Table 4 Micro-level comparison of triplet extraction results on representative Chinese clinical sentences.

From: Fine-tuned large language models with structured prompts enable efficient construction of lung cancer knowledge graphs

Challenge

Example sentence (Chinese)

ChatGLM baseline output

KGLM output

KGLM+Prompt output

Synonymy

患者确诊为慢性阻塞性肺疾病(慢阻肺), 目前接受支气管扩张剂治疗。

“The patient was diagnosed with chronic obstructive pulmonary disease (COPD) and is currently receiving bronchodilator therapy.”

[(慢性阻塞性肺疾病, 治疗, 支气管扩张剂), (慢阻肺, 治疗, 支气管扩张剂)]

[(Chronic Obstructive Pulmonary Disease, treatment, bronchodilator), (COPD, treatment, bronchodilator)]

(Same disease treated as two distinct entities)

[(慢阻肺, 治疗, 支气管扩张剂)]

[(COPD, treatment, bronchodilator)]

[(慢阻肺, 治疗, 支气管扩张剂)]

[(COPD, treatment, bronchodilator)]

(Proper focus on disease; no extraneous triples)

TCM terminology

患者诊断为肺癌, 采用清肺散结汤进行辅助治疗。

“Patient diagnosed with lung cancer, using Qingfei Sanjie Tang as adjunct therapy.”

[(肺癌, 治疗, 辅助治疗)]

[(Lung Cancer, treatment, Adjuvant therapy)]

(Misses the specific TCM therapy)

[(肺癌, 治疗, 清肺散结汤辅助治疗)]

[(Lung Cancer, treatment, Adjuvant therapy with Qingfei Sanjie Decoction)]

[(肺癌, 治疗, 清肺散结汤)]

[(Lung Cancer, treatment, Qingfei Sanjie Decoction)]

(Extracts the TCM formula name accurately)

Nested structure

患者确诊肺癌, 给予奥希替尼靶向治疗, 每日80mg。

“Patient diagnosed with lung cancer, given Osimertinib targeted therapy, 80 mg daily.”

[(肺癌, 治疗, 奥希替尼每日80mg)]

[(Lung Cancer, treatment, Osimertinib 80 mg daily)]

(Drug and dosage conflated in one triple)

[(肺癌, 治疗, 奥希替尼)]

[(Lung Cancer, treatment, Osimertinib)]

(Dosage omitted)

[(肺癌, 治疗, 奥希替尼), (奥希替尼, 剂量, 每日80mg)]

[(Lung Cancer, treatment, Osimertinib), (Osimertinib, dosage, 80 mg daily)]

(Properly separates drug and dosage into two triples)