An Early Subarachnoid Hemorrhage (SAH) Predictive Model Development And The Related Cell Signaling Pathways Analysis

Subarachnoid hemorrhage(SAH)represents that blood flows into subarachnoid space—the area between the arachnoid membrane and the pia mater surrounding the brain.As a clinical syndrome,because SAH is high mortality,high disability rate and poor clinical prognosis,it is always the focus of the basic and clinical medicine.In order to reduce the morbidity and mortality of SAH,many studies have been carried out on the pathogenesis of SAH.In recent years,studies have found that early brain injury(EBI)may be the main cause of poor prognosis in SAH patients.Therefore,the current popular SAH study is to explore such therapeutic drugs and effective targets that not only can reduce EBI after SAH and related complications,but also early predict and prevent SAH.Since current studies for SAH drugs select drugs from the perspective of clinical trials without considering the effect of related drugs from the perspective of cell signaling pathway,we put forward our first scientific question: can commonly clinical used LCN2 effectively intervene or treat SAH from the perspective of cell signaling pathway?On the other hand,because previous studies usually find the effective biomarkers for SAH prediction and treatment at a narrow molecular range,we put forward the second scientific question: could we choose potential biomarkers at genome-wide level for SAN by considering effect of LCN2 drug?Previously,we usually predict SAH by diagnostic imaging and clinical automation data,which results in inaccurate predictive power.Here,we put forward the third scientific question: could we use key genes to build classifiers for SAH early prediction,and improve predictive power by employing ensemble-learning model?Here,we employ following innovations to answer the above three scientific questions:(1)We use intervention experiments to screen out the candidate genes susceptible to SAH drug(LCN2).Then,we employ hypergeometric statistical analysis to choose the key signaling pathways from the candidates under different experimental conditions.(2)We integrate e-Bayes,SVMRFE,SPCA and statistical tests into experimental data to find the key genes by considering both SAH disease and LCN2 pharmacodynamics factors.(3)We integrate LR,SVM and Naive-Bayes algorithms into ensemble learning model to build an early SAH predictive platform.This study first use LCN2 intervention experiments to obtain the differential genes under various experimental conditions.Then,we employ hypergeometric statistical analysis to choose the signaling pathways from the candidates under different experimental conditions.Since the manually reviewed evidences demonstrate that several key signaling pathways stimulated by LCN2 could alleviate or promote SAH,we consider that LCN2 drug is able to intervene or treat SAH from the perspective of cell signaling pathway.Next,we use mathematical algorithms to choose five key genes(Tk1?Cyr61?Olig1?Pcolce2 and Slc6a9),which are sensitive to both SAH disease and LCN2 treatment.Finally,the five genes are employed as the classifiers to build a SAH early predictive ensemble-learning model,which outperforms classical LR,Naive-Bayes and SVM models.Thus,we consider that our study could provide new ideas and clue for the future study of clinical treatment of subarachnoid hemorrhage(SAH)and related diseases.