Evaluating proximity metrics for gene expression data: A hybrid model integrating data mining and machine learning techniques for disease diagnosis systems

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F44555601%3A13440%2F25%3A43899179" target="_blank" >RIV/44555601:13440/25:43899179 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.sciencedirect.com/science/article/pii/S1746809425006263" target="_blank" >https://www.sciencedirect.com/science/article/pii/S1746809425006263</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1016/j.bspc.2025.108115" target="_blank" >10.1016/j.bspc.2025.108115</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Evaluating proximity metrics for gene expression data: A hybrid model integrating data mining and machine learning techniques for disease diagnosis systems

Popis výsledku v původním jazyce

This study presents the development and application of a hybrid model for evaluating proximity metrics in high-dimensional gene expression data, integrating data mining and machine learning methods within a comprehensive framework. The research focuses on the comparative analysis of correlation distance, mutual information-based and Wasserstein metrics, assessing their effectiveness for clustering and classification tasks. In the initial modeling stage using gene expression data from over 6,000 patient samples covering 13 cancer types (TCGA dataset), the proposed model achieved classification accuracy exceeding 95.9% and a weighted F1-score above 95.8%. External validation using Alzheimer&apos;s (GSE174367) and Type 2 Diabetes (GSE81608) datasets confirmed the model&apos;s generalizability, with accuracy values reaching 96.28% and 97.43%, and weighted F1-scores of 96.26% and 97.41%, respectively. A stacking model was implemented to enhance classification robustness, compensating for potential clustering errors and delivering consistent performance across varying metrics and cluster structures. The proposed data processing pipeline ensures automated, standardized, and scalable analysis of large-scale gene expression datasets, aligning with the principles of personalized medicine.

Název v anglickém jazyce

Evaluating proximity metrics for gene expression data: A hybrid model integrating data mining and machine learning techniques for disease diagnosis systems

Popis výsledku anglicky

This study presents the development and application of a hybrid model for evaluating proximity metrics in high-dimensional gene expression data, integrating data mining and machine learning methods within a comprehensive framework. The research focuses on the comparative analysis of correlation distance, mutual information-based and Wasserstein metrics, assessing their effectiveness for clustering and classification tasks. In the initial modeling stage using gene expression data from over 6,000 patient samples covering 13 cancer types (TCGA dataset), the proposed model achieved classification accuracy exceeding 95.9% and a weighted F1-score above 95.8%. External validation using Alzheimer&apos;s (GSE174367) and Type 2 Diabetes (GSE81608) datasets confirmed the model&apos;s generalizability, with accuracy values reaching 96.28% and 97.43%, and weighted F1-scores of 96.26% and 97.41%, respectively. A stacking model was implemented to enhance classification robustness, compensating for potential clustering errors and delivering consistent performance across varying metrics and cluster structures. The proposed data processing pipeline ensures automated, standardized, and scalable analysis of large-scale gene expression datasets, aligning with the principles of personalized medicine.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Biomedical signal processing and control

ISSN

1746-8094

e-ISSN

1746-8108

Svazek periodika

2025

Číslo periodika v rámci svazku

110

Stát vydavatele periodika

GB - Spojené království Velké Británie a Severního Irska

Počet stran výsledku

17

Strana od-do

108-115

Kód UT WoS článku

001529361000002

EID výsledku v databázi Scopus

2-s2.0-105009419762