Scientific　elites　are　regarded　as　engines　that,　for　any　nation,　drive　scientific　and　technological　progress　and　social　development.　It　is　increasingly　important　to　explore　the　research　performance　of　scientific　elites.　Over　the　past　studies,　the　approaches　with　machine　learning　or　neural　networks　have　been　extensively　applied　in　the　task　of　time-series　data　prediction.　We　tried　to　use　Long　short-term　memory　(LSTM)　neural　network　for　predicting　the　productivity　and　influence　of　scientists　who　has　been　selected　as　academicians　of　American　Academy　of　Sciences　in　the　field　of　biology　during　period　of　2008　to　2015.　From　experimental　results,　we　found　that　forecasting　results　with　using　LSTM　neural　network　can　better　predict　the　research　performance　of　academicians.　Using　neural　network　approaches　to　predict　research　performance　is　an　important　attempt　in　the　field　of　informetric,　we　will　explore　and　compare　more　applications　of　neural　networks　in　research　performance　prediction　of　outstanding　scientists　in　the　future.　Furthermore,　we　will　further　study　the　impact　of　the　election　of　academicians　on　the　career　development　of　scientists.　©　2021　CEUR-WS.　All　rights　reserved.