%0 Journal Article %A Nikola Minic %A Svetlana Zunic %A Snezana Zunic %T NeuroStation – Statistical software based on artificial intelligence and pattern recognition for NSCLC development prediction through comprehensive biomarker analysis %D 2011 %J European Respiratory Journal %P p4437 %V 38 %N Suppl 55 %X A feed-forward artificial neural network is the perfect way to organize gathered information into precisely defined categories regardless blurred borders among them. Well-trained neural network with sufficient number of neurons in processing layer will act as an expert with highly increased learning and data processing skills and reduced error possibility below 0.001.NeuroStation statistical software is uniquely made for analysis of three biomarker categories: apoptotic, cytological and cytochemical parameters. Initial learning database contains 24 patients (9 non-smokers, 6 smokers and 9 NSCLC patients), plotted into 21-dimensional space neural network based on observed biomarkers. After thorough analysis and testing, the system offers a reliable prediction in percentages what is the probability of NSCLC development based on any collected biomarker for a new person.An analyst can further adjust statistical factors, such as influence ratio diameter and dispersion formula. Although the sample of 24 patients usually does not fulfill requirements for standard statistic tests, this network demonstrated significant results both in sense of reliability and mean squared error reduction.Software NeuroStation has remarkable potential in becoming essential tool both for in scientific purposes and for medical staff who work with NSCLC endangered patients. %U