segmentWalking {adept}

Walking Stride Pattern Segmentation from Raw Accelerometry Data via ADEPT

Description

Segment walking stride pattern from a raw accelerometry data time-series (x,y,z) via Adaptive Empirical Pattern Transformation (ADEPT). Default algorithm parameters are optimized for a wrist-worn sensor and were evaluated with data collected in the free-living environment.

Usage

segmentWalking(
  xyz,
  xyz.fs,
  template,
  sim_MIN = 0.85,
  dur_MIN = 0.8,
  dur_MAX = 1.4,
  ptp_r_MIN = 0.2,
  ptp_r_MAX = 2,
  vmc_r_MIN = 0.05,
  vmc_r_MAX = 0.5,
  mean_abs_diff_med_p_MAX = 0.5,
  mean_abs_diff_med_t_MAX = 0.2,
  mean_abs_diff_dur_MAX = 0.2,
  compute.template.idx = FALSE,
  run.parallel = FALSE,
  run.parallel.cores = 1
)

Arguments

xyz

A numeric matrix (or data frame) of n × 3 dimension. Three-dimensional raw accelerometry data time-series; acceleration measurements (x,y,z) collected along three orthogonal axes by the sensor's accelerometer.

xyz.fs

A numeric scalar. Frequency at which a time-series (x,y,z) is collected, expressed in a number of observations per second.

template

A list of numeric vectors, or a numeric vector. Distinct pattern template(s) of walking stride.

sim_MIN

numeric scalar. Minimum value of correlation between pattern template(s) and (r_t)_t vector magnitude of accelerometry data. Default used is 0.85.

dur_MIN

A numeric scalar. Minimum value of a stride duration allowed to be identified. Expressed in seconds. Default used is 0.8.

dur_MAX

A numeric scalar. Maximum value of a stride duration allowed to be identified. Expressed in seconds. Default used is 1.4.

ptp_r_MIN

A numeric scalar. Minimum value of "peak to peak" difference in (r_t)_t vector magnitude data of a stride. Default used is 0.2.

ptp_r_MAX

A numeric scalar. Maximum value of "peak to peak" difference in (r_t)_t vector magnitude data of a stride. Default used is 2.0

vmc_r_MIN

A numeric scalar. Minimum value of VMC in (r_t)_t vector magnitude data of a stride. Default used is 0.05.

vmc_r_MAX

A numeric scalar. Maximum value of VMC in (r_t)_t vector magnitude data of a stride. Default used is 0.5.

mean_abs_diff_med_p_MAX

A numeric scalar. Maximum value of MAD* of Azimuth (az_)_t median for 3 subsequent valid strides. Here, MAD* stands for mean of 2 absolute differences between 3 subsequent values. Default used is 0.5.

mean_abs_diff_med_t_MAX

A numeric scalar. Maximum value of MAD* of Elevation (el_)_t median for 3 subsequent valid strides. Here, MAD* stands for mean of 2 absolute differences between 3 subsequent values. Default used is 0.2.

mean_abs_diff_dur_MAX

A numeric scalar. Maximum value of MAD* of duration time for 3 subsequent valid strides. Here, MAD* stands for mean of 2 absolute differences between 3 subsequent values. Default used is 0.2.

compute.template.idx

A logical scalar. Whether or not to compute and return information about which of the provided pattern templates yielded a similarity matrix value that corresponds to an identified pattern occurrence. Setting to TRUE may increase computation time. Default is FALSE.

run.parallel

A logical scalar. Whether or not to use parallel execution in the algorithm with parallel package. Default is FALSE. DOES NOT WORK ON WINDOWS.

run.parallel.cores

An integer scalar. The number of cores to use for parallel execution. Defaults to 1L (no parallel). DOES NOT WORK ON WINDOWS.

Value

A data.frame with segmentation results. Each row describes one identified pattern occurrence:

• tau_i - row index of xyz where pattern starts,

• T_i - pattern duration, expressed in xyz vector length,

• sim_i - similarity between a pattern and best-fit template; see segmentPattern for details,

• template_i - if compute.template.idx equals TRUE: index of a template best matched to x; if compute.template.idx equals FALSE: NA,

• is_walking_i – 1 if a pattern is identified as walking stride; 0 otherwise.

Examples

library(adeptdata)
library(dplyr)
library(adept)
xyz <-
  adeptdata::acc_walking_IU %>%
  filter(loc_id == "left_wrist", subj_id == "id86237981") %>%
  arrange(time_s) %>%
  select(v1 = x, v2 = y, v3 = z) %>%
  as.matrix()
# define raw accelerometry data sample frequency
xyz.fs <- 100

# define template list based on predefined templates
template_mat <- adeptdata::stride_template$left_wrist[[3]]
template <- list(
  template_mat[1, ],
  template_mat[2, ],
  template_mat[3, ]
)

# run walking segmentation
# (parallel supported, except for Windows; see run.parallel, run.parallel.cores args)
segmentWalking(xyz, xyz.fs, template)