Visual Attention Dynamics in Children with Special Needs During Eye-Tracking-Based Adaptive Learning Game Sessions in Pontianak, Indonesia

Abstract

Background: Children with special needs (CSN), particularly those with attention disorders and autism spectrum conditions, may experience difficulty sustaining visual attention consistently. Eye-tracking technology allows visual attention to be measured objectively through eye-movement characteristics. Integrating eye tracking with an adaptive learning game has the potential to produce visual stimulation that is more responsive to individual user characteristics. Objective: This study aimed to analyze changes in visual-attention indicators of children with special needs across three measurement sessions using an eye-tracking-based adaptive learning game. Methods: This study used a quantitative approach with a one-group repeated-measures design. Sixteen children with special needs receiving services at the Disability Services and Assessment Center (LDAC) of Pontianak City served as subjects. Each participant underwent three measurement sessions on different days using an adaptive learning game system integrated with an eye tracker. The visual-attention indicators analyzed were mean fixation duration, saccade count, mean saccade distance, and gaze transition rate. Descriptive analysis was performed by calculating means and standard deviations. Differences across the three measurements were analyzed using the Friedman test, given the repeated-measures design on the same subjects. Post hoc analysis was conducted using the Wilcoxon signed-rank test with Bonferroni correction. Results: Mean fixation duration increased from 494.50 ± 75.72 ms at the first measurement to 541.10 ± 108.74 ms and 564.87 ± 136.77 ms at the second and third measurements, but the difference across measurements was not significant (p = 0.087). Saccade count showed a significant change (p = 0.002), with the mean rising from 129.81 ± 83.31 to 223.25 ± 99.93, then decreasing to 147.69 ± 79.84. Gaze transition rate also showed a significant difference (p = 0.003), from 7.58 ± 2.66 times/second to 13.68 ± 5.56 times/second and then 9.18 ± 3.72 times/second. Mean saccade distance showed no significant difference (p = 0.087), changing from 273.03 ± 46.66 pixels, decreasing to 242.06 ± 32.52 pixels, and increasing again to 257.24 ± 42.13 pixels. Conclusion: The eye-tracking-based adaptive learning game was associated with changes in eye-movement dynamics, particularly saccade count and gaze transition rate, across three measurement sessions. However, these changes did not follow a linear improvement pattern across all indicators. The results indicate an active oculomotor exploration phase at the second session followed by stabilization at the third session, a pattern relevant for the design of intervention sessions for children with special needs.