An artificial neural network model to diagnose non-obstructive azoospermia based on RNA-binding protein-related genes

Abstract

Non-obstructive azoospermia (NOA) is a severe form of male infertility, but its pathological mechanisms and diagnostic biomarkers remain obscure. Since the dysregulation of RNA-binding proteins (RBPs) had nonnegligible effects on spermatogenesis, we aimed to investigate the functions and diagnosis values of RBPs in NOA. 58 testicular samples (control = 11, NOA = 47) from Gene Expression Omnibus (GEO) were set as the training cohort. Three public datasets, containing GSE45885 (control = 4, NOA = 27), GSE45887 (control = 4, NOA = 16), and GSE145467 (control = 10, NOA = 10), and 44 clinical samples from the local hospital (control = 27, NOA = 17) were used for validation. Through a series of bioinformatical analyses and machine learning algorithms, including genomic difference detection, protein-protein interaction network analysis, LASSO, SVM-RFE, and Boruta, DDX20 and NCBP2 were determined as significant predictors of NOA. Single-cell RNA sequencing of 432 testicular cell samples from NOA patients indicated that DDX20 and NCBP2 were associated with spermatogenesis (false discovery rate < 0.05). Based on the transcriptome expressions of DDX20 and NCBP2, we constructed multiple diagnosis models using logistic regression, random forest, and artificial neural network (ANN). The ANN model exhibited the most reliable predictive performance in the training cohort (AUC = 0.840), GSE45885 (AUC = 0.731), GSE45887 (AUC = 0.781), GSE145467 (AUC = 0.850), and local cohort (AUC = 0.623). Totally, an ANN diagnosis model based on RBP DDX20 and NCBP2 was developed and externally validated in NOA, functioning as a promising tool in clinical practice.

Predicting Male Infertility Using Artificial Neural Networks: A Review of the Literature

“…Therefore, to answer the PICO questions formulated in the previous section, we organized and schematically summarized them in the upcoming tables. Indeed, the studies were divided based on the general topic they dealt with; therefore, the specific variables considered in each model included sperm retrieval (Table 2, four studies [37][38][39][40]); sperm quality, further divided into the investigation of sperm quality and morphology (Table 3a, seventeen studies [41][42][43][44][45][46][47][48][49][50][51][52][53][54][55][56][57]) and quality of sperm and environmental factors (Table 3b, four studies [58][59][60][61]); non-obstructive azoospermia (Table 4, three studies [62][63][64]); IVF outcome (Table 5, three studies [65][66][67]); environmental and medical factors (Table 6, twelve studies [68][69][70][71][72][73][74][75]…”

“…Some studies [58,62,63,66] are actually comparative analyses, encompassing the utilization of diverse AI techniques, including ANNs, DT, and SVM, for the detection of male fertility. These methodologies demonstrated differing degrees of accuracy, emphasizing the significance of choosing the suitable algorithm for specific tasks.…”

“…These methodologies demonstrated differing degrees of accuracy, emphasizing the significance of choosing the suitable algorithm for specific tasks. ANNs, in association with the LR, have exhibited promise in the prediction of biochemical parameters associated with male infertility [63]. The adaptability of ANNs permits one to consider intricate relationships within the data, thereby facilitating precise predictions.…”

“…Figure2. Risk of Bias (RoB) analysis of retrospective and prospective studies performed using PROBAST[37][38][39][40][41][42][43][44]47,48,[54][55][56][57][58][59][60][61]63,64,71,72,[74][75][76][77]79].…”

Clinical management of nonobstructive azoospermia: An update

“…Combining these markers with known variables may allow for a better model to predict sperm retrieval. Moreover, based on bioinformatics analysis and machine learning algorithms, including genomic differences, protein–protein interactions, and network analysis, NOA predictors include DDX20 and NCBP2, and single‐cell RNA sequencing showed that these genes were significantly associated with spermatogenesis, leading to an artificial neural network diagnostic model 59 . Artificial intelligence technology is also expected to be applied to object detection of rare sperm identification in seminiferous tubules retrieved by micro‐TESE, in addition to biomarker identification 60 .…”