PROBLEM TO BE SOLVED: To provide a method for highly accurately predicting an occurrence time of an event, such as development of a specific disease, death, or the like, by using a huge volume of analysis results of clinical tests.SOLUTION: The event occurrence time prediction device comprises: input means to which one or a plurality of types of first analysis results obtained from a first analysis object are input and processing means that executes a step of calculating a score indicating a possibility of occurrence of an event in the first analysis object within a predetermined period from acquisition of the first analysis results, on the basis of the first analysis results input in the input means. The score is obtained through machine learning that uses one or a plurality of types of second analysis results obtained from a second analysis object for which occurrence of an event is known, before the occurrence of the event, and a lapse time from acquisition of the second analysis results to occurrence of the event in the second analysis object, and uses, as teacher data, whether or not the time of the acquisition of the second analysis results and the time of the occurrence of the event exist in the predetermined period.SELECTED DRAWING: NoneCOPYRIGHT: (C)2017,JPO&INPIT【課題】、膨大な量の臨床検査の解析結果を活用し、特定の疾患の発症や死亡等のイベントの発生時期を高い精度で予測する手法を提供する。【解決手段】第1解析対象から得られた1種類又は複数種類の第1解析結果が入力される入力手段と、入力手段に入力された第1解析結果に基づき、第1解析結果の取得時から所定の期間内に、第1解析対象においてイベントが発生する可能性を示すスコアを算出するステップを実行する処理手段と、を少なくとも備え、スコアは、イベントが発生したことが既知の第2解析対象から、イベント発生前に得られた1種類又は複数種類の第2解析結果と、第2解析結果の取得時から第2解析対象におけるイベント発生時までの経過時間と、を用いて、第2解析結果の取得時とイベント発生時とが所定の期間内に存在するか否かを教師データとした機械学習により得られる。【選択図】なし
