Study on Enrichment and Classification of Cancer-Related Fatigue Patients with Qi and Blood Deficiency Syndrome Based on Decision Tree Algorithm
Author：SHI Jiyan1,2, GU Shanshan2, ZENG Yumei2, LIANG Weijie3, YI Danhui3, XU Yun2
Unit：1.Graduate School, Beijing University of Chinese Medicine, Beijing 100029, China; 2.Xiyuan Hospital, China Academy of Chinese Medical Sciences, Beijing 100091, China; 3.School of Statistics, Renmin University of China, Beijing 100872, China
Quote：引用：史纪言，谷珊珊，曾钰梅，梁玮杰，易丹辉，许云.基于决策树的癌因性疲乏气血两虚证患者富集分类研究[J].中医药导报,2025,31(9):117-121,145.
DOI：10.13862/j.cn43-1446/r.2025.09.020
PDF：
Download PDF
Abstract：Objective: To construct a classification prediction
model for cancer-related fatigue (CRF) patients with qi and blood deficiency
syndrome. Methods: CRF patients with qi and blood deficiency syndrome from 29
sub-centers nationwide (October 2019 to April 2022) were included. A population
classification prediction model was constructed based on the decision tree
algorithm. Results: Data from 591 patients were incorporated into the model,
including the mean score of the Piper Fatigue Scale (PFS), blood routine
examination indicators, and TCM diagnostic symptoms for qi and blood deficiency
syndrome. Red blood cell (RBC) count and white blood cell (WBC) count were
identified as key predictors for population classification. The model's
prediction accuracy reached 82.49%. Relying on a single type of variable was
found unsuitable for constructing an effective classification model using the
decision tree algorithm. Conclusion: The decision tree-based classification
prediction model for CRF patients with qi and blood deficiency syndrome
demonstrates relatively good performance by combining macro (symptom) and micro
(laboratory) indicators for effective population enrichment and classification.
Key words：cancer-related fatigue (CRF); Qi and blood deficiency syndrome; decision tree; predictive model; information enrichment
摘要：目的：构建癌因性疲乏（CRF）气血两虚证人群的分类预测模型。方法：以2019年10月至2022年4月期间全国29家分中心符合纳入标准的CRF气血两虚证患者为研究对象，基于决策树算法构建人群分类预测模型。结果：建模共纳入591例患者信息，将Piper疲乏调查量表（PFS）总均分、血常规指标、CRF气血两虚证诊断症状纳入决策树模型，其中外周血红细胞计数（RBC）、外周血白细胞计数（WBC）水平对人群划分起主要决定作用，经测试模型预测准确率为82.49%；仅纳入单一变量不适合使用决策树算法构建CRF气血两虚证人群分类预测模型。结论：基于决策树算法构建的CRF气血两虚证人群分类预测模型，结合宏观、微观指标进行人群富集分类，具有较为良好的性能。
关键词：癌因性疲乏；气血两虚证；决策树；预测模型；信息富集
Release time：2026-01-08
click-through rate：167