Early antidepressant treatment response prediction in major depression using clinical and TPH2 DNA methylation features based on machine learning approaches

Table 2 Formulas in classifiers

Data set	Feature selection	Formulas	Number
Only demographic and clinical features	Without RFE	w2yn ~ gen + age + les + ctq + his + edu + mar + fst + mon + times + fstyn + HAMD0	12
Only demographic and clinical features	With RFE	w2yn ~ HAMD0 + age + times + les + ctq + gen + edu + fst + fstyn + mar	10
Only TPH2 CpGs	Without RFE	w2yn ~ t1 + t2 + … + t38^a	35
Only TPH2 CpGs	With RFE	w2yn ~ t6 + t12 + t13 + t11 + t29 + t17 + t15 + t32 + t33 + t16 + t2 + t18 + t14 + t28 + t20 + t31	16
Both	Without RFE	w2yn ~ gen + age + les + ctq + his + edu + mar + fst + mon + times + fstyn + HAMD0 + t1 + t2 + … + t38 ^a	47
Both	With RFE	w2yn ~ t12 + t17 + t6 + t15 + t13 + t32 + gen + t11 + times + age + t14 + t29 + HAMD0 + t16 + fst + t18 + mar + les + t31 + t33 + ctq + t4 + t2	23

^a 35 CpGs features were included in the model, and 3 CpGs (TPH2-7-99、TPH2-7-142、TPH2-7-170) were excluded with more than 45% missing values

ISSN: 1471-244X