Determining aqueous humor immune mediators of intraocular diseases by artificial intelligence
Purpose
Research on immune-related factors has great clinical significance in the field of ophthalmology, because analysis of immune mediators contributes to diagnosis and treatment. In this study, we used artificial intelligence using support vector machine (SVM) to analyze 27 immune-related factors measured in aqueous humor samples from patients with intraocular diseases.
Methods
Aqueous humor samples were collected from 339 eyes with 13 intraocular diseases at the Tokyo Medical University Hospital. 27 immune mediators were measured by CBA flex kit. The measurements were converted to natural logarithm. 20% of the samples from each disease were randomly extracted and used as test data. The remaining 80% were used as training data to create a prediction model. Data were analyzed by SVM in R (version 3.5.1).
Results
In the 1st prediction, anterior chamber was 64% correct answer rate. It was 78% up to the 2nd prediction, and 90% up to the 3rd prediction. Concerning vitreoretinal lymphoma and glaucoma, the sensitivity for both was 100%, while corrected answer rates were 100% and 88.9%, respectively. Acute retinal necrosis showed sensitivity of 100% but correct answer rate was 63.6%.
Conclusion
As a result of analyzing only 27 immune-related factors in intraocular fluids by SVM, aqueous humor was useful for classification of several intraocular diseases.
Conflict of interest
No
Author 1
Last name
USUI
Initials of first name(s)
Y
Department
Ophthalmology, Tokyo Medical University Hospital
City
Shinjuku-ku
Country
Japan
Author 2
Last name
Nezu
Initials of first name(s)
N
Department
Ophthalmology, Tokyo Medical University Hospital
City
Shinjuku-ku
Country
Japan
Author 3
Last name
Tsubota
Initials of first name(s)
K
Department
Ophthalmology, Tokyo Medical University Hospital
City
Shinjuku-ku
Country
Japan
Author 4
Last name
Goto
Initials of first name(s)
H
Department
Ophthalmology, Tokyo Medical University Hospital
City
Shinjuku-ku
Country
Japan
This website uses cookies to ensure you get the best experience on our website.
Learn more