향에서 약 3배, 역방향에서 최대 10배 증가하였다. 반면, 시표적과의 위치 오차는 큰 변동 없이 유지되었다. 안구운동반응 분류 성능 비교 결과, 샘플링 주파수 90Hz 이상에서는 정확도 .98, 정밀도 .99, F1 점수 .99, 일치도 계수 .95이상을 유지하였으나, 60Hz 이하에서는 F1 점수가 .91 이하, 일치도 계수는 .63 이하로 급격히 저하되었다. 본 연구결과는 신속보기 안구운동 연구에서 시선추적 장비의 샘플링 주파수가 분석 정확도와 신뢰도에 실질적인 영향을 미친다는 점을 시사하며, 300Hz 기준의 시선 데이터 분석 수준을 안정적으로 유지하기 위해서는 최소 90Hz 이상의 샘플링 주파수 확보가 필요함을 제안한다.

This study quantitatively investigated the effects of the sampling frequency of eye-tracking equipment on gaze metrics and response classification performance in a saccade-based eye movement task. Thirty healthy older adults performed an interleaved prosaccade and antisaccade task. Eye gaze data were collected at a baseline frequency of 300 Hz and then downsampled to a range of 30-200 Hz for analysis. The lower sampling frequencies led to significant changes in gaze metrics. Specifically, saccade latency increased by more than 27 ms, whereas saccade amplitude decreased by approximately 2-3°. Additionally, peak velocity dropped by 90-120°/s. Fixation duration increased by as much as 397 ms, and fixation variability rose sharply by approximately three-fold in the prosaccade condition and up to 10-fold in the antisaccade condition. Conversely, the position error relative to the target remained mostly stable. The performance of gaze response classification was assessed using evaluation metrics. At sampling frequencies of ≥90 Hz, the performance remained stable, with accuracy ≥.98, precision ≥.99, F1 score ≥.99, and Cohen’s kappa ≥.95. However, at sampling frequencies of ≤60 Hz, the F1 score dropped to ≤.91, and Cohen’s kappa fell to ≤.63. These findings suggest that the sampling frequency substantially affects the accuracy and reliability of gaze analysis in eye tracking-based saccadic movement studies. A minimum sampling frequency of 90 Hz is recommended to ensure the extraction of gaze metrics and the performance of response classification.