An autocorrelation analysis of distance measurements to detect differences in visual behavior between educational systems and between emmetropic and myopic children

Aim and Research Question(s)

The aim of this thesis is to contribute to an interdisciplinary study investigating risk factors for myopia using wearable technology. This contribution is made by subjecting the data collected in a previously conducted study from children (8-10 years) in two Hong Kong school systems from a local (n=28) and an international school (n=27) to an extended data analysis.

Research Question: Do near-work behavioral dynamics differ in emmetropic and myopic children, and in different classroom environments?

Background

Short-sightedness has become a worldwide public health issue [1]. It is estimated that by the end of this decade, one-third of the world’s population could be affected by myopia [2]. The health problem of increasing and rapidly progressing myopia during childhood is that children may develop high myopia leading to several sight-threatening complications [3].

Methods

The data recorded in the study included the children's working distances recorded at high frequency (10 samples/ second) during a typical 90-minute lesson in their classroom using a head-mounted LIDAR.

In consultation with the clinical partners involved in the study, who approved the research question, analysis methods were explored to assess behavioral dynamics. For the analysis, a program was implemented that was responsible for analyzing the visual dynamics and performing statistical tests.

Visual behavior was expressed as dioptric viewing distance (1/working distance). To quantify the dynamic nature of the visual behaviour, autocorrelation was calculated, i.e., the correlation of a dioptric distance with a delayed copy of itself, as a function of delay.

Results and Discussion

The dioptric distances of local schoolchildren changed significantly slower than those of international schoolchildren (0.33±0.10 s-1 vs. 0.66±0.24 s-1; Kolmogorov–Smirnov statistic (KSs) = 0.77, P<0.0001).
No statistical difference was found in the dioptric distance dynamics between emmetropic and myopic children in the same school using the two-sampled KS test (local school KSs = 0.4, P = 0.19 and international school KSs = 0.26, P = 0.68)

Conclusion

The quantitative analysis based on LIDAR data revealed sizable differences in the dynamic visual behaviors of Hong Kong children attending two schools that employ contrasting pedagogical approaches, but not between emmetropic and myopic children within the same school.
The results indicate that the education system plays a predominant role in determining children’s visual behavior and thus can be an environmental risk factor for myopia development.

References

[1] R. Pararajasegaram, ‘VISION 2020-the right to sight: from strategies to action’, Am. J. Ophthalmol., vol. 128, no. 3, pp. 359–360, Sep. 1999, doi: 10.1016/s0002-9394(99)00251-2.
[2] J. Myopia et al., The impact of myopia and high myopia. Report of the Joint World Health Organization-Brien Holden Vision Institute Global Scientific Meeting on Myopia. 2015. [Online]. Available: https://myopiainstitute.org/wp-content/uploads/2020/10/Myopia_report_020517.pdf
[3] K. Ohno-Matsui et al., ‘International photographic classification and grading system for myopic maculopathy’, Am. J. Ophthalmol., vol. 159, no. 5, pp. 877-883.e7, May 2015, doi: 10.1016/j.ajo.2015.01.022.

Project's source code; Github id: leonharj-fh;
Repository: mdh_thesis_analyses_eye_tracker