Shu Jiong, Shao Jingyuan, Liu Lingling, Huang Xiang, Mao Yuxiang, Chen Huabao, Cui Xiangli, Li Bingbing, Jia Jie, Fei Zhenle, Hu Zongtao, Yang Xianjun, Chu Yannan, Wang Hongzhi
Hefei Cancer Hospital of CAS, Institute of Health and Medical Technology, Hefei Institutes of Physical Science, Chinese Academy of Sciences (CAS), Hefei, 230031, Anhui, China.
University of Science and Technology of China, Hefei, 230026, Anhui, China.
NPJ Digit Med. 2025 Mar 3;8(1):136. doi: 10.1038/s41746-025-01537-x.
In tumor radiotherapy, monitoring patient body position is crucial for improving efficacy and reducing complications. We developed a contact-based body position monitoring system using pressure sensors and artificial intelligence, enabling non-invasive, radiation-free, real-time monitoring. The system consists of two pressure-sensitive mattresses with 6400 piezoresistive pressure points each, placed under the scapulae and sacrococcygeal region to monitor resistance values for center of gravity calculation. Using data from 251 cancer patients across 1046 sessions, a random forest algorithm achieved an area under the curve (AUC) of 0.995. Internal validation revealed a true positive rate (TPR) of 95.5% and accuracy (ACC) of 96.8%. Overall accuracy exceeded 90%, providing an effective and low-cost solution for continuous body position monitoring during radiotherapy.
在肿瘤放射治疗中,监测患者体位对于提高疗效和减少并发症至关重要。我们开发了一种基于接触的体位监测系统,该系统使用压力传感器和人工智能,能够进行非侵入性、无辐射的实时监测。该系统由两个压敏床垫组成,每个床垫有6400个压阻压力点,分别放置在肩胛骨和骶尾区域下方,以监测电阻值用于计算重心。利用来自251名癌症患者1046个疗程的数据,随机森林算法的曲线下面积(AUC)达到0.995。内部验证显示真阳性率(TPR)为95.5%,准确率(ACC)为96.8%。总体准确率超过90%,为放疗期间的连续体位监测提供了一种有效且低成本的解决方案。