Skip to main content

Table 2 Reasons that cause the validation questions with correct answers listed in the system to lead to incorrect answers

From: Development of a novel drug information provision system for Kampo medicine using natural language processing technology

Classification

Reason for incorrect answer

Example of validation question

(Original text was Japanese)

System answer

Answers by

question authors

1

Problem with structure

of NLC

1–1

Misclassification by NLC®

(Category-sorting NLC® misclassification)

I want to know about patients to whom Yokukansan should be administered carefully.

抑肝散服用際、慎重に投与すべき患者について知りたい。

Carefully administer Kampo formulas to patients with cardiovascular disorders

(An answer in a different category NLC® with the answer by question author)

Careful administration of Yokukansan

Misclassification by NLC®

(Category NLC® misclassification)

In what patients is Shakuyakukanzoto contraindicated?

芍薬甘草湯禁忌はどんな患者さんですか?

Indication of Shakuyakukanzoto (An answer in the same category NLC® with the answer by question author)

Contraindications to Shakuyakukanzoto

1–2

Covering more than one category or answers (Concomitant drug)

Is it okay to take Hochuekkito and Shakuyakukanzoto together?

補中益気湯と芍薬甘草湯と飲み合わせは?

I do not have an answer. *

Drug Interactions of Shakuyakukanzoto

1–3

Question text structure

What side effects does Hochuekkito have other than pseudoaldosteronism?

補中益気湯は偽アルドステロン症以外に他にどような副作用がありますか?

Contents about pseudoaldosteronism of Hochuekkito

Adverse reactions of Hochuekkito

2

Problem with validation question

2 − 1

Ambiguous question

Can I use Yokukansan for patients with gastrointestinal upset?

胃腸調子が良くない患者さんに抑肝散を使ってもいいですか?

I do not have an answer. *

Careful administration of Yokukansan

2–2

Questions about content not mentioned in the package insert

Can Hochuekkito be used with patients taking antihypertensive drugs?

降圧剤を服用している患者さんに補中益気湯を併用してもよいですか?

Description of Hochuekkito

Drug Interactions of Hochuekkito

  1. When the threshold of confidence rate is set to 61%
  2. NLC® Natural Language classifier®