Inge Cuc

Systematic Mapping of a Process to Support MLOps in Small and Medium-Sized Software Development Companies

2024-08-26T05:10:44+00:00

Introduction: Currently, software development companies (SDCs) have begun to incorporate Machine Learning into their projects, which has allowed Machine Learning models to move from the experimentation stage to the production stage. This is where Machine Learning Operations (MLOps) comes in, with the goal of bridging the gap between operations, development, and data science teams.

Objective: Conduct research on the current state of knowledge regarding the adoption of MLOps in SDCs through systematic literature mapping, with the aim of studying, identifying, and understanding initiatives, solutions, and issues related to work in this area.

Method: A systematic literature mapping was carried out using a defined protocol, which included the development of research questions and the implementation of a search strategy in six databases. Primary articles were then selected based on pre-established inclusion and exclusion criteria. The research questions were addressed based on the results obtained, allowing the findings to be classified and characterized. Finally, the results were analyzed, and the corresponding conclusions were presented.

Results: The findings of this article reflect the efforts of the scientific community to define the principles, roles, artifacts, technologies, challenges, and crucial factors for the implementation of Machine Learning Operations in SDCs.

Conclusions: The research questions are answered, which reveal the main challenges in implementing Machine Learning.

Synergies between Digital Transformation and Sustainability: A Scientific Mapping of Project Management in Industry 4.0

2026-03-02T22:41:52+00:00

Introduction: Currently, the interaction between digital transformation, sustainability and project management integrates the academic discussion associated with Industry 4.0. The accelerated incorporation of intelligent technologies, together with the growing demands for responsible production models, stimulates the growth of scientific production in this field. Unfortunately, this growth lacks integrative analyses to understand its evolution and thematic structure.

Objective: Examine global research trends on digital transformation, sustainability, and project management, from a bibliometric approach to identify influential authors, emerging lines of research, and citation patterns.

Method: Records indexed in Scopus, processed using Bibliometrix (RStudio), VOSviewer and Gephi, were analyzed. Applying the PRISMA protocol, 87 documents were selected that met the established criteria. The analysis addressed scientific productivity, collaborative networks, specialized sources, institutions with greater presence and co-occurrence of keywords.

Results: There is evidence of a progressive growth in literature, with India, Italy and Spain as nations with the greatest contribution. Journal of Cleaner Production and Sustainability are positioned as the most influential journals, while authors such as Ghobakhloo and Frank concentrate the largest number of citations. Concepts such as digitalization, Industry 4.0, project management and artificial intelligence are recurrent, reflecting the incorporation of advanced technologies in sustainable management approaches.

Conclusions: Digitalization works as a catalyst to move towards more efficient and sustainable management models. There are important gaps in Latin American production and opportunities to develop integrated frameworks that articulate technological innovation and sustainability in project management processes.

Symbolic regression model based on harris hawks optimization for temperature rediction in bifacial PV modules

2025-08-22T17:25:01+00:00

Introduction: Accurate temperature prediction in bifacial photovoltaic (PV) modules is crucial for optimizing energy efficiency and system longevity. This study presents a symbolic regression model optimized using the Harris Hawks Optimization (HHO) algorithm and compares its performance with a Genetic Algorithm (GA)-based symbolic model and statistical methods, using real-world data from a 26.6 MW bifacial PV plant in Colombia.

Objective: To develop an interpretable symbolic regression model to predict the temperature of bifacial PV modules with solar trackers, using solar radiation and solar time as input variables.

Method: Four models were designed and compared: multiple linear regression (MLR), gradient descent-enhanced MLR, symbolic regression with GA, and symbolic regression with HHO. A one-year dataset with 5-minute resolution was used. Correlation and normality analyses were conducted, and model performance was assessed using RMSE and R² metrics.

Results: The gradient descent-enhanced MLR model showed the best performance (RMSE: 4.92; R²: 0.86), followed by the SR-GA model (RMSE: 7.14; R²: 0.71). The SR-HHO model exhibited faster convergence and better performance with smaller datasets, though it showed lower accuracy with larger data volumes (RMSE: 13.91; R²: 0.09).

Conclusions: Symbolic models are effective for interpreting thermal behavior in bifacial PV modules. HHO is computationally efficient with small datasets, while GA provides more stable performance with large datasets. A hybrid approach combining both algorithms is recommended to improve predictive performance.

Use of artificial intelligence for patient health management in healthcare institutions. Benefits and risks

2025-03-24T19:22:18+00:00

Introduction: Artificial intelligence (AI) in the health sector promises great impacts, allowing humanity to face challenges such as early detection of diseases, accurate diagnoses, personalization of treatments and optimization of processes in health service providers.

Objective: To examine recent literature on the implications of using AI in healthcare management, exploring both its benefits and associated concerns.

Method: The research is descriptive based on documentary analysis of articles published between 01/01/2018 and 06/30/2024, in English and Spanish, in databases such as Scopus and Web of Science Sciences, published between the years 2018-2024 are considered.

Results: This review highlights the potential of AI to improve diagnostic accuracy, personalize treatments and reduce medical errors, acting as a cognitive assistant in clinical workflows. However, there are acknowledged risks related to ethics, data privacy, and regulatory development.

Conclusions: While AI presents a significant opportunity for digital transformation to improve health management for people and healthcare providers, its responsible implementation requires clear regulatory frameworks and strategies focused on people (patients and healthcare professionals). In this way, the aim is to take advantage of its benefits, mitigating risks and promoting reliable, equitable and efficient medical care. Therefore, this review is important to understand different positions and to continue with the assessment of the impacts and research on specific topics.

Proposal for an Artificial Neural Network Model to Predict the Success or Failure of Projects Based on the PMBOK Approach

2026-03-18T21:34:28+00:00

Globally, companies and organizations are increasingly implementing methodological tools or best practice standards for project management, such as PMBOK; however, it is important to recognize that, given the robustness of these approaches, a high level of expertise is required from project managers. In this context, the use of predictive tools can contribute to informed decision-making across the different stages of the project life cycle. Thus, this study proposes as its main contribution a novel model based on neural networks for predicting the success or failure of a project managed under PMBOK, using project features such as the resources involved, the project group according to PMBOK, and project indicators. For the development of this work, an adaptation of the CRISP-DM methodology into four phases was carried out. As a result, the model achieved consistent fitting over 100 training epochs, reaching an accuracy above 95% after 60–70 epochs, which suggests excellent fitting and generalization capabilities in predicting project outcomes. In conclusion, the model represents a key contribution to project management under PMBOK, with its differential factor being the inclusion of resources, the project group attribute, and the vector representation of qualitative project features identified within the indicator attribute.

Implementation science frameworks for evaluating primary health care interventions in low- and middle-income countries: a scoping review

2025-12-15T04:32:08+00:00

Primary Health Care (PHC) is understood and operationalized in multiple ways, and this variability complicates how implementation is assessed in low- and middle-income countries (LMICs). We conducted a scoping review to map how PHC-oriented interventions in LMICs have been evaluated using the Reach, Effectiveness, Adoption, Implementation, and Maintenance (RE-AIM) framework and other implementation science approaches. Eighteen eligible articles were included. The interventions covered prevention, early detection, and management of chronic communicable and non-communicable conditions, in addition to maternal, child, and mental health. Most studies reflected selective PHC models, and Africa contributed the largest share of publications. RE-AIM was explicitly applied in seven studies; the remaining articles used other implementation science frameworks, most often within mixed-method designs. Overall, findings aligned with the stated study designs and generally followed appropriate reporting practices. This review offers a pragmatic map of how implementation science frameworks are being used to evaluate PHC interventions across LMIC setting.

Design and Implementation of an Accessible Monitoring System for a Wind Energy System in a Power-to-Gas Pilot Plant Located in Middle Guajira

2025-06-26T20:58:07+00:00

This study presents the development and implementation of a low-cost data acquisition system (SAD.1) for real-time monitoring of a wind energy system installed at a Power-to-Gas (PtG) pilot plant located in Middle Guajira. The objective was to design a functional and cost-effective tool capable of recording and analyzing the energy performance of the wind system, thereby contributing to the efficient management of the PtG pilot plant. The system uses Modbus RTU communication under the RS485 standard, with an Arduino Uno microcontroller and a MAX485 module for signal conversion to TTL levels. Data are extracted from the wind inverter and transmitted through a serial port to a PC, where a Python script processes them, computes five-minute averages, and stores them in .CSV files. Operational profiles were established over one year of monitoring, identifying higher generation between 7:00 a.m. and 6:00 p.m. and low or no production during nighttime hours. SAD.1 makes it possible to evaluate the behavior of the wind system both individually and in integration with the photovoltaic system, providing key data to optimize the energy planning and management of the PtG plant.

A reproducible layered directed acyclic graph protocol for secondary-data studies

2026-02-11T14:12:41+00:00

Introduction: Secondary-data studies (clinical, administrative or sensor records) embed selection, measurement and missing-data processes that are often treated as “preprocessing”, leaving key causal assumptions implicit and hard to audit.

Objective: To propose a reproducible protocol that makes causal assumptions explicit in secondary-data studies using layered directed acyclic graphs.

Methodology: The protocol separates a material causal system (exposure→outcome) and three additional layers: (i) selection/observability, (ii) operationalisation and measurement (constructs vs. recorded proxies) and (iii) missing-data mechanisms. It includes estimand and time alignment, extraction rules and cohort construction, plus quality control based on graph source code and figure traceability.

Results: Outputs include a reporting checklist (Table 1), a decision→threat→graph-signature mapping with conditional mitigations (Table 2), a related-work gap synthesis (Table 3), layered DAG examples (Figs. 1–9) and reproducible evaluations quantifying typical biases and sensitivity to missingness (Tables 4–7).

Conclusions: The layered approach makes selection, measurement and missingness assumptions explicit; improves reproducibility through code–figure–text traceability; and operationalises alignment between design/analysis choices and the estimand, limiting claims that exceed the available evidence. A minimal applied case shows that observation/missingness decisions can materially shift the estimand, motivating explicit sensitivity reporting (Δ) and denominators.