Detect Atrial Fibrillation / Frequent Ectopy from RR Irregularity

Classifies the cardiac rhythm of each channel as regular sinus, atrial fibrillation (AF), or frequent ectopy from the irregularity of the RR interval series. HRV metrics are physiologically undefined in AF, so this gate is used to invalidate them (see rhythm_check in ecgHRVtime, ecgHRVfreq and ecgHRVnonlinear).

Usage

ecgDetectAF(
  rr,
  window_beats = NULL,
  ectopic_frac_max = 0.05,
  rmssd_ratio_af = 0.15,
  entropy_af = 0.7,
  n_bins = 16L,
  drr_range = c(-600, 600),
  threshold_ms = 300,
  min_beats = 20L
)

Arguments

rr

A data.frame with columns channel and rr_ms, as returned by ecgRRintervals.

window_beats

If non-NULL, an integer window length (in beats): instead of a single per-channel verdict, a continuous AF probability is returned for each non-overlapping window of window_beats beats (see Value).

ectopic_frac_max

Maximum ectopic fraction for a "sinus" verdict (default 0.05).

rmssd_ratio_af

Minimum RMSSD/mean-RR ratio consistent with AF (default 0.15).

entropy_af

Minimum normalized dRR Shannon entropy consistent with AF (default 0.7, range 0-1).

n_bins

Number of histogram bins for the dRR entropy (default 16).

drr_range

Numeric length-2 dRR window (ms) for the entropy histogram (default c(-600, 600)).

threshold_ms

Deviation threshold (ms) passed to ecgQualityCheck for the ectopic fraction (default 300).

min_beats

Minimum RR intervals required to classify; fewer yields NA (default 20).

Value

With window_beats = NULL (default), a data.frame with one row per channel and columns channel, rhythm ("sinus", "AF" or "frequent_ectopy"), ectopic_frac, rmssd_ratio, drr_entropy, af_prob (a continuous AF probability in 0-1) and n_beats. With window_beats set, a data.frame with one row per non-overlapping window and columns channel, window, beat_start, n_beats, af_prob and its components (rmssd_ratio, drr_entropy, sample_entropy, sd1_sd2).

Details

Three descriptors are combined: the ectopic-beat fraction from ecgQualityCheck, the RMSSD/mean-RR normalized irregularity index, and the normalized Shannon entropy of the successive-difference (dRR) histogram. AF requires both large beat-to-beat variability (rmssd_ratio) and a broad, disordered dRR distribution (drr_entropy); a large but concentrated irregularity (e.g. an ectopic couplet) falls through to "frequent_ectopy".

References

Tateno, K. & Glass, L. (2001). Automatic detection of atrial fibrillation using the coefficient of variation and density histograms of RR and dRR intervals. Medical & Biological Engineering & Computing, 39(6), 664-671. Sarkar, S., Ritscher, D. & Mehra, R. (2008). A detector for a chronic implantable atrial tachyarrhythmia monitor. IEEE Transactions on Biomedical Engineering, 55(3), 1219-1224.

See also

ecgQualityCheck, ecgHRVtime

Examples

set.seed(1)
rr <- data.frame(channel = 1L, rr_ms = 800 + rnorm(100, sd = 20),
                 time_sec = cumsum(rep(0.8, 100)))
ecgDetectAF(rr)
#>   channel rhythm ectopic_frac rmssd_ratio drr_entropy   af_prob n_beats
#> 1       1  sinus            0  0.03164052   0.2485076 0.5644174     100