背景 结直肠癌是全球最常见的三大肿瘤种类之一,到目前为止,对肿瘤复发早期预警是提高结直肠癌患者术后生 存率的重要课题。 当前,结直肠癌肿瘤患者的复发预测主要依靠临床医师根据患者的临床检查、血液检验及影像检查等临床 资料进行经验性判断,缺乏全面、可靠的客观依据,可能使患者错过最佳干预时机。 近几年出现的新型诊断方法费用相对昂 贵而难以普及;同时也是一种“事后汇报”模式,即肿瘤已在体内发展到一定程度才能检出。 方法 本研究是一项基于回顾性 队列数据的前瞻性研究项目,拟纳入来自电子科技大学附属医院· 四川省人民医院胃肠外科及急诊中心 2013 年 1 月至 2018 年 10 月的所有结直肠癌术后患者。 根据排除纳入标准筛选患者,拟收集纳入结肠癌患者术前与术后 5 年内的临床数据,将数 据进行清洗、预处理后分类与增强,最终采用数据驱动与机理互动的方式进行动力学建模,构建能够早期预测结直肠癌患者 5 年内肿瘤复发的监测与预测数字孪生模型。 结论 目前研究表明,结直肠癌术后 5 年的复发率高达 30%,目前临床急需对于结 直肠癌术后复发早期预警的可靠工具,数字孪生技术能够实现多维度数据处理,建立多模块模型预测,并建立具有时间序列 性的预测模型,适用于结直肠癌患者复发的预测研究。 本研究拟基于数字孪生技术,建立结直肠癌患者术后 5 年内肿瘤复发 的预测模型,以期实现对肿瘤复发的监测及早期预警,降低结直肠癌术后复发患者死亡率。

Background Colorectal cancer ranks among the top three most prevalent cancers worldwide posing a significant challenge despite surgical intervention. Detecting tumor recurrence early is paramount to improving the survival outcomes of patients post-colorectal cancer surgery. Presently recurrence prediction relies heavily on clinicians' subjective judgment drawing from clinical examination blood tests imaging and other clinical data. However this method lacks objective reliability and risks patients missing optimal intervention opportunities. While new diagnostic methods have emerged they are often costly and challenging to implement universally. Moreover they typically operate reactively detecting tumors only after significant in vivo development serving as a 􀆴 debriefing􀆶 mode. Method This study is based on retrospective cohort data collected from patients with colorectal cancer admitted to the Department of Gastrointestinal Surgery and Emergency Surgery at Sichuan Provincial People 's Hospital Affiliated Hospital of University of Electronic Science and Technology of China between January 2013 and October 2018. Stringent exclusion criteria will be applied during patient selection and clinical data from colon cancer patients are gathered spanning five years before and after surgery. The collected data will undergo thorough cleaning preprocessing classification and enrichment processes to create an AI-ready dataset. Leveraging a combination of data-driven and mechanistic modeling approaches we aim to develop a digital twin model capable of monitoring and predicting colorectal cancer recurrence within a five-year timeframe. Conclusion Current studies have shown that the recurrence rate of colorectal cancer five years after surgery is as high as 30% and a reliable tool for early warning of postoperative recurrence of colorectal cancer is urgently needed in clinic. Digital twin technology can realize multi-dimensional data processing establish multi -module model prediction and establish a prediction model with time series which is suitable for the prediction of colorectal cancer patients' recurrence. Based on digital twin technology this study intends to establish a prediction model of tumor recurrence in patients with colorectal cancer within 5 years after surgery in order to achieve the monitoring and early warning of tumor recurrence and reduce the mortality of patients with colorectal cancer recurrence after surgery.