Development of a classification system based on corneal biomechanical properties using artificial intelligence predicting keratoconus severity

Table 5 Confusion matrix of keratoconus detection by LDA, optimized RF (DKI), CBI and TBI on validation dataset

		Reference diagnosis (healthy – KC)
		Healthy	KC	S_n/S_p	Accuracy
Final LDA	Healthy	31	7	93%/94%	93%
Final LDA	KC	2	86	93%/94%	93%
DKI	Healthy	30	6	94%/91%	93%
DKI	KC	3	87	94%/91%	93%
CBI	Healthy	31	11	88%/94%	90%
CBI	KC	2	82	88%/94%	90%
TBI	Healthy	32	0	100%/97%	99%
TBI	KC	1	93	100%/97%	99%

CBI Corvis biomechanical index; DKI Dresden keratoconus index based on optimized RF model; KC keratoconus; LDA linear discriminant analysis; S_n sensitivity; S_p specificity; TBI tomographic and biomechanical index. Bold signifies correct prediction by LDA or RF model

ISSN: 2326-0254