Conference Agenda

Monitoring water quality remains a persistent challenge in rural communities of emerging countries, where traditional methods are often costly, slow, and difficult to implement. In recent years, the integration of Internet of Things (IoT) technologies and Machine Learning (ML) techniques has emerged as an efficient alternative to improve the accuracy, continuity, and predictive capacity of water monitoring systems. This Systematic Literature Review aimed to identify and analyze the approaches, parameters, technological platforms, and models used in studies that combine IoT and ML for water quality monitoring in rural settings. A structured search was conducted exclusively in the Scopus database, resulting in the selection of 26 primary studies, which were analyzed using a PICOC-based extraction matrix. The results show that most investigations focus on rural environments and developing countries, with rivers being the most monitored water bodies and the Water Quality Index (WQI) the most frequently employed parameter. Additionally, fixed IoT nodes and ML models for predicting physicochemical variables were the most common technological solutions. Although these technologies show strong potential, the field still presents limitations, including a lack of standardized metrics and insufficient reporting of system costs and energy autonomy. Overall, the convergence of IoT and ML represents a promising pathway for strengthening water quality monitoring in rural communities, provided that future research improves methodological rigor and technical transparency.

Food waste represents a critical structural challenge to the sustainability and profitability of the global food service industry. This study presents a Systematic Literature Review (SLR) under the PRISMA protocol, analyzing 59 high-impact articles indexed in Scopus and Web of Science during the period 2023–2026. The objective was to evaluate the effectiveness of predictive and optimization technologies in mitigating waste and improving operational efficiency. The results reveal a technological dichotomy: while optimization models predominate (54.55%) for resource planning, machine learning (40.00%) is established as the superior tool for demand forecasting in volatile environments. However, a critical implementation gap was identified: 59.32% of the studies validate their models exclusively using mathematical error metrics (RMSE, MAPE), while only 25.42% report tangible operational KPIs such as return on investment or volumetric waste reduction. It is concluded that, in order to move from theoretical precision to industrial utility, future research must integrate economic loss functions into algorithmic training.

Furthermore, this study develops and evaluates a hybrid architecture for algorithmic trading in the USD/PEN foreign exchange market, where a Gaussian Hidden Markov Model (HMM) infers latent market regimes and a deep reinforcement learning (DRL) agent optimizes Buy/Sell/Hold actions within an MDP formulation.

Moreover, a reproducible data pipeline is implemented to integrate heterogeneous financial and macroeconomic sources, producing a consolidated dataset with 4,126 observations and 1,113 variables spanning 2010-02-02 to 2025-12-05, with continuity checks to support stable learning.

The inferred regime signal (e.g., bearish/volatile, sideways, bullish) is incorporated as exogenous context to mitigate market non-stationarity and improve policy learning.

Finally, out-of-sample benchmarking shows that the RL-based strategy outperforms supervised baselines and Buy & Hold, achieving the highest cumulative return (28.64%) and improved risk-adjusted performance when regime context is included (Sharpe 0.116 for RL+HMM vs 0.081 for RL-only), while remaining robust in adverse periods where the passive benchmark incurs losses (-14.56%).

In addition, training dynamics suggest PPO yields more stable convergence than alternative DRL methods in a noisy financial environment.

Keywords: USD/PEN; Hidden Markov Models; market regimes; deep reinforcement learning; PPO; algorithmic trading; risk management; CRISP-DM..

This research work described the issues regarding vehicle monitoring management in Trujillo, Peru, and aimed to propose a computer vision system to optimize multi-camera tracking in the year 2026. The study was descriptive-propositional with a quantitative approach; a questionnaire was applied to a non-probabilistic sample of 50 control center operators. The diagnosis revealed critical deficiencies in current operations, highlighting high discontinuity in inter-camera tracking (4.2) and visual fatigue in screen comparison (4.0), demonstrating the inefficiency of manual processes. In response, the UrbanSight technical proposal was designed, integrating the YOLOv11 model for vehicle detection and optimizing the DeepSORT algorithm through Re-Identification (ReID) techniques trained with the VeRi-776 dataset. It was concluded that this technological integration allows maintaining vehicle identity persistence across video surveillance networks, representing a viable solution to automate traceability and reduce reliance on the human factor.

Early detection of acute respiratory infections is critical for timely public health interventions, especially in vulnerable populations. This study presents a hierarchical ensemble framework with stacking applied to Peru's National Epidemiological Surveillance System (RENACE), covering data from 2000 to 2024 for the Arequipa region (130,970 records). The framework integrates six diverse base models—XGBoost, LightGBM, CatBoost, Random Forest, Extra Trees, and Ridge Regression—combined using RidgeCV meta-learning to predict six simultaneous targets: pneumonia cases, hospitalizations, and deaths for children under 5 and adults over 60. Using comprehensive spatiotemporal feature engineering (more than 80 features including lags, moving statistics, seasonal patterns, and geographic aggregations), the stacking ensemble achieved exceptional performance with R²=0.9957, MAE=0.0005, and RMSE=0.0082, outperforming all individual models. Notably, Ridge regression achieved R²=0.9999, indicating an almost perfect fit to the aggregated departmental data. The proposed system demonstrates strong potential as an operational early warning tool for resource allocation and epidemic preparedness in developing countries with limited surveillance infrastructure.

This study presents a data-driven workflow to optimize artificial lift system (ALS) selection for wells in the Oriente Basin, Ecuador. The goal is to support engineers in choosing the most suitable ALS based on production and operational characteristics. Proper ALS selection is critical to maintain stable production, reduce unnecessary energy use, and minimize failures caused by mismatches between reservoir conditions and lift mechanisms. Historical well and production data were compiled and processed through data cleaning, feature engineering, and class-balancing techniques to improve representation of underused ALS categories. Multiple multiclass machine-learning classifiers were trained to predict the recommended ALS using key operational parameters. The best model was embedded in a web-based application that allows users to input well data and obtain data-driven ALS recommendations.
Among the evaluated algorithms, XGBoost and a Stacking ensemble achieved the strongest performance, with test accuracies above 99%, while Random Forest and Decision Tree models reached about 96%. Overall evaluation shows high predictive capability. Comparison with field installations yielded an agreement of 83.3% between model recommendations and deployed ALSs. Although field choices do not always match model outputs, results indicate that the models provide robust and consistent predictions under current data conditions. The proposed workflow demonstrates that machine-learning–based ALS selection is a practical and reliable decision-support approach for wells with similar characteristics. Its use can help standardize selection criteria, reduce operational uncertainty, and improve production management efficiency across assets.