TABLE IV.

Performance of predicting smoking status and BMI from the GGMP microbiome data (species level).

Smoking Status	ROC-AUC	F1-Score	Time (s)
DeepEn-Phy (nParameter=158,166)	0.7016	0.6748	2,213
Vanilla MLP (nParameter=171,289)	0.6471	0.6383	652
PopPhy-CNN (nParameter=238,658)	0.6122	0.5946	833
Random Forest (nTree=500)	0.6583	0.5409	408
Gradient Boosting (nTree=100)	0.6610	0.5887	14
Logistic LASSO (L1 λ=0.001)	0.6600	0.6491	19
Logistic Ridge (L2 λ=0.001)	0.6445	0.6615	17
BMI	MSE	RMSE	Time (s)
DeepEn-Phy (nParameter=51,636)	12.5559	3.5434	1,211
Vanilla MLP (nParameter=67,939)	16.3519	4.0437	346
PopPhy-CNN	–	–	–
Random Forest (nTree=500)	12.7475	3.5704	1,325
Gradient Boosting (nTree=100)	12.7675	3.5732	12
Linear LASSO (L1 λ=0.005)	15.1966	3.8983	38
Linear Ridge (L2 λ=0.005)	21.1825	4.6024	6