Machine learning algorithms’ accuracy in predicting kidney disease progression: a systematic review and meta-analysis

Table 2 Summary of meta-analysis and subgroup analysis

Subgroup	Number of ML algorithms	Sensitivity (95% CI)	Specificity (95% CI)	AUC (95% CI)	Correlation coefficient	β	DOR
Total DTA	23	0.68 (0.58–0.77)	0.88 (0.83–0.92)	0.87 (0.84–0.90)	− 0.53	0.015	16.34
Type of KD
CKD	15	0.64 (0.49–0.77)	0.84 (0.74–0.91)	0.82 (0.79–0.85)	− 0.77	− 0.036	9.31
IgAN	8	0.74 (0.71–0.77)	0.93 (0.91–0.95)	0.78 (0.74–0.81)	− 1.0	3.781	39.27
ML algorithm type
Classification	16	0.64 (0.50–0.76)	0.87 (0.79–0.92)	0.84 (0.81–0.87)	− 0.66	0.021	11.75
Regression	7	0.80 (0.74–0.84)	0.91 (0.86–0.95)	N/A	1.0	6.044	41.09
Dataset type
Training set	11	0.56 (0.37–0.73)	0.90 (0.80–0.95)	0.83 (0.80–0.86)	− 0.57	0.074	11.40
Testing set	12	0.79 (0.76–0.82)	0.86 (0.81–0.90)	0.81 (0.77–0.84)	− 1.0	3.693	23.33
Pathology
Y	11	0.71 (0.66–0.76)	0.89 (0.80–0.94)	N/A	1	1.086^a	19.46
N	12	0.65 (0.46–0.81)	0.87 (0.78–0.93)	0.86 (0.83–0.89)	− 0.53	− 0.172	12.92
Race
Asian	16	0.64 (0.49–0.77)	0.84 (0.75–0.91)	0.82 (0.79–0.86)	− 0.76	− 0.042	9.53
Not Asian	7	0.74 (0.71–0.77)	0.93 (0.91–0.95)	0.78 (0.74–0.81)	− 1	3.806	10.95

DTA diagnostic test accuracy, KD kidney disease, CKD chronic kidney disease, IgAN Immunoglobulin A Nephropathy, ML machine learning, Y Yes, N No
^aP < 0.01

ISSN: 1472-6947