The　research　contribution　of　High-Performance　Computing　and　Applications　(HPCA),　combined　with　biomedical　text　related　to　generic　and　phenotypes,　has　been　published　at　various　levels　to　optimize　user　demand.　However,　there　is　still　a　huge　challenge　in　eliminating　ambiguities　that　occur　in　biomedical　text　mining　due　to　the　use　of　the　same　expression　in　the　investigation　of　the　genes　disease.　This　article　uses　two　combined　approaches　that　represent　identification　and　detection　in　high-performance　computing　and　Applications　(HPCA)　system　for　the　evaluation　of　gene-disease　and　phenotype　processes　in　biomedicine.　The　Current　approaches　use　machine　learning　as　a　positive　training　phrase　P　and　a　gene　as　the　negative　training　set　N　for　identification,　but　these　approaches　are　only　of　the　low-noise　negative　set　which　is　the　performance　of　the　quality　of　the　HPCA　System　affects,　our　first　contribution　paper　proposes　a　new　implementation　based　on　positive　unlabeled　Disease　gene　Identification　(PUDI)　to　construct　our　classifier.　The　second　contribution　we　introduce　high　precision　and　recall　values　which　achieve　term　classification　accuracy　between　0.96　and　0.99,　precision　values　between　0.95　and　0.99,　and　recall　values　between　0.96　and　0.99.　Definitively,　the　F-score　value　obtained　confirmed　outperformed　results　on　the　classifiers　with　high　accuracy　over　a　wide　range　of　parameters.　©　2019　IEEE.