Quality control in green coffee bean production is critical for food safety and economic sustainability. A persistent gap in existing automated inspection systems is their inability to capture both sides of each bean, which leads to systematic under-detection of surface defects. This work presents three contributions: (i) a novel mechatronic apparatus that mechanically guarantees dual-sided imaging of every bean, (ii) a public 12-class dataset of green coffee bean defects, and (iii) an embedded, real-time inspection pipeline validated on low-cost hardware. The apparatus sequentially presents each bean, from a standard 350 g sample, to two 16-megapixel cameras under controlled LED illumination. A dataset of 9600 images spanning 12 classes (11 defects and 1 normal) was generated from expert-classified samples and enriched through data augmentation. Four convolutional neural network (CNN) architectures, VGG-16, VGG-19, ResNet-50 and YOLOv8, were trained and benchmarked using precision, recall, F1-score and mean average precision. YOLOv8 achieved the best overall performance, with a precision of 97.4%, a recall of 99.6%, an F1-score of 0.930 and a mean average precision of 96.5%, outperforming VGG-16 (accuracy 86.07%), VGG-19 (accuracy 67.03%) and ResNet-50 (accuracy 87.76%). Dual-sided acquisition raised mean per-class detection accuracy from 0.727 to 0.908, a relative gain of 25.7% over an equivalent single-sided configuration. Deployed in real-time “track” mode on a Raspberry Pi 4, the system simultaneously classifies defects and counts beans by category, processing a 350 g sample in approximately 38 min. Combining mechanical innovation with lightweight deep learning enables practical, scalable, and cost-effective quality control for laboratories specialized in coffee analysis.

Over the past decade, serious games and virtual reality have gained increasing relevance in upper-limb rehabilitation, yet desktop virtual reality solutions often suffer from reduced spatial correspondence and limited sensory feedback. This work presents the design and preliminary evaluation of a desktop virtual reality-based serious game that combines Leap Motion Controller hand tracking with a custom wireless vibrotactile wearable device to support upper-limb rehabilitation training. Three training scenarios were implemented to target pronation/supination, pinch grip, ulnar/radial deviation, and wrist, elbow, and finger flexion/extension. Usability (System Usability Scale, SUS), user experience (short AttrakDiff), and perceived workload (Raw NASA-TLX), together with functionality and perception questionnaires, were collected from healthy participants randomly assigned to two groups (Group 1: n=13, LMC only; Group 2: n=9, LMC plus wearable). Across all instruments, the configuration including the wearable device tended to obtain higher usability ratings, more desirable pragmatic and hedonic quality scores, and lower overall workload means than the LMC-only configuration, with moderate effect sizes but limited statistical power due to the small samples. Participants in the wearable condition also reported clearer feedback, a perceived improvement in movement precision, and a stronger perceived alignment between real and virtual actions. These findings suggest that the proposed system may serve as a promising user-centered prototype for desktop VR-based upper-limb rehabilitation and provide preliminary design evidence to support future clinical and kinematic validation studies with larger cohorts.

Dendritic cell (DC) immunotherapy is a promising approach for treating cancers such as melanoma and prostate cancer. Although DC-based vaccines can elicit potent anti-tumor immune responses, dosing schedules in both preclinical and clinical settings are often chosen empirically rather than through quantitative optimization. In this work, we develop an enhanced mathematical model of tumor-immune dynamics that incorporates a more realistic tumor growth law and an estimated immune-response delay, enabling the systematic design of DC vaccination protocols. Tumor-growth and immunotherapy parameters were calibrated using experimental melanoma data and two metaheuristic optimization methods: Genetic Algorithm and Particle Swarm Optimization. Using the calibrated model, we derived vaccination schedules consisting of three injections totaling 2.4 × 106 DCs. Despite using the same total dose as the baseline four-injection protocol, the optimized schedules reduced tumor burden by approximately 52% over a 5000-h window, as measured by the area under the tumor-time curve, while also lowering the number of administrations. These results demonstrate that effective tumor control can be achieved without increasing treatment intensity and with substantially fewer vaccinations than previously assumed. Prior optimization studies often required cumulative doses exceeding 1 × 107 cells to obtain comparable therapeutic effects. In contrast, our findings show that metaheuristic algorithms can produce dose-efficient and biologically grounded schedules that significantly enhance treatment performance. This work highlights the value of computational optimization as a decision-support tool for designing efficient and clinically meaningful DC immunotherapy protocols.

Medical simulators provide a safe environment for practising crucial procedures, particularly in virtual simulators where objective and quantitative data can be collected for developing machine learning algorithms for automatic expertise classification. This survey analyses 13 automatic evaluation systems used in medical simulators and identifies best practices for integrating ML algorithms. Among these systems, nine employed commercial simulators, particularly NeuroVR and the Da Vinci robotic systems, while four utilised custom simulators. The survey outlines the main steps in the integration of machine learning algorithms: data collection, metric generation and selection, training, and testing. Metric selection was identified as a crucial factor affecting both the accuracy of the algorithm and the comprehension of the evaluation. Typically, multiple machine learning algorithms were applied to the same dataset to compare results and identify the most effective model. Overall, this survey suggests that transparent algorithms are preferable, as they enhance physicians' understanding.

Public authorities and private companies have used video cameras as part of surveillance systems, and one of their objectives is the rapid detection of physically violent actions. This task is usually performed by human visual inspection, which is labor-intensive. For this reason, different deep learning models have been implemented to remove the human eye from this task, yielding positive results. One of the main problems in detecting physical violence in videos is the variety of scenarios that can exist, which leads to different models being trained on datasets, leading them to detect physical violence in only one or a few types of videos. In this work, we present an approach for physical violence detection on images obtained from video based on threshold active learning, that increases the classifier’s robustness in environments where it was not trained. The proposed approach consists of two stages: In the first stage, pre-trained neural network models are trained on initial datasets, and we use a threshold (μ) to identify those images that the classifier considers ambiguous or hard to classify. Then, they are included in the training dataset, and the model is retrained to improve its classification performance. In the second stage, we test the model with video images from other environments, and we again employ (μ) to detect ambiguous images that a human expert analyzes to determine the real class or delete the ambiguity on them. After that, the ambiguous images are added to the original training set and the classifier is retrained; this process is repeated while ambiguous images exist. The model is a hybrid neural network that uses transfer learning and a threshold μ to detect physical violence on images obtained from video files successfully. In this active learning process, the classifier can detect physical violence in different environments, where the main contribution is the method used to obtain a threshold μ (which is based on the neural network output) that allows human experts to contribute to the classification process to obtain more robust neural networks and high-quality datasets. The experimental results show the proposed approach’s effectiveness in detecting physical violence, where it is trained using an initial dataset, and new images are added to improve its robustness in diverse environments.

Violence against women captured in videos and surveillance systems necessitates effective identification to enable appropriate reactions for controlling and mitigating of its effects in public spaces and the potential apprehension of aggressors. While several algorithms have been developed for violence detection, their evaluation has primarily focused on controlled scenarios with clear differentiation between violent and non-violent scenes, representing two-class identification problems. However, real-world situations often present challenges where specific actions, such as hugs or effusive greetings, fall into an ambiguous class that is difficult to classify. Consequently, this transforms into a multi-class identification problem. In this study, we assess the performance of three pre-trained models, namely VGG16, ResNet50, and InceptionV3, to evaluate their efficacy in addressing the multi-class identification challenges. Furthermore, we compare their performance against datasets consisting of two-class classifications, where the models generally exhibit satisfactory results. Our analysis reveals that the models struggle to differentiate the ambiguous scenes effectively, with Inception V3 achieving a 0

This paper presents a mathematical model derived from an equivalent electrical circuit to describe the dynamic behavior of the high-voltage terminal of a tandem Van de Graaff accelerator. Two approaches are presented for modeling the transit time of the current flowing through the corona needles. The first one considers an equivalent self-inductance in the corona triode, whereas, in the second one, the transit time is represented by a delay in the corona current. The validation of the proposed models was carried out through experimental tests developed at the National Nuclear Research Institute of Mexico. Furthermore, two strategies for controlling the slow variations of the terminal voltage limited by the slow response of the control loop based on corona discharge are evaluated: a Proportional–Integral–Derivative controller and a sliding mode controller. The Root-Mean-Squared Error calculation leads to the conclusion that both control strategies are suitable for regulating the voltage at the accelerator potential terminal. However, the sliding mode controller leads to an overshoot-free response and a shorter settling time.

The use of renewable energy has experienced significant growth as part of efforts to reduce fossil fuel consumption. Mexico, due to its geographical location, holds significant potential for solar energy development. However, this valuable resource remains underutilized, largely due to the low efficiency of photovoltaic panels. To meet user demand for power, implementing a power converter and control algorithms is essential. This article focuses on the implications of not having a voltage converter and analyzes the performance of various controllers applied to the Maximum Power Point tracking problem. Specifically, it examines Proportional Integral Derivative control, first-order sliding mode control, and Super Twisting sliding mode control under abrupt and smooth changes in temperature and irradiance. The objective is to determine the most efficient and reliable control strategy for solving the stated problem.

The Romberg test is widely used for static postural balance assessment, but the effect of foot position on center of pressure (CoP) parameters remains unclear. This study compared CoP parameters in 47 participants aged 13 to 71 years, with two foot positions: a comfortable stance and feet together. This broad age range was used to capture significant diversity in physical development, balance, and health conditions, ensuring results representative of variability in a general population. Participants performed the test with eyes open (EO) and closed (EC). CoP parameters such as mean velocity, range, standard deviation, area of the 95% confidence ellipse, mean frequency, and Romberg ratio were measured. Paired t-tests were used to evaluate the effect of vision and foot position. The results showed that both vision and foot position significantly influenced postural stability, with greater sway in the EC and feet together conditions, especially in the mediolateral direction. For example, mean velocity increased from 10.02 ± 4.09 mm/s in EO-comfortable stance to 21.75 ± 8.93 mm/s in EC-feet together. These findings highlight the importance of standardizing foot position and considering the effect of vision when assessing postural stability.

This paper proposes a robust maximum power point tracking algorithm based on a super twisting sliding modes controller. The underlying idea is solving the classical trajectory tracking control problem where the maximum power point defines the reference path. This trajectory is determined through two approaches: a) using the simplest linear and multiple regression models that can be constructed from the solar irradiance and temperature, and b) considering optimum operating parameters derived from the photovoltaic system’s characteristics. The proposal is compared with the classical methods Perturbation and Observation and Incremental Conductance, as well as with two recently reported hybrid algorithm based on Artificial Neural Networks: one uses the Levenberg-Marquardt algorithm and the other applies Bayesian regularization to generate current and voltage references, respectively. Both use a Proportional-Integral-Derivative controller to solve the maximum power point tracking problem. Numerical simulations confirm the effectiveness of the method proposed in this work regarding convergence time, power efficiency, and amplitude of oscillations. Furthermore, it has been shown that, although no significant differences in the system response are observed with respect to the Artificial Neural Networks-based methods, the proposed algorithm with a reference generated through a linear regression constitutes a low-complexity solution that does not require a temperature sensor to efficiently solve the maximum power point tracking problem.

This paper presents a detailed analysis of the relation between physical characteristics and defects of green coffee beans and the sensory profile that influence the sensory notes of fragrance, aroma, flavor, and aftertaste of coffee. Machine learning models were used to identify the variables of importance and identify the ways in which these variables affect the sensory note of coffee, to determine which algorithm and its hyperparameters have greater precision in determining the sensory values of coffee such as floral, fruity, herbal, nutty, caramel, chocolate, spicy, resinous, pyrolytic, earthy, fermented, and phenolic. The result indicates the relationship and importance that exist between the physical variables, defects, and size of the green coffee bean, with respect to their respective sensory notes. The data of the proposed system demonstrate that by combining the scores of several experts, a precision can be achieved analogously to that obtained by cupping experts; therefore, the possibility of errors induced by human concerns such as fatigue or subjectivity is reduced.

One of the most popular renewable energy sources is photovoltaic energy; however, its main drawback is the low conversion efficiency. Optimal system operation requires efficient tracking of the maximum power point representing the maximum energy that can be extracted from the photovoltaic system. One of the main problems is the presence of partial shadows. In this scenario, the system's power presents multiple peaks; a robust control technique is required to properly track the Global Maximum Power Point (GMPP). This paper proposes using a Sliding Mode Controller (SMC) for the GMPP tracking which is a robust strategy for coping with changing environmental conditions and partial shadows. Numerical simulation results show the effectiveness of the proposal.

Introducción: En el Estado de México no existen investigaciones que proporcionen información para toma de decisiones y administración de recursos relacionados con la atención de las lesiones por causa externa (LCE).Objetivo: Describir las LCE en un servicio de urgencias durante un periodo de cinco años.Método: Se diseñó un estudio retrospectivo con pacientes que ingresaron al servicio de urgencias (2010-2015) por diagnóstico de LCE.Se realizó análisis descriptivo y de clúster.Resultados: En el servicio de urgencias, 16.59 % de las atenciones derivaron de LCE.Se incluyeron 16 567 pacientes de 14 a 99 años (media o promedio = 37.7, DE = 17.28), 69.2 % fue del sexo masculino.Las LCE principalmente ocurrieron en la vía pública (26.3 %) y en el hogar (23.7 %).Las causas más frecuentes fueron agresiones fuera del hogar (32.7 %), en promedio a los 34 años; caídas (25 %) en promedio a los 45 años; accidentes ocasionados por vehículos de motor (9.7 %), en promedio a los 33 años.El análisis por clúster identificó cuatro grupos: agresiones fuera del hogar 32.7 % (5417), contactos traumáticos 26.30 % ( 4363), accidentes de tránsito 15.9 % (2,640) y caídas 25 % (4147).Conclusión: Las LCE relacionadas con vehículos de motor mostraron consecuencias más severas.

Background The timely detection of fall risk or balance impairment in older adults is transcendental because, based on a reliable diagnosis, clinical actions can be taken to prevent accidents. This study presents a statistical model to estimate the fall risk from the center of pressure (CoP) data. Methods This study is a cross-sectional analysis from a cohort of community-dwelling older adults aged 60 and over living in Mexico City. CoP balance assessments were conducted in 414 older adults (72.2% females) with a mean age of 70.23 ± 6.68, using a modified and previously validated Wii Balance Board (MWBB) platform. From this information, 78 CoP indexes were calculated and analyzed. Multiple logistic regression models were fitted in order to estimate the relationship between balance alteration and the CoP indexes and other covariables. Results The CoP velocity index in the Antero-Posterior direction with open eyes (MVELAPOE) had the best value of area under the curve (AUC) to identify a balance alteration (0.714), and in the adjusted model, AUC was increased to 0.827. Older adults with their mean velocity higher than 14.24 mm/s had more risk of presenting a balance alteration than those below this value (OR (Odd Ratio) = 2.94, p<0.001, 95% C.I.(Confidence Interval) 1.68–5.15). Individuals with increased age and BMI were more likely to present a balance alteration (OR 1.17, p<0.001, 95% C.I. 1.12–1.23; OR 1.17, p<0.001, 95% C.I. 1.10–1.25). Contrary to what is reported in the literature, sex was not associated with presenting a balance alteration (p = 0.441, 95% C.I. 0.70–2.27). Significance The proposed model had a discriminatory capacity higher than those estimated by similar means and resources to this research and was implemented in an embedded standalone system which is low-cost, portable, and easy-to-use, ideal for non-laboratory environments. The authors recommend using this technology to support and complement the clinical tools to attend to the serious public health problem represented by falls in older adults.

Today, computer vision algorithms are very important for different fields and applications, such as closed-circuit television security, health status monitoring, and recognizing a specific person or object and robotics. Regarding this topic, the present paper deals with a recent review of the literature on computer vision algorithms (recognition and tracking of faces, bodies, and objects) oriented towards socially assistive robot applications. The performance, frames per second (FPS) processing speed, and hardware implemented to run the algorithms are highlighted by comparing the available solutions. Moreover, this paper provides general information for researchers interested in knowing which vision algorithms are available, enabling them to select the one that is most suitable to include in their robotic system applications.

Virtual environments (VEs) and haptic devices increase patients’ motivation. Furthermore, they observe their performance during rehabilitation. However, some of these technologies present disadvantages because they do not consider therapists’ needs and experience. This research presents the development and usability evaluation of an upper limb rehabilitation system based on a user-centered design approach for patients with moderate or mild stroke that can perform active rehabilitation. The system consists of a virtual environment with four virtual scenarios and a developed haptic device with vibrotactile feedback, and it can be visualized using a monitor or a Head-Mounted Display (HMD). Two evaluations were carried out; in the first one, five therapists evaluated the system’s usability using a monitor through the System Usability Scale, the user experience with the AttrakDiff questionnaire, and the functionality with customized items. As a result of these tests, improvements were made to the system. The second evaluation was carried out by ten volunteers who evaluated the usability, user experience, and performance with a monitor and HMD. A comparison of the therapist and volunteer scores has shown an increase in the usability evaluation (from 78 to >85), the hedonic score rose from 0.6 to 2.23, the pragmatic qualities from 1.25 to 2.20, and the attractiveness from 1.3 to 2.95. Additionally, the haptic device and the VE showed no relevant difference between their performance when using a monitor or HMD. The results show that the proposed system has the characteristics to be a helpful tool for therapists and upper limb rehabilitation.

OBJECTIVE:This paper presents the design and development of a new electronic portable device to assess the human balance of the human body during standing, using a minimal number of sensors and peripheral components. This device is aimed to evaluate human balance in environments outside of specialized laboratories, such as small clinics and therapy offices. APPROACH:The design is based on previous designs using three or more resistive force sensors attached to the feet, however in the present work, the sensors were attached on an adjustable platform, to fit several sizes of feet. Furthermore, all the signal acquisition, process, storage and display are executed by an embedded electronic system, thus avoiding the use of computers and external peripherals. A new method to compute the CoP using only two sensors per foot was developed and tested in a group of 50 university students, (17 women and 33 men), 26.04 ± 4.94 years. MAIN RESULTS:It was developed a portable electronic system to measure the trajectory of the CoP and to calculate the indexes values derived from it. The system is capable to discriminate between measuring situations (open and closed eyes), using only two sensors per foot (p < 0.0001). A comparison between the values obtained for young subjects using the proposed device, and the values reported in the literature showed a similar tendency. SIGNIFICANCE:The results indicate that the proposed system is a good, low-cost, and easy-to-use alternative tool for researchers and clinicians interested in the evaluation of human balance, especially if the measurements must be done outside laboratories.

Background and ObjectiveTraditional methods to determine stress and anxiety in academic environments consist of the application of questionnaires, but the main disadvantage is that the results depend on the students’ self-perception. Being able to detect anxiety-related stress levels in a simple and objective way contributes greatly to dealing with low performance and school drop-out by students.MethodsThe main contribution of this study is to identify the physiological features that could be used as predictors of stressful activities and states of anxiety in academic environments using an Arduino board and low-cost sensors. A test with 21 students was conducted, and a stress-inducing protocol was proposed and 21 physiological features of five signals were analyzed. In addition, the State-Trait Anxiety Inventory (STAI) was used to assess the level of anxiety for each student. Four classifiers were compared to find the physiological feature subset that provides the best accuracy to identify states of stress and anxiety.ResultsThe stress due to activities performed by students can be identified with an accuracy greater than 90% (Kappa = 0.84) using the k-Nearest Neighbors classifier, using data from heart rate, skin temperature and oximetry signals and four physiological features. Meanwhile, the identification of anxiety was achieved with an accuracy greater than 95% (Kappa = 0.90) using the SVM classifier with data from the galvanic skin response (GSR) signal and three physiological features.ConclusionsThe results provide a clue that anxiety detection in academic environments could be done using the analysis of physiological signals instead of STAI test scores. Besides, the results suggest that physiological features could be used to develop stress recognition systems to help teachers to identify the stressful tasks in an academic environment or to develop anxiety recognition systems to help students to control their level of anxiety when they are performing either academic tasks or exams.

This work presents the design and development of a new alternative tool to measure the Center of Pressure (CoP) displacements, intended to evaluate the human balance. The device is based on a modified commercial balance board used for video games, resulting in a low-cost, portable device capable of computing the CoP, providing 24 of the most used indexes to test the human balance. The proposed standalone device runs on rechargeable batteries, weighs only 3.5 kg, and has a data storage capacity for over 1000 tests. Visual and auditory instructions assist its user interface. Thus, contrary to the commercial systems designed for laboratory use, this device enables the measurement of quantitative balance parameters in non-laboratory places, allowing the study of the balance of vulnerable populations directly on their typical environments. To evaluate the device, 20 older adults (68.60 ± 1.23 years) were tested, and the resulting values were compared with a similar study using a force platform; 19 indexes showed a similarity with those reported using force platform and 12 of these were statistically equivalent. The proposed device represents an open-source alternative tool for researchers and healthcare personnel to acquire reliable data to evaluate human balance.

In this work, the authors propose the use of a new electronic measurement platform to assess the performance of goalkeepers based on their cognitive and motor skills. A test was run in order to show how the GoMeSy platform could be used by coaches to evaluate the penalty shot-stopping skills of goalkeepers, consequently, the performance of 22 players was measured. In the case of the cognitive skills, there is not a significant difference between novice and expert goalkeepers (p = 0.333), meanwhile, in the case of the motor skills, there is a significant difference (p = 0.006). In addition, there is a moderate, positive correlation between weight and motion time (r = 0.437). In general, this study demonstrates that the GoMeSy platform allows coaches to evaluate with accuracy the penalty shot-stopping skills of goalkeepers. The findings of the test emphasize the fact that coaches do not have to focus only on the goalkeepers' physical training (motor skills), but they should also consider developing the reaction time (cognitive skills). The results also highlight that although the body mass index depends on the height and weight of each subject, coaches should focus primarily on monitoring the weight of players to improve their performance.