Kang Ka-Won, Choi Yumin, Lim Hyung-Jun, Kwak Kunye, Choi Yoon Seok, Park Yong, Kim Byung Soo, Lee Kwang-Sig, Ahn Ki Hoon
Division of Hematology-Oncology, Department of Internal Medicine, Korea University College of Medicine, Seoul, Republic of Korea.
School of Mechanical Engineering, Korea University, Seoul, Republic of Korea.
Sci Rep. 2025 Feb 12;15(1):5174. doi: 10.1038/s41598-025-89419-w.
The main treatment goal for immune thrombocytopenia (ITP) is bleeding risk reduction, particularly during procedures. While adjusting platelet thresholds with ITP treatments is recommended, platelet transfusions are commonly used despite controversial benefits. We evaluated the effectiveness of platelet transfusion in reducing post-procedure bleeding risk and identified predictive indicators of bleeding. A nationally representative database was used to develop a model predicting post-procedure bleeding risk in patients with ITP. Machine learning analyses, including random forest feature importance and Shapley additive explanations (SHAP) values, assessed 34 risk factors, including the platelet transfusion amount. The random forest model had an area under the receiver-operating characteristic curve of 93.6%. Key variables influencing bleeding risk included platelet transfusion amount, high-risk procedure, anticoagulant use, anemia, age, ITP treatment, and newly diagnosed ITP, all positively correlated with bleeding risk. Conversely, no antiplatelet or anticoagulant use and moderate- or low-risk procedures were negatively associated with bleeding risk. SHAP plots showed that platelet transfusion amount correlated with high-risk procedures, and bleeding risk increased with age in high-risk procedures. Bleeding risk in patients with ITP is primarily determined by procedural risk and patient condition, rather than platelet transfusion. Minimizing unnecessary platelet transfusions and addressing bleeding risk factors pre-procedure is crucial.
免疫性血小板减少症（ITP）的主要治疗目标是降低出血风险，尤其是在手术过程中。虽然建议通过ITP治疗来调整血小板阈值，但尽管血小板输注的益处存在争议，但仍普遍使用。我们评估了血小板输注在降低术后出血风险方面的有效性，并确定了出血的预测指标。使用一个具有全国代表性的数据库来建立一个预测ITP患者术后出血风险的模型。机器学习分析，包括随机森林特征重要性和夏普利值（SHAP），评估了34个风险因素，包括血小板输注量。随机森林模型的受试者操作特征曲线下面积为93.6%。影响出血风险的关键变量包括血小板输注量、高风险手术、抗凝剂使用、贫血、年龄、ITP治疗以及新诊断的ITP，所有这些都与出血风险呈正相关。相反，未使用抗血小板或抗凝剂以及中低风险手术与出血风险呈负相关。SHAP图显示血小板输注量与高风险手术相关，并且在高风险手术中出血风险随年龄增加。ITP患者的出血风险主要由手术风险和患者状况决定，而非血小板输注。尽量减少不必要的血小板输注并在术前解决出血风险因素至关重要。
