Complete Gazepoint Downstream Workflow

In eyeprocess: Harmonize Eye-Tracking, Pupillometry, Biometrics, and Psychometric Process Data

knitr::opts_chunk$set(collapse = TRUE, comment = "#>")

Purpose

run_gazepoint_workflow() executes the full research-data workflow from a real Gazepoint Analysis folder:

1. canonical eye_dataset import;
2. file-pair, timebase, coordinate, sampling-rate, and signal-quality audits;
3. contiguous media-run reconstruction as person-by-item-by-trial intervals;
4. vendor-fixation and AOI summaries;
5. short-gap pupil interpolation, optional filtering, and blink detection;
6. valid-only biometric summaries while preserving native values;
7. gaze, pupil, biometric, AOI, and QC plots;
8. one-row-per-person-item-trial process tables;
9. response templates and IRT-ready long/matrix structures;
10. canonical exports, provenance, source fingerprints, and reproducible reports.

The workflow does not manufacture response scores. When no observed responses are supplied, the result is marked process_ready_response_pending.

Minimal workflow

library(eyeprocess)

source_dir <- "path/to/eyeprocess-validation-corpus/cases/gazepoint-analysis-v7.2.0-demo"
output_dir <- "path/to/eyeprocess-downstream-output"

result <- run_gazepoint_workflow(
  source_dir,
  output_dir = output_dir,
  overwrite = TRUE
)

result
validate_gazepoint_workflow(result)

Explicit specification

Pupil baseline correction is deliberately disabled by default. The first samples after media onset are not automatically equivalent to a pre-stimulus baseline.

spec <- gazepoint_workflow_spec(
  expected_sampling_rate = 60,
  minimum_valid_gaze = 0.80,
  minimum_valid_pupil = 0.70,
  pupil_interpolation = "linear",
  pupil_max_gap_ms = 150,
  pupil_filter = "median",
  pupil_window = 5,
  pupil_baseline = "none",
  create_plots = TRUE,
  create_html_report = TRUE,
  retain_raw = TRUE
)

Item labels and conditions

By default, item_id equals Gazepoint MEDIA_ID. A study-specific mapping can supply meaningful item and condition labels.

item_map <- data.frame(
  stimulus_id = c("0", "1"),
  item_id = c("item_control", "item_treatment"),
  condition_id = c("control", "treatment")
)

result <- run_gazepoint_workflow(
  source_dir,
  output_dir,
  item_map = item_map,
  spec = spec,
  overwrite = TRUE
)

Adding observed responses

Responses may be supplied now or joined later using the generated irt/response-template.csv file.

responses <- data.frame(
  participant_id = c("User 3", "User 3"),
  item_id = c("item_control", "item_treatment"),
  response = c("yes", "no"),
  score = c(1, 1),
  response_time = c(6.1, 7.4)
)

result <- run_gazepoint_workflow(
  source_dir,
  output_dir,
  responses = responses,
  item_map = item_map,
  spec = spec,
  overwrite = TRUE
)

The workflow creates response and response-time matrices only when the relevant observations are available. It does not fit IRT automatically; model adequacy, sample size, item count, dimensionality, and process-covariate assumptions must be evaluated first.

Output structure

eyeprocess-downstream-output/
├── canonical-dataset/
├── qc/
├── tables/
├── irt/
├── plots/
│   ├── summary/
│   ├── gaze/
│   ├── fixations/
│   ├── pupil/
│   └── biometrics/
├── gazepoint-workflow-report.md
├── gazepoint-workflow-report.html
├── workflow-result.rds
├── workflow-spec.rds
├── source-fingerprint.csv
├── session-info.txt
└── rerun-workflow.R

Interpretation boundaries

Fixations are not automatically attention; dwell time is not automatically difficulty; pupil dilation is not automatically cognitive load; and GSR or heart rate does not identify a specific emotion. The report preserves these interpretive safeguards alongside the analysis outputs.

eyeprocess documentation built on Sept. 28, 2026, 5:08 p.m.