Abstract: Background: Tumor recurrence is an abomination for hepatocellular carcinoma (HCC) patients receiving local treatment. Purpose: To build a hybrid machine learning model to recommend optimized first treatment (Laparoscopic hepatectomy (LH) or Microwave ablation (MWA)) for naïve single 3-5cm HCC patients based on early recurrence (ER, ≤2 years) probability. Methods: This retrospective study collected 20 semantic variables of 582 patients (LH:300, MWA:282) from 13 hospitals with at least 24 months follow-up. Both groups were divided into training, validation and test set, respectively. Five algorithms (Logistics Regression, Random Forest, Neural Network, Stochastic Gradient Boosting (SGB) and eXtreme Gradient Boosting (XGB)) were used for model building. Model with highest AUC in validation set of LH and MWA was selected to connect as a hybrid model which made decision based on ER probability. Model testing was performed in a comprehensive set composing of LH and MWA test set. Results: Four variables in each group were selected to build LH and MWA model, respectively. LH-XGB model (AUC=0.744) and MWA-SGM (AUC=0.750) model were selected for model building. In comprehensive set, a treatment confusion matrix was established based on recommended and actual treatment. The predicted ER probabilities were comparable with the actual ER rates for various types of patients in matrix (p>0.05). ER rate of patients whose actual treatment consistent with recommendation was lower than that of inconsistent patients (LH:21.2%vs46.2%, p=0.042; MWA:26.3%vs54.1%, p=0.048). By recommending optimal treatment, hybrid model can significantly reduce ER probability from 38.2% to 25.6% for overall patients (p<0.001). Conclusions: The hybrid model can accurately predict ER probability of different treatments, and thereby provide a reliable evidence to make optimal treatment decision for patients with single 3-5cm HCC.


中文翻译：

---

基于语义信息的混合机器学习模型可以优化单纯 3-5cm HCC 患者的治疗决策

摘要： 背景：对于接受局部治疗的肝细胞癌（HCC）患者来说，肿瘤复发是令人厌恶的。目的：建立一个混合机器学习模型，根据早期复发（ER，≤2 年）的概率推荐优化的首次治疗（腹腔镜肝切除术（LH）或微波消融术（MWA））用于初始单发 3-5cm HCC 患者。方法：这项回顾性研究收集了来自 13 家医院的 582 名患者（LH：300，MWA：282）的 20 个语义变量，并进行了至少 24 个月的随访。两组分别分为训练集、验证集和测试集。五种算法（逻辑回归、随机森林、神经网络、随机梯度提升（SGB）和极限梯度提升（XGB））用于模型构建。选择在 LH 和 MWA 的验证集中具有最高 AUC 的模型作为基于 ER 概率做出决策的混合模型进行连接。模型测试是在由 LH 和 MWA 测试集组成的综合集合中进行的。结果：每组选取4个变量分别建立LH和MWA模型。选择LH-XGB模型（AUC=0.744）和MWA-SGM（AUC=0.750）模型进行建模。在综合集合中，根据推荐治疗和实际治疗建立治疗混淆矩阵。预测的 ER 概率与矩阵中各种类型患者的实际 ER 率相当（p>0.05）。实际治疗与推荐一致的患者的 ER 率低于不一致的患者（LH：21.2%vs46.2%，p=0.042；MWA：26.3%vs54.1%，p=0.048）。通过推荐最佳治疗，混合模型可以将整体患者的 ER 概率从 38.2% 显着降低至 25.6% (p<0.001)。结论：混合模型能够准确预测不同治疗方案的ER概率，从而为单发3~5cm HCC患者做出最优治疗决策提供可靠依据。