Children myopia influence factor evaluation method based on causal learning

• Main Authors: JIA HANZHU, KWON BONGJO, YANG XU, ZHAO WUYAN, WANG JIAYUN
• Format: Patent
• Language: Chinese; English
• Published: 04.06.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATIONTECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING ORPROCESSING OF MEDICAL OR HEALTHCARE DATA; INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTEDFOR SPECIFIC APPLICATION FIELDS; PHYSICS

The invention discloses a child myopia influence factor evaluation method based on causal learning. The method comprises the following steps: preprocessing a child ophthalmology data set; constructing a four-layer feed-forward neural network prediction model according to the data features, and abstracting the four-layer feed-forward neural network prediction model into a structural causal model; performing causal discovery on the children ophthalmology data set by using a PC algorithm based on degenerated Gaussian, constructing a causal graph, and dividing input layer neurons into three classes according to the generated causal graph; for the three types of neurons under different conditions, the methods of directly calculating the average causal effect, calculating the average causal effect after correcting bias and calculating the average causal effect after eliminating mixing are respectively adopted for attribution. According to the method, the influence factors possibly causing children myopia are resear