Proceedings of the 4th National Big Data Health Science Conference

Table 1 (abstract P15). Performance of single-task models vs. stock MTL models vs. our MTL+TAL models

Tasks	Single-task models from literature		MTL model		MTL + TAL model
Tasks	AUC-ROC	F-Score	AUC-ROC	F-Score	AUC-ROC	F-Score
Morbidity:
SMM	0.7617 XGBoost	0.8071	0.8658	0.7993	0.8878	0.8186
Hemorrhage	0.8482 DNN	0.7897	0.8365	0.8177	0.8450	0.8285
Eclampsia	0.8110 RF	0.8273	0.8374	0.8350	0.8495	0.8442
Chronic:
Cardio. Disease	0.7113 DT	0.7193	0.7735	0.7408	0.7951	0.7542
Hypertension	0.8195 XGBoost	0.8352	0.8877	0.8561	0.9124	0.9237
Diabetes	0.7327 XGBoost	0.7391	0.8306	0.7985	0.8425	0.8716
Obesity	0.7421 RF	0.7394	0.8527	0.8294	0.8892	0.8853
Mental	0.8664 RF	0.8110	0.8704	0.8535	0.9039	0.8633
Substance Use	0.7845 LR	0.6585	0.5718	0.7210	0.7785	0.7109

ISSN: 1753-6561