Gradilla Hernández, Misael SebastiánFernández del Castillo Barrón, Alberto2025-01-042024-12-03Fernández del Castillo Barrón, A. (2024). Environmental assessment of urban rivers through a dual lens approach: machine learning based water quality analysis and metagenomic characterization of contamination effects (Tesis doctoral]. Instituto Tecnológico y de Estudios Superiores de Monterrey. Recuperado de: https://hdl.handle.net/11285/702967https://hdl.handle.net/11285/702967https://doi.org/10.60473/ritec.43https://orcid.org/0000-0002-8236-4400Urban rivers are critical ecosystems increasingly threatened by pollution. Effective water quality monitoring and contamination assessment are essential for informed management decisions. The Santiago River, a key hydrologic system in Mexico, has become one of the country’s most polluted rivers, posing significant ecological risks and public health concerns for nearby communities. This study underscores the urgent need for comprehensive environmental evaluation and enhanced monitoring approaches. Chapter one introduces the motivation behind monitoring water quality in highly polluted rivers, presenting the problem statement and contextual background of the Santiago River basin. It outlines the research question and provides an overview of the proposed dual-lens approach: combining water quality analysis via machine learning algorithms with metagenomic characterization of contamination effects. Key contributions of this work to the field are also highlighted. Chapter two reviews global monitoring strategies from highly polluted rivers, focusing on nine rivers across developed and developing countries to offer a comparative perspective on water quality management needs. In Chapter three, regression and classification machine learning models are developed to predict the Santiago River Water Quality Index (SR-WQI), designed as complementary tools to strengthen the current monitoring program. Chapter four analyzes the historical water quality patterns of the Santiago River to identify the most variable and representative data for training machine learning models. This chapter also reveals that redundant data can hinder model performance by leading to overfitting. Chapter five investigates spatial variations in the microbial composition of Santiago River sediments and examines correlations with water quality. Using high-throughput sequencing, potential microbial biomarkers were identified and impacts of physicochemical parameters and heavy metals on microbial communities were assessed. Finally, chapter five highlight the main findings of this thesis and covers some limitations, perspectives for future research and final remarks.TextoengopenAccesshttp://creativecommons.org/licenses/by-nc-nd/4.0INGENIERÍA Y TECNOLOGÍA::CIENCIAS TECNOLÓGICAS::INGENIERÍA Y TECNOLOGÍA DEL MEDIO AMBIENTE::CONTROL DE LA CONTAMINACIÓN DEL AGUATechnologyTesis Doctorado / doctoral ThesisContiene información que aún no ha sido publicada.https://orcid.org/0000-0003-4050-6797Water qualityUrban riverEnvironmental assesmentMachine learningMetagenomics966466