肠癌切除术复发预测标准数据集

本数据集专为结直肠癌术后复发风险的智能预测任务构建，旨在通过机器学习技术解决传统预后评估依赖单一TNM分期、缺乏个体化动态监测且预测精度受限的问题。数据采用高维结构化特征格式，适配目前主流临床预测模型的训练场景。
输入端（Features）： 结合人口信息、住院信息、诊断信息、病理信息、检验信息、随访信息等。输出端（Labels）： 严格基于患者长期随访结果判定的复发状态标签，涵盖复发的时间节点与结局。
数据源采用全量“临床真实数据”架构。所有数据均源自权威省级三甲肿瘤专科医院的脱敏病例库，未引入任何合成数据，最大程度保留了真实世界中复杂的医学特征分布。全量数据经由“病理专家复核+长期随访闭环验证”的严格质检，是训练高鲁棒性复发预测模型、辅助医生制定个体化辅助化疗与随访方案的高质量核心语料。

This dataset is specifically constructed for the intelligent prediction task of postoperative recurrence risk of colorectal cancer, aiming to address the limitations of traditional prognostic assessment—including sole reliance on TNM staging, lack of individualized dynamic monitoring, and restricted prediction accuracy—through machine learning techniques.
The dataset adopts a high-dimensional structured feature format, which is compatible with the training scenarios of current mainstream clinical prediction models.
Input End (Features): Combines demographic information, hospitalization information, diagnostic information, pathological information, laboratory test information, follow-up information and other relevant clinical data.
Output End (Labels): Recurrence status labels strictly determined based on the long-term follow-up results of patients, covering the recurrence time points and clinical outcomes.
The data source adopts a full-scale "real-world clinical data" architecture. All data are derived from the anonymized case databases of authoritative provincial-level tertiary tumor hospitals, without introducing any synthetic data, and maximally preserves the complex medical feature distribution in real-world clinical settings.
The full dataset has undergone strict quality inspection via "pathologist review + long-term follow-up closed-loop verification", making it a high-quality core corpus for training high-robustness recurrence prediction models and assisting clinicians in formulating individualized adjuvant chemotherapy and follow-up plans.

搜集汇总

背景与挑战

背景概述

该数据集专为结直肠癌术后复发风险的智能预测任务设计，旨在通过机器学习技术改进传统预后评估方法。它包含来自权威医院的多维度临床特征和基于长期随访的复发标签，经过严格质量控制，适用于训练高鲁棒性的预测模型。