Developing a nomogram model for predicting non-obstructive azoospermia using machine learning techniques.

Journal: Scientific reports
PMID:

Abstract

Azoospermia, defined by the absence of sperm in the ejaculate, manifests as obstructive azoospermia (OA) or non-obstructive azoospermia (NOA). Reliable predictive models utilizing biomarkers could aid in clinical decision-making. This study included 352 azoospermia patients, with 152 diagnosed with OA and 200 with NOA. The data were randomly divided into a training set (244 cases) and a validation set (108 cases) for machine learning analysis. The training set was utilized for univariate and multivariate logistic regression to identify key predictors of NOA. Following this, nine machine learning. This study included 352 azoospermia patients, with 152 diagnosed with OA and 200 with NOA. The data were randomly divided into a training set (244 cases) and a validation set (108 cases) for machine learning analysis. The training set was utilized for univariate and multivariate logistic regression to identify key predictors of NOA. Following this, nine machine learning methods were employed to refine the prediction model. A novel nomogram model was developed, and its predictive performance was evaluated using receiver operating characteristic curves, calibration plots, and decision curve analysis. Univariate and multivariate logistic regression analyses identified semen pH and follicle-stimulating hormone (FSH) as positive predictors of NOA, while mean testicular volume (MTV) and inhibin B (INHB) were negatively correlated with NOA. Among nine machine learning methods evaluated, the Gradient Boosting Decision Trees achieved the highest performance with an area under the curve (AUC) of 0.974, whereas Random Forest showed the lowest AUC at 0.953. The nomogram model, incorporating these four factors, demonstrated robust predictive performance with AUCs of 0.984 in the training set and 0.976 in the validation set. Calibration and decision curve analysis confirmed the model's accuracy and clinical utility. Optimal cut-off points for biomarkers were identified: FSH at 7.50 IU/L (AUC = 0.96), INHB at 43.45 pg/ml (AUC = 0.95), MTV at 9.92 ml (AUC = 0.91), and semen pH at 6.95 (AUC = 0.71). The novel nomogram model incorporating FSH, INHB, MTV, and pH effectively predicts NOA in patients. This model offers a valuable tool for personalized diagnosis and management of azoospermia.

Authors

  • Hong Xiao
    Department of Computer and Information Science and Institute for Data Science, University of Mississippi, University, Mississippi 38677, United States.
  • Yi-Lang Ding
    Department of Andrology and Sexual Medicine, First Affiliated Hospital of Fujian Medical University, Fuzhou, 350005, China.
  • Chao Wang
    College of Agriculture, Shanxi Agricultural University, Taigu, Shanxi, China.
  • Peng Yang
  • Qiang Chen
    School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, Jiangsu, China.
  • Hao-Nan He
    Department of Andrology and Sexual Medicine, First Affiliated Hospital of Fujian Medical University, Fuzhou, 350005, China.
  • Ruijie Yao
    State Key Laboratory of Genetic Engineering, MOE Engineering Research Center of Gene Technology, School of Life Sciences, Fudan University, Shanghai 200438, China.
  • Hai-Lin Huang
    Department of Andrology and Sexual Medicine, First Affiliated Hospital of Fujian Medical University, Fuzhou, 350005, China.
  • Xi Chen
    Department of Critical care medicine, Shenzhen Hospital, Southern Medical University, Guangdong, Shenzhen, China.
  • Mao-Yuan Wang
    Department of Andrology and Sexual Medicine, First Affiliated Hospital of Fujian Medical University, Fuzhou, 350005, China.
  • Song-Xi Tang
    Department of Andrology and Sexual Medicine, First Affiliated Hospital of Fujian Medical University, Fuzhou, 350005, China.
  • Hui-Liang Zhou
    Department of Andrology and Sexual Medicine, First Affiliated Hospital of Fujian Medical University, Fuzhou, 350005, China. zhlpaper@fjmu.edu.cn.