Method for diagnosing keratoconus case based on XGBoost+SVM hybrid machine learning

Abstract

The invention provides a method for diagnosing a keratoconus case based on XGBoost+SVM hybrid machine learning. The method includes the following steps that: the corneal examination data of ophthalmicpatients are collected, each corneal sample is labeled with a category label keratoconus, suspected keratoconus and normal cornea by ophthalmologists, and the labeled corneal samples are adopted as training sample data; the feature values of the features of the corneal sample data are normalized and are mapped to an interval [0, 1]; the XKoost is used to expand the features of the sample data, and an expanded feature set is adopted as the training features of the samples; and based on the training features of the sample data, an SVM diagnosis model is trained and constructed; and the diagnosis model is adopted to diagnose and predict new cases. Tests have shown that the diagnostic effect of the method has met the requirements of clinical applications. The method can be used for screeningthe keratoconus, especially suspected keratoconus, can reduce dependence on the diagnosis of medical experts, and can basically improve diagnostic efficiency and accuracy.

Description

technical field [0001] The invention belongs to the field of ophthalmic medical diagnosis, and relates to machine learning technology, in particular to a method for diagnosing keratoconus disease based on XGBoost+SVM hybrid machine learning. Background technique [0002] Keratoconus (keratoconus) is a primary corneal degenerative disease characterized by corneal dilation, which causes the central or paracentral area of ​​the cornea to bulge forward into a cone and produces highly irregular myopic astigmatism and different visual impairments. A stand-alone disorder that can also be part of multiple syndromes. It mostly occurs around puberty and is not accompanied by inflammation. In the advanced stage, acute corneal edema, scarring, and severe visual impairment can occur. Overt keratoconus is easy to diagnose. Early diagnosis is difficult when the appearance and slit lamp findings are not typical. The most effective method is corneal topography. But even so, the diagnost...

AI Technical Summary

Problems solved by technology

But even so, the diagnostic criteria for early keratoconus are not completely unified. For relatively complicated cases of suspected keratoconus, detailed consultation and discussion by experienced experts are often required to give a relatively accurate diagnosis; at the same time, a large number of cases, limited experts, The complex corneal parameters add great difficulty to the early diagnosis of keratoconus

Images