Feature　selection　is　one　of　the　popular　techniques　used　to　reduce　the　number　of　features　by　eliminating　noisy,　unreliable,　and　unnecessary　data　without　affecting　the　classification　accuracy.　Metaheuristic　algorithms　were　widely　incorporated　by　researchers　to　search　for　the　best　possible　features　in　simplifying　and　enhancing　dataset　feature.　This　is　because　the　traditional　optimization　techniques　have　drawback　of　suffering　from　entrapment　into　local　optima　when　handling　a　dataset　with　large　number　of　features.　In　this　study,　the　capability　of　African　vultures　optimization　algorithm　(AVOA)　in　conducting　feature　selection　on　medical　datasets　while　preserving　classification　accuracy　is　investigated.　Eight　medical　datasets　retrieved　from　UCI　machine　learning　repository　are　used　to　evaluation　performance　of　AVOA　in　feature　selection　and　compare　with　other　algorithms　known　as　opposition-based　differential　evolution　algorithms　(CO-DE),　particle　swarm　optimization　(PSO),　hybrid　canonical　differential　evolutionary　particle　swarm　optimization　(hC-DEEPSO),　and　multi-verse　optimizer　(MVO).　Comparative　study　reports　that　AVOA　produces　the　best　mean　accuracy　(i.e.,　82.9%)　in　six　out　of　eight　medical　datasets　and　lowest　number　of　features　(i.e.,　around　24　features)　in　four　out　of　eight　medical　datasets.　AVOA　can　outperform　other　competitive　algorithm　in　the　selected　medical　dataset　as　it　has　most　of　the　mean　accuracy　and　lowest　number　of　features.　©　2023,　The　Author(s),　under　exclusive　license　to　Springer　Nature　Singapore　Pte　Ltd.