Assessing the status of malaria transmission in endemic areas becomes increasingly challenging as countries approach elimination and infections become rare. Here, we evaluated the use of multiplex antibody response data to malaria-specific antigens to classify recent and historical infections of differentially exposed populations in three provinces in the Philippines. We utilized samples (n=9132) from health-facility based cross-sectional surveys in Palawan (ongoing malaria transmission), Occidental Mindoro (limited transmission), and Bataan (no transmission) and quantified antibody responses against 8 Plasmodium falciparum and 6 P. vivax-specific antigens. Different statistical and machine learning analytical methods were used to examine associations between antigen-specific antibody responses with malaria incidence, and the ability to predict recent or historical exposure. Consistent with the provinces′ endemicity status, antibody levels and seroprevalence were consistently highest in Palawan and lowest in Bataan. A machine learning (ML) approach (Random Forest model) using identified responses to 4 antigens (PfGLURP R2, Etramp5.Ag1, GEXP18 and PfMSP119) gave better predictions for P. falciparum infection (positive by microscopy, RDT and/or PCR) or likely recent exposure in Palawan (AUC: 0.9591, CI 0.9497-0.9684) than mixture models calculating seropositivity to individual antigens. Meanwhile, employing the same ML approaches for the vivax-specific antigens did not improve predictions for recent P. vivax infections. Still, the antigen panel was overall able to confirm the absence of recent exposure to P. falciparum and P. vivax in both Occidental Mindoro and Bataan through single and ensemble ML approaches. Seroprevalence and seroconversion rates based on cumulative exposure markers AMA1 and MSP119 showed accurate trends of historical P. falciparum and P. vivax transmission in the 3 sites. Our study emphasizes the utility of serological markers in predicting recent and historical exposure in a sub-national elimination setting, establishes baseline antibody data for monitoring risk in malaria-endemic areas in the Philippines, and also highlights the potential use of machine learning models using multiplex antibody responses to accurately assess the malaria transmission status of countries aiming for elimination.