Neural network and spline-based regression for the prediction of salivary hypofunction in patients undergoing radiation therapy.
Neural network and spline-based regression for the prediction of salivary hypofunction in patients undergoing radiation therapy.
复制标题
DOI：
10.1186/s13014-023-02274-9
复制
发表时间：
2023-05-08
期刊：
影响因子：
--
通讯作者：
中科院分区：
文献类型：
--
作者：
这项研究利用了大量的头颈部癌症患者的回溯性队列，以开发机器学习模型，以根据腮腺的剂量-体积直方图来预测辐射引起的肝细胞减退。用510例头颈癌患者放疗前后的唾液流率对唾液功能低下的三种预测模型进行拟合：(1)Lyman-Kutcher-Burman(LKB)模型；(2)基于Spline的模型；(3)神经网络。第四个使用文献报道的参数值的LKB型模型被包括以供参考。使用截断相关的AUC分析来评估预测性能。神经网络模型在LKB模型中占主导地位，在每个截止点表现出更好的预测性能，根据所选择的截止点，AUC从0.75到0.83不等。基于样条线的模型几乎主导了LKB模型，拟合的LKB模型只有在0.55的临界值时表现得更好。样条线模型的AUC范围在0.75到0.84之间，具体取决于所选的截止值。LKB模型的预测能力最低，AUC的范围分别为0.70~0.80(拟合)和0.67~0.77(文献报道)。我们的神经网络模型比LKB和其他机器学习方法表现出更好的性能，并提供了临床上有用的唾液功能低下预测，而不依赖于总结测量。
This study leverages a large retrospective cohort of head and neck cancer patients in order to develop machine learning models to predict radiation induced hyposalivation from dose-volume histograms of the parotid glands. The pre and post-radiotherapy salivary flow rates of 510 head and neck cancer patients were used to fit three predictive models of salivary hypofunction, (1) the Lyman-Kutcher-Burman (LKB) model, (2) a spline-based model, (3) a neural network. A fourth LKB-type model using literature reported parameter values was included for reference. Predictive performance was evaluated using a cut-off dependent AUC analysis. The neural network model dominated the LKB models demonstrating better predictive performance at every cutoff with AUCs ranging from 0.75 to 0.83 depending on the cutoff selected. The spline-based model nearly dominated the LKB models with the fitted LKB model only performing better at the 0.55 cutoff. The AUCs for the spline model ranged from 0.75 to 0.84 depending on the cutoff chosen. The LKB models had the lowest predictive ability with AUCs ranging from 0.70 to 0.80 (fitted) and 0.67 to 0.77 (literature reported). Our neural network model showed improved performance over the LKB and alternative machine learning approaches and provided clinically useful predictions of salivary hypofunction without relying on summary measures.
登录
查看更多内容
DOI：
10.1016/s0360-3016(01)01717-5
发表时间：
2001-11-15
影响因子：
7
作者：
Roesink, JM;Moerland, MA;Terhaard, CHJ
通讯作者：
Terhaard, CHJ
影响因子：
17.1
作者：
van Luijk P;Pringle S;Deasy JO;Moiseenko VV;Faber H;Hovan A;Baanstra M;van der Laan HP;Kierkels RG;van der Schaaf A;Witjes MJ;Schippers JM;Brandenburg S;Langendijk JA;Wu J;Coppes RP
通讯作者：
Coppes RP
DOI：
10.1016/j.ijrobp.2009.09.040
发表时间：
2010-03-01
影响因子：
7
作者：
Bentzen, Soren M.;Constine, Louis S.;Deasy, Joseph O.;Eisbruch, Avi;Jackson, Andrew;Marks, Lawrence B.;Ten Haken, Randall K.;Yorke, Ellen D.
通讯作者：
Yorke, Ellen D.
DOI：
10.1002/hed.20735
发表时间：
2008-04-01
影响因子：
2.9
作者：
Nordgren, Mats;Hammerlid, Eva;Jannert, Magnus
通讯作者：
Jannert, Magnus
DOI：
10.1016/j.ijrobp.2018.06.048
发表时间：
2018-11-15
期刊：
International journal of radiation oncology, biology, physics
影响因子：
--
作者：
Rosen BS;Hawkins PG;Polan DF;Balter JM;Brock KK;Kamp JD;Lockhart CM;Eisbruch A;Mierzwa ML;Ten Haken RK;El Naqa I
通讯作者：
El Naqa I