MEDICAL DRUG ADVERSE EVENT EXTRACTION METHOD AND DEVICE

PROBLEM TO BE SOLVED: To provide a medical drug adverse event extraction device that can accurately extract a combination of a medical drug and an adverse event related to the medical drug.SOLUTION: A medical drug adverse event extraction method is configured to: extract a medical event from medical...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: KUMANO AIKO, ENDO AYUMI, YAMADA KAORI, KOMAMINE MAKI, MORINAGA SATOSHI, NARITA KAZUYO, KOSAKA YUKI
Format: Patent
Sprache:eng ; jpn
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:PROBLEM TO BE SOLVED: To provide a medical drug adverse event extraction device that can accurately extract a combination of a medical drug and an adverse event related to the medical drug.SOLUTION: A medical drug adverse event extraction method is configured to: extract a medical event from medical information data relating to a patient for each of a combination of positive examples, a combination of negative examples, and a combination of non-positive nor negative examples, which are a combination of a medical drug and an accident/sickness; generate attribute data on the basis of time-series information on the adverse event; learn a determination model on the basis of attribute data on the positive example and negative example; and input attribute data corresponding to the combination of the non-positive nor negative examples to the determination model to obtain a score. As the medical event, the medical event is configured to include at least one of a medical practice carried out to the patient including the accident/sickness observed in the patient and an event indicating that the medical practice was carried out accompanied by the medical practice. 【課題】医薬品とその医薬品に関連した有害事象との組み合わせを精度よく抽出できる医薬品有害事象抽出装置を提供する。【解決手段】医薬品と傷病の組み合わせである、正例の組み合わせ、負例の組み合わせ、及び正例でも負例でもない組み合わせの各々ごとに、患者に関する医療情報データから医療イベントを抽出し、医療イベントの時系列情報に基づいて属性データを生成する。正例及び負例の属性データに基づいて判別モデルを学習し、正例でも負例でもない組み合わせに対応する属性データを判別モデルに入力してスコアを求める。医療イベントとして、当患者において観察された傷病を含むとともに、患者に対して行われた医療行為及びその医療行為に付随してその医療行為が行われたことを示すイベントの少なくとも一方を含んでいるものを用いる。【選択図】図1