		The Random Forests machine learning method is a useful tool to help ophthalmologists distinguish between the visual fields (VFs) of individuals with healthy eyes and those with preperimetric open-angle glaucoma (OAG) in which glaucoma will manifest, according to research from Japan published in 

Investigative Ophthalmology & Visual Science

		The investigators classified all VFs before an initial diagnosis of manifest glaucoma as preperimetric glaucoma VFs. They used the 30-2 threshold test programme of an automated perimeter (Humphrey Field Analyzer, Carl Zeiss Meditec) to obtain a series of 171 VFs in 53 eyes of 51 individuals with OAG or suspected OAG and from 108 healthy eyes of 87 individuals. They used the Random Forests method to calculate the area under the receiver operating characteristic curve (AROC) in distinguishing between the VFs of those with preperimetric glaucoma and those with healthy eyes; they used 52 total deviation (TD) values, mean deviation (MD) and pattern standard deviation (PSD) as predictors.

		The researchers found a significant difference in MD and PSD between VFs of those with healthy eyes and the VFs of those with eyes with preperimetric glaucoma. They also observed a significant difference in the TD values between healthy VFs and preperimetric VFs at 25 out of 52 test points. The AROC obtained using the Random Forests method was 79%.

Articles

The Random Forests machine learning method is a useful tool to help ophthalmologists distinguish between the visual fields (VFs) of individuals with healthy eyes and those with preperimetric open-angle glaucoma (OAG) in which glaucoma will manifest, according to research from Japan.