Πλοήγηση ανά Συγγραφέας "Droumalia, Foteini"

Τώρα δείχνει 1 - 1 of 1
  • Μικρογραφία εικόνας
    Τεκμήριο
    Gene expression and gene regulatory network analysis with statistical methods and machine learning algorithms.
    (ΕΛ.ΜΕ.ΠΑ., ΣΧΟΛΗ ΜΗΧΑΝΙΚΩΝ (ΣΜΗΧ), Τμήμα Ηλεκτρολόγων Μηχανικών και Μηχανικών Υπολογιστών, 2023-04-10) Droumalia, Foteini; Δρουμαλιά, Φωτεινή
    Determining the best approach and data type for Pathway Analysis is a significant difficulty for the field of diagnostic medicine. The findings of recent studies indicate their preference for Machine Learning algorithms and the utilization of continuous gene expression values rather than binary values. However, due to the limitations of Machine Learning and non-binary values, there were efforts to produce new hybrid techniques that exploit the benefits of statistical methods and discrete values. The purpose of this study was to identify the most effective approach for Pathway Analysis utilizing currently available tools and to compare the results to previous research. This was accomplished by implementing the scoring methodology for several Pathway Analysis tools and employing a Decision Tree algorithm to assess the outcomes. The tools selected for implementation were TAPPA, PRS, TEAK, DEAP, GraphiteWeb, MinePath and HiPathia, among which PRS displayed the highest rate of accuracy, while HiPathia, which performed equally well, achieved the shortest execution time; overall, Machine Learning-based techniques outperformed those based on statistics. The outcomes obtained are consistent with prior literature, which have shown that non-binary data hold more information and that Machine Learning methods offer new opportunities for use in health. Unfortunately, a major issue that prevented our re-search from extracting common significant sub-pathways among the tools is probably related to the fact that there are numerous genomic platforms available; as a result genes cannot be recognized across different datasets.