Chief Supervisor

Prof. KEE Chea-su

Project Title

Synopsis

Myopia is becoming a global public health problem due to its high prevalence rate and complex aetiology. Several studies showed that the development of myopia involves both genetic and environmental factors, and sometimes there may be relevant multiple structural and functional anomalies found in myopic eyes. Up to now, no prediction system has been widely accepted and validated for myopia development.

Machine learning methods build a mathematical model based on the large-scale training data, not by explicit rules but relying on inherent patterns instead, and make predictions in the testing data. Compared with statistical modelling, machine learning techniques can improve the predictive performance by using more flexible models, especially for the complex data with high dimensionality, non-linearity, and heterogeneity. Since the reasons leading to myopia are complicated and various factors might contribute to the myopia development, it is expected that the machine learning based method may have better performance, combining more complex variables and information.