The study examined the associations of serum sTREM-1 and sTREM-2 levels with mortality and neurologic prognosis in 120 patients resuscitated after cardiac arrest (including 32 survivors and 88 nonsurvivors), compared with 30 healthy volunteers. Both sTREM-1 and sTREM-2 levels increased after return of spontaneous circulation (ROSC) on days 1, 3, and 5, with greater increases in nonsurvivors. The primary objective was the prediction of 28-day all-cause mortality, the secondary 3-month neurological prognosis. 11 features including sTREM-1 and sTREM-2 were used to develop the machine learning models. The XGBoost and Random Forest models achieved the best results in predicting both mortality and neurological prognosis. sTREM-1 was more effective than sTREM-2 for predicting mortality and neurologic status after ROSC. These models have demonstrated higher accuracy than conventional clinical scoring systems.