Discussion on machine learning technology to predict tacrolimus blood concentration in patients with nephrotic syndrome and membranous nephropathy in real-world settings

Table 5 Performance comparison of models according to number of features

No. of Features	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
Accuracy	0.7167	0.7167	0.7167	0.7056	0.7000	0.7056	0.7111	0.7333	0.7258	0.7268	0.7278	0.7167	0.7222	0.7167	0.7278	0.7333	0.7222
F-beta	0.9267	0.9267	0.9267	0.9011	0.9045	0.8823	0.9071	0.9124	0.9102	0.9093	0.9064	0.9038	0.9051	0.8944	0.9064	0.9170	0.9143
Recall	1.0000	1.0000	1.0000	0.9612	0.9690	0.9302	0.9690	0.9690	0.9664	0.9745	0.9612	0.9612	0.9612	0.9457	0.9612	0.9767	0.9767
Precision	0.7167	0.7167	0.7167	0.7209	0.7143	0.7317	0.7225	0.7396	0.7384	0.7174	0.7381	0.7294	0.7337	0.7349	0.7381	0.7368	0.7283
AUC	0.5000	0.5000	0.500	0.5100	0.4943	0.5337	0.5139	0.5531	0.5374	0.5315	0.5492	0.5296	0.5394	0.5415	0.5492	0.5472	0.5276

ISSN: 1472-6947