Risk Assessment of High Myopia in Primary School Students using Bayesian Network Inference

The prevalence of myopia is increasing substantially among young adults and progressive high myopia can cause sight-threatening ocular complications. Hence, an accurate and effective risk assessment for high myopia is important to the control of the myopia progression. In this paper, a Bayesian network-based risk assessment model (BNRAM) was developed by integrating domain knowledge and clinical data. Specifically, association rules were applied on a real data set containing clinical records of primary school students with myopia. The valuable and meaning association relationship between risk factors and the severity of myopia was mined. Then, the risk analysis based on a complex network was illustrated to explore the evolution regularity of myopia. Bayesian network inference was further utilized to achieve the risk prediction. Experimental results showed that the proposed BNAM could predict the onset of high myopia. This research provides evidence for personalized interventions to control myopia in primary school students.