Ciencias Exactas y Ciencias de la Salud

Permanent URI for this collectionhttps://hdl.handle.net/11285/551039

Pertenecen a esta colección Tesis y Trabajos de grado de las Maestrías correspondientes a las Escuelas de Ingeniería y Ciencias así como a Medicina y Ciencias de la Salud.

Browse

Search Results

Now showing 1 - 10 of 25

Deep learning framework to predict and generate new fluorescent molecules from experimental data
(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2025-07) Azizi, Mina; Aguirre Soto, Héctor Alán; emipsanchez; Ray, Mallar; Bernal Neira, David Esteban; Mendoza Cortés, José Luis; School of Engineering and Sciences; Campus Monterrey; Flores Tlacuahuac, Antonio
Fluorescent molecules play important roles in biological imaging, diagnostics, and materials science. However, identifying efficient and effective fluorophores remains challenging, as traditional trial-and-error experimentation and in silico computations are both costly and time-consuming. To address this, this thesis presents a deep learn- ing approach to streamline the discovery process by predicting optical properties and generating novel fluorescent molecules directly from experimental data. The study is based on FluoDB, a publicly available dataset collected from the literature, containing over 55,000 fluorophore–solvent pairs with experimentally measured optical prop- erties. Graph Convolutional Network (GCN) models were trained to predict four key optical properties and effec- tively captured complex structure–property relationships, achieving R² values ranging from 0.49 to 0.87 across the different targets. A Conditional Variational Autoencoder (CVAE) was also implemented to generate novel fluores- cent molecules based on solvent identity and target absorption range. In total, 2573 valid and structurally diverse molecules were generated, with a variety of predicted optical behaviors. Together, the predictive model and genera- tive models provide a useful and data-driven approach to accelerate exploration and design of functional fluorescent materials.
Environmental assessment of urban rivers through a dual lens approach: machine learning based water quality analysis and metagenomic characterization of contamination effects
(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2024-12-03) Fernández del Castillo Barrón, Alberto; Gradilla Hernández, Misael Sebastián; emipsanchez; García González, Alejandro; Pacheco Moscoa, Adriana; Brown, Lee; Oscar Alejandro Aguilar Jiménez; School of Engineering and Sciences; Campus Monterrey; Senés Guerrero, Carolina
Urban 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.
Explainable AI for trading 50 consumer discretionary stocks in the S&P 500
(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2024-12) Sanromán Iñiguez, Paulina Monserrat; Mendoza Montoya Omar; emipsanchez; Antelis Ortiz, Javier Mauricio; Guizar Mateos, Isaí; School of Engineering and Sciences; Sede EGADE Monterrey; Bernal Ponce, Luis Arturo
This document presents a study that merges computer science techniques with finance, focusing on the development of an Explainable Supervised Machine Learning (SML) model aimed at achieving a balance between predictive accuracy and interpretability in price forecasting for Algorithmic Trading (AT). Utilizing SHAP (SHapley Additive exPlanations), both global explanations are provided to facilitate feature selection and determine the importance of various macroeconomic and technical indicators derived from historical data of 50 companies within the Consumer Discretionary sector of the S&P 500 Index. The study also employs hyperparameter tuning on lagged values to assess whether the price movements from one day can effectively predict subsequent market prices. Algorithmic Trading (AT) currently constitutes approximately 60% to 75% of total trading activity in U.S. equity markets, European financial markets, and major Asian capital markets (Groette, 2024). Projections indicate a significant growth trajectory for this sector. The driving force behind this expansion is the advancement of Artificial Intelligence (AI). As AI models incorporate more data, they tend to become increasingly intricate and opaque, evolving into what are commonly referred to as black box models. This complexity raises critical concerns surrounding explainability, interpretability, and transparency, as well as adherence to regulatory standards. Neglecting these issues can lead to severe market disruptions, including panic selling, liquidity evaporation, increased asset correlations, and a lack of clarity regarding the decision-making processes of AI models. Such challenges underscore the imperative for developing transparent and interpretable AI solutions in AT to mitigate risks and enhance market stability.
Modeling of carbon sequestration and productivity for maize and oats crops using artificial neural network
(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2024-11-25) Aguilar Chavez, Fernanda; Valiente-Banuet, Juan Ignacio; emipsanchez; Clarke Crespo, Emilio; González Viejo, Claudia; School of Engineering and Sciences; Campus Querétaro
Climate change presents a critical challenge to global food security, especially as the global population continues to rise. A major driver of this phenomenon is the accumulation of greenhouse gases, particularly CO₂, which intensifies Earth's warming. Key contributors to elevated CO₂ levels include fossil fuel combustion and agricultural activities. However, agricultural systems have the potential to mitigate this effect by capturing atmospheric CO₂. Notably, few models account for the net CO₂ flux in agricultural systems, which is critical for understanding their true carbon sequestration potential. This study introduces a machine learning-based approach to model CO₂ sequestration and productivity in two forage crops, a variety of maize (Zea mays) and oats (Avena sativa), under diverse environmental conditions. The model leverages critical variables such as degree days, NDVI, and water balance. Using an artificial neural network (ANN), the study achieved robust predictive accuracy for both crops, with determination coefficients (R) of 0.95 for maize and 0.96 for oats, and low mean squared errors (MSE = 0.02). These results highlight the model’s high performance and reliability, offering a valuable tool for predicting carbon sequestration and productivity in forage crops while addressing a key gap in net CO₂ flux modeling.
The role of capitalization and character repetition in identifying depression on social Media: a bilingual approach
(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2024-11-24) Burgueño Paz, Luis Humberto; Zareei, Mahdi; emipsanchez; Roshan Biswal, Rajesh; School of Engineering and Sciences; Campus Monterrey; García Ceja, Enrique Alejandro
Depression is a mental disorder that affects millions of people worldwide, but a significant portion of the affected people don’t receive adequate treatment. There has been an increasing interest from researchers to detect this condition through social media posts in order to prompt for early treatment. However, most of the research has been focused on the Caucasian Western English-speaking population, limiting the applicability of their findings across diverse cultural contexts. While research has shown the use of nonverbal cues to convey sentiment, their role on depression detection remains under-explored. This thesis aims to assess the effect of nonverbal cues, specifically capitalization and character repetition, on depression detection using datasets both in English and Spanish. This effect was explored through three existing datasets. The first dataset included a collection of Reddit posts and comments in the English language and was selected to assess the effect on a dataset coming from one of the most reputable mental health competitions in Natural Language Processing. The second dataset consisted of a collection of Spanish- language messages from Telegram to verify whether findings in the English language would hold for Spanish. The third dataset, also built from Reddit posts, was used to analyze the impact of these features when classifying by depression severity levels rather than binary labels. Four classifiers were used throughout this research: Logistic Regression, Random Forest, Support Vector Machine, and Neural Network. Overall, the impact of capitalization and character repetition for depression detection was found to be minimal. These features had the most effect on English Reddit data with binary labels, while showing limited impact on Spanish data or when classifying by severity levels. Additionally, models using only character repetition outperformed those relying on capitalization features.
Mechanical characterization and design of square honeycombs with the aid of additive manufacturing and AI
(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2024-05-07) Herrera Ramos, Gustavo; Cuan Urquizo, Enrique; emimmayorquin; Román Flores, Armando; Mora Córdova, Ángel; Escuela de Ingeniería y Ciencias; Campus Monterrey; Batres Prieto, Rafael
Metamaterials offer a viable mean to attain targeted mechanical characteristics tailored to particular loading conditions. Aperiodic metamaterials provide higher tailorability of mechanical behavior by providing a customizable deformation mode, properties, and mechanical response. Artificial intelligence has enhanced metamaterial design by discerning correlations between parameters and mechanical characteristics. This work studies two types of gradation on square honeycombs: wall thickness and wall angle. The studied gradation characteristics were wall inclination, pattern distribution, and direction. Fifteen designs were proposed, each combining different gradation characteristics. The designs were additively manufactured with PLA on an FFF 3D printer and experimentally tested under compression. The effects of the gradation characteristics on the mechanical response, mechanical properties, and deformation mode were analyzed. The results confirmed the influence of gradation on the mechanical behavior of the structures. The gradation characteristics influence specific properties or responses, such as a 30% energy absorption difference between graded honeycombs with aligned and not aligned walls. The metamodel evolutionary optimizer (MEVO) algorithm was used to assist in the design of a tailored square honeycomb with an angle gradation to minimize the displacement of a designated point in the structure. The algorithm was tested on multiple nonconventional loading scenarios to prove its versatility.
Machine learning-guided production of a nanoemulsion for delivery of anacardic acid
(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2024) Prieto Medrano, Cassandra Guadalupe; Perfecto Avalos, Yocanxóchitl; emimmayorquin; Sánchez Ante, Gildardo; Sánchez López, Angélica Lizeth; Zavala Martínez, Araceli; School of Engineering and Sciences; Campus Guadalajara; García Varela, Rebeca
Bioactive molecules from plants remain an important source of drug candidates for many diseases. However, such molecules have poor in vivo performance due to low water solubility leading to inadequate distribution. Oil-in-water nanoemulsion drug delivery systems can help counteract these limitations by improving drug distribution even into highly resistant tissues. A limitation preventing the widespread adoption of nanoemulsion drug delivery systems is the expensive, time-consuming development process by trial and error. In this study, we developed a nanoemulsion design strategy guided by machine learning. We retrieved and aggregated data such as average particle size and polydispersity index associated to nanoemulsion composition to construct a comprehensive dataset from available literature. A predictive machine learning model was used to identify improved self-nanoemulsifying system formulations, including olive oil as oily base and combinations of Tween 20, Tween 80, glycerol, and soy lecithin. The predictive power of the model was assessed by estimating the successful self-nanoemulsification through transmittance, and later confirmed by analyzing the formulations using Dynamic Light Scattering. As an experimental model, the nanoemulsions were loaded with an organic extract from Amphipterygium adstringens (a plant native to Mexico known as cuachalalate) containing anacardic acid. Encapsulation efficiency was measured by UHPLC, and the antiproliferative activity of the preparations was evaluated on HEPG2, a human hepatic cancer cell line, and HEK-293, a normal-like human embryonic kidney cell line. The machine learning model was able to accurately predict a successful formulation 81% of the time. The best-performing formulation, a combination of 10% olive oil, 60% Tween 20, and 30% glycerol, exhibited average particle size of 162.8±26 nm, with a polydispersity index of 0.234±0.03, and full encapsulation efficiency given the assay used. The naked nanoemulsion presented no toxicity in the normal-like cell line but exerted an inhibitory effect on the cancer cell line. Moreover, loading the plant extract into the selected formulation increased the cytotoxic effect on the cancer cell line in comparison to the naked nanoemulsion, the extract alone, and pure anacardic acid alone, yielding an IC50 value of 5.9±1.27 µM. These results suggest that the formulation identified by the model was a successful carrier of the plant extract and molecule of interest. This study presents a proof of concept on how artificial intelligence can reduce the development pipeline of nanoemulsified drug delivery systems.
Facilitating early detection of depression through conversational audios and machine learning techniques
(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2023-06-21) Noriega Quirós, Isabella; Trejo Rodriguez, Luis Ángel; puemcuervo, emipsanchez; González Mendoza, Miguel; Brena Pinero, Ramón Felipe; Figueroa López, Carlos Gonzalo; Escuela de Ingeniería y Ciencias; Campus Monterrey
Mental health is becoming a trending topic amongst society. The relevance of it in our lives is being studied in order to achieve a better comprehension for our well-being. Studies have shown that both anxiety and depression greatly affect higher education student’s performance and development, as well as post-graduate life. Early detection of depression, or other mental health issues, could lead to sooner evaluation and support. As humans go through life, many stressful situations arise. This is not possible to avoid. Nevertheless, our resilience to stress is the factor that estimates how much stress we can handle until reaching alerting levels of a possible mental disorder. This research intends to use machine learning techniques to deliver an accurate classification from depressive indicators based on conversational audios. The result provided will be used by an algorithm to analyze the individual’s state, and with the combination of conversational audios and the psychophysiological profile, it will identify early symptoms of the illness, which will alert the individual in time to act.
Aplicación de modelos de aprendizaje automático para la predicción de eventos adversos secundarios a vacunas COVID-19 y detección de factores importantes
(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2023-05) Molina Puentes, Mayra Alejandra; Falcón Morales, Luis Eduardo; emimmayorquin; Roshan Biswal, Rajesh; Sánchez Ante, Gildardo; Escuela de Ingeniería y Ciencias (EIC); Campus Guadalajara
COVID-19 disease caused by the SARS-CoV-2 virus was the third leading cause of death in the United States for 2022, claiming more than six million lives worldwide since the outbreak began in 2019. COVID-19 vaccines have been available for more than two years, yet vaccine hesitancy still prevails to this day. One of the factors to vaccine hesitancy is the concern for vaccine safety and adverse reactions. The main objective of this work is to apply machine learning algorithms and develop a model to predict if an individual will have a serious adverse reaction or death based on patient information, medical history, and vaccine information. Additionally, through the application of feature importance techniques this study aims to identify potential risk factors for serious adverse reactions. Six machine learning algorithms were chosen for this study: Logistic Regression, Decision trees, Support Vector Machines (SVM), Naïve Bayes (NB), Random Forest (RF), K-Nearest Neighbors (kNN). The best result was achieved through Random Forest to predict a lethal adverse event post vaccination with an accuracy of 91.37%. Decision Tree provides an accuracy of 64.83% when predicting a severe adverse event. Age, gender, vaccine manufacturer and vaccine dose contribute the most to serious adverse reactions while age, gender and symptoms contribute the most towards patient death.
Machine learning detection of severity level of maladaptive plasticity in tinnitus and neuropathic pain
(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2023) González Sánchez, Andrea; Alonso Valerdi, Luz María; emimmayorquin; Ibarra Zarate, David Isaac; Román Godínez, Israel; Tamez Peña, José Gerardo; Montemayor Zolezzi, Daniela; Escuela de Ingeniería y Ciencias; Campus Monterrey
Tinnitus and NP datasets were analyzed separately and in conjunction, under the hypothesis that they share an underlying mechanism of maladaptive plasticity. Linear and non-linear features were extracted from the EEG signal data, including power spectral density, Shannon entropy and imaginary coherence between channel data. Feature selection with BorutaSHAP wrapper method was applied as part of the model construction process. Classification was performed using both traditional ML algorithms, Random Forest, Support Vector Machine, and k-Nearest Neighbors, and then implementing a prediction voting mechanism, as well as the deep learning neural network EEGNet. In general, SVM had the best performance across experiments. For the tinnitus dataset, the ensemble classifier had the highest accuracy of 50.08%. For the NP dataset, SVM had the best performance measured through an accuracy of 58.43%. For the BC dataset, the highest accuracy score of 42.46% was obtained by SVM. For the EEGNet implementation, the average accuracy obtained in the tinnitus dataset was 51%, the NP dataset was 97.92%, and the BC dataset was 74.31%. EEGNet was the best performing model, particularly for the NP dataset. The top selected features of the feature selection algorithm suggests gamma as a potential biomarker for the detection of maladaptive plasticity.

Ciencias Exactas y Ciencias de la Salud

Browse

Filters

Settings

Sort By

Results per page

Search Results