps in any single day.
  • StepsDayMax: Maximum steps in any single day.
  • Steps_Hour{XX}: Steps in the hour bin {XX} (00–23).

Steps -- Weekends Only

  • TotalSteps_Weekend: Total step count on weekends.
  • StepsDayAvg_Weekend: Mean daily steps on weekends.
  • StepsDayMed_Weekend: Median daily steps on weekends.
  • StepsDayMin_Weekend: Minimum daily steps on a weekend day.
  • StepsDayMax_Weekend: Maximum daily steps on a weekend day.
  • Steps_Hour{XX}_Weekend: Steps in the hour bin {XX} (00-23) on weekends.

Steps -- Weekdays Only

  • TotalSteps_Weekday: Total step count on weekdays.
  • StepsDayAvg_Weekday: Mean daily steps on weekdays.
  • StepsDayMed_Weekday: Median daily steps on weekdays.
  • StepsDayMin_Weekday: Minimum daily steps on a weekday.
  • StepsDayMax_Weekday: Maximum daily steps on a weekday.
  • Steps_Hour{XX}_Weekday: Steps in the hour bin {XX} (00-23) on weekdays.

Walking

  • TotalWalking(mins): Total walking duration.
  • WalkingDayAvg(mins): Average daily walking duration.
  • WalkingDayMed(mins): Median daily walking duration.
  • WalkingDayMin(mins): Minimum walking duration in any single day.
  • WalkingDayMax(mins): Maximum walking duration in any single day.
  • Walking(mins)_Hour{XX}: Walking time in the hour bin {XX} (00-23).

(Weekend and Weekday subsections analogous to Steps above.)

Cadence

  • CadencePeak1(steps/min): Highest cadence per day.
  • CadencePeak30(steps/min): Mean cadence of the thirty most active one-minute epochs per day.
  • Cadence95th(steps/min): 95th percentile of cadence per day.
    • Note: Cadence metrics are computed from minutes classified as walking; if fewer than a minimum threshold (default 5) are present in a day, values will be NaN.

(Weekend and Weekday subsections analogous to Steps above.)

Steps Distribution

  • Steps5thAt: Average clock time when 5% of total daily steps are reached.
  • Steps25thAt: Average clock time when 25% of total daily steps are are reached.
  • Steps50thAt: Average clock time when half (50%) of total daily steps are reached.
  • Steps75thAt: Average clock time when 75% of total daily steps are reached.
  • Steps95thAt: Average clock time when 95% of total daily steps are reached.

Crude vs. Adjusted Estimates

Crude estimates represent raw metrics calculated directly from observed data. Adjusted estimates compensate for missing time-series values by imputing each absent timepoint with the average value at that same clock time across all other recorded days. To derive adjusted totals and daily summaries, any gaps in the required 24‑hour span are similarly imputed; if data remain missing after this process, the estimate is reported as NaN. Adjusted metrics are labeled with an "Adjusted" suffix—for example, StepsDayAvgAdjusted_Weekend.

Default wear thresholds used for adjusted aggregations

  • Day: 21 hours minimum (--min-wear-per-day 1260).
  • Hour: 50 minutes minimum (--min-wear-per-hour 50).
  • Minute: 30 seconds minimum (--min-wear-per-minute 0.5).

These thresholds can be changed via CLI flags. Adjusted daily/cadence summaries form a 7‑day representative week before final aggregation; very short recordings may yield NaNs.

CSV Output Schemas

All CSV files are gzipped (.csv.gz). Column schemas:

  • Steps.csv.gz
    • time, Steps (per-window step counts; SSL uses 10 s windows).
  • StepTimes.csv.gz
    • time (one row per detected step).
  • Minutely.csv.gz / MinutelyAdjusted.csv.gz
    • Filename, Time, Steps, ENMO(mg) (Adjusted spans full 24 h grid with imputation where applicable).
  • Hourly.csv.gz / HourlyAdjusted.csv.gz
    • Filename, Time, Steps, ENMO(mg).
  • Daily.csv.gz / DailyAdjusted.csv.gz
    • Filename, Date, Walk(mins), Steps, Steps5thAt, Steps25thAt, Steps50thAt, Steps75thAt, Steps95thAt, CadencePeak1(steps/min), CadencePeak30(steps/min), Cadence95th(steps/min), ENMO(mg).
  • Bouts.csv.gz
    • Filename, StartTime, EndTime, Duration(mins), TimeSinceLast(mins), Steps, Cadence(steps/min), CadenceSD(steps/min), Cadence25th(steps/min), Cadence50th(steps/min), Cadence75th(steps/min), ENMO(mg), ENMOMed(mg).

Random Forest Feature List

The table below describes the handcrafted features used as inputs to the Random Forest model. These features are extracted from the Euclidean norm of the triaxial accelerometer data.

Feature Name Description Units
Moment features
avg Mean g
std Standard deviation g
skew Skewness
kurt Kurtosis
Quantile Features
min Minimum g
q25 Lower quartile g
med Median g
q75 Upper quartile g
max Maximum g
Autocorrelation features
acf_1st_max Maximum autocorrelation
acf_1st_max_loc Location of 1st autocorrelation maximum s
acf_1st_min Minimum autocorrelation
acf_1st_min_loc Location of 1st autocorrelation minimum s
acf_zeros Number of autocorrelation zero-crossings
Spectral features
pentropy Signal's spectral entropy nats
power Signal's total power g2/s
f1, f2, f3 1st, 2nd and 3rd dominant frequencies Hz
p1, p2, p3 Power spectral densities of respective dominant frequencies g2/s
fft0, fft1, fft2, ... Power spectral density for frequencies 0Hz, 1Hz, 2Hz, ... g2/s
Peak features
npeaks Number of peaks in the signal per second 1/s
peaks_avg_promin Average prominence of peaks g
peaks_min_promin Minimum prominence of peaks g
peaks_max_promin Maximum prominence of peaks g