Programa del congreso

1. Introduction and Objective

We have previously developed and in vivo validated threadlike, injectable, neuromuscular microstimulators. These devices operate wirelessly through the rectification of bursts of high-frequency current delivered to tissues by volume conduction. These implants resemble small leads: a thin, elongated and flexible silicone body (< 1 mm diameter, 35 mm length) containing in its middle a hermetic titanium capsule housing the electronics. Two platinum-iridium helical coils form the two electrodes at opposite ends and their interconnections to the capsule [1].

As any active implantable medical device (AIMD), these microstimulators will experience repeated mechanical stresses within the corrosive environment of the body, which can affect their mechanical properties, leading to failure. Long-term and complex in vivo tests are mandatory to test the reliability of AIMDs. However, mechanical tests using accelerated testing protocols may reveal mechanical failures in a short-term and present an ethical alternative to reduce the number of animal tests.

Regulatory agencies and organizations have established standards for the mechanical testing of AIMDs. The most common tests include tensile, fatigue and fracture, compression, flexion, vibration, shock and also combinations of them [2].

For these injectable intramuscular microstimulators that mostly experience flexural stresses caused by the contraction and relaxation of muscles, we have developed a customized test bench to perform flex bending fatigue tests that combine tension and compression forces.

2. Methods

The machine was designed to perform repetitive bending of the microstimulators submerged in a heated saline bath to test their reliability in simulated in vivo conditions. The support structure is made with aluminum profiles to provide stability and robustness. The fixtures, gears and ball-bearings are manufactured in polymers with good thermal and chemical resistance to avoid corrosion and reduce wear. The step-motor that actuates the machine is controlled by a Raspberry Pi that also provides an interface to select angle, frequency and number of cycles for the test.

Following guidance documents [3], the fixture dimensions are 110% larger than the diameter of the device under test and have a curvature radius of 6 mm. The machine configuration considers a maximum bending angle of 25 degrees. These parameters consider the worst case of range of motion and stresses observed in chronic experiments with rabbits.

3. Results and Discussion

The machine is capable of performing over 7.5 million cycles for over 43 days, which is equivalent to 10 years of life-span based on a median span of a person's gait of 1.03 s [4] and an adult’s average walking count of 2,000 steps per day, using a frequency of 2 Hz, as suggested by standards [5], with a contraction time of 310 ms and a relaxation time of 205 ms. The fixtures can fix the implant at different points along its length, allowing to test the interconnection points that are considered the most susceptible to mechanical failure.

References

[1] A. García-Moreno, A. Comerma-Montells, M. Tudela-Pi, J. Minguillon, L. Becerra-Fajardo, and A. Ivorra, “Wireless networks of injectable microelectronic stimulators based on rectification of volume conducted high frequency currents”, J. Neural Eng., vol. 19, no. 5, p. 056015, 2022, doi: 10.1088/1741-2552/ac8dc4.

[2] International Standard, “Implants for surgery – Active implantable medical devices – Part 1: General requirements for safety, marking and for information to be provided by the manufacturer” ISO 14708-1:2014(E), second edition, August 2014.

[3] U.S. Department of Health and Human Services, Food and Drug Administration (FDA), Center for Devices and Radiological Health (CDRH), “Guidance for the Submission of Research and Marketing Applications for Permanent Pacemaker Leads and for Pacemaker Lead Adaptor 510(k) Submissions.” Guidance for Industry, November 2000. [Available at: https://www.fda.gov/media/71740/download]

[4] J. B. Webster and B. J. Darter, “Principles of Normal and Pathologic Gait,” Atlas of Orthoses and Assistive Devices, Elsevier, 5th edition, pp. 49–62, 2019, doi: 10.1016/B978-0-323-48323-0.00004-4.

[5] European Committee for Standardization (CEN), European Electrotechnical Committee for Standardization (CENELEC), “Active implantable medical devices - Part 2-1: Particular requirements for active implantable medical devices intended to treat bradyarrhythmia (cardiac pacemakers).” Standard EN 45502-2-1, 2003.

The blood–brain barrier (BBB) is essential for protecting the brain from harmful substances but also severely restricts central nervous system (CNS) drug delivery, contributing to the failure of nearly 80% of neurodegenerative disease (NDD) drug candidates in clinical trials. To address the need for more physiologically relevant preclinical models, we present a three-dimensional (3D) bioprinted human BBB platform using microvalve-based embedded 3D printing.

Our low-viscosity bioink, formulated from natural polymers and brain microvascular endothelial cells (BMECs), recreates a biologically compatible microenvironment that more closely mimics the native BBB than synthetic alternatives. This strategy enables the reproducible fabrication of 3D ring-shaped scaffolds with high structural fidelity. Up to 48 constructs can be automatically printed within minutes while maintaining excellent cell integrity and viability, supported by precise droplet deposition and minimal shear stress.

The platform further accommodates the incorporation of additional neural cell types, allowing systematic exploration of neurovascular interactions and dynamic crosstalk between cellular compartments. Its architecture also supports targeted delivery of drugs, nanoparticles, or imaging agents into the luminal channel, making it suitable for probing brain physiology, disease mechanisms, therapeutic efficacy and safety, and neuroimaging probe development.

Collectively, these findings establish a robust, scalable, and cost-effective 3D bioprinted BBB model that holds significant promise for bridging the gap between preclinical studies and clinical translation in NDD drug development.

Las heridas crónicas y úlceras cutáneas representan un desafío clínico debido a su limitada capacidad de regeneración, alta incidencia de infecciones y riesgo de complicaciones graves como amputaciones. En este estudio se propone un apósito avanzado basado en plasma rico en factores de crecimiento (PRFC), con propiedades homogéneas y estandarizadas. Para su liberación controlada, se diseñó un soporte de policaprolactona (PCL) biodegradable y biocompatible, funcionalizado químicamente para incorporar factores de crecimiento. Los resultados obtenidos demuestran que el apósito desarrollado promueve de manera significativa la proliferación celular en comparación con los controles sin factores de crecimiento

1. Introduction

Extracellular matrix (ECM) mechanics play a critical role in directing cellular behavior. Considerable efforts focus on developing in vitro platforms that replicate the mechanical properties of biological tissues in health and disease (1). Natural hydrogels are promising as they resemble native environments but require mechanical tuning. This is particularly relevant for the myocardium, which sustains continuous loading. Yet their mechanical properties remain poorly characterized across scales, from the microscale sensed by cells to the macroscale of whole tissue, and especially under strain.

2. Objective

To characterize the multiscale mechanical properties of collagen hydrogels using atomic force microscopy (AFM) and rheometry with different crosslinking degrees and stretches applied.

3. Methodology

To measure the micromechanical stiffness with AFM, a device to stretch samples, compatible with simultaneous AFM measurements, was first designed and built (Figs. 1A,B). Collagen I was isolated from rat tails and diluted at 10 mg/mL. To produce photocrosslinkable hydrogels, 10 µL of Ru/SPS photocrosslinker (Advanced Biomatrix, #5248) was added to the collagen I mixture and then pipetted on top of the stretching device preparing gels of 50 µL volume and left it at 37 ºC for 30 min for thermal crosslinking. Afterwards, gels were illuminated with blue light (455 nm) for 0, 5 or 30 minutes. Then, the stretching device was placed on the AFM stage and the hydrogels were measured at strains up to 30% to extract the Young’s modulus. In parallel, myocardial ECM derived hydrogels from swine hearts were produced by following already described protocol (2). The produced hydrogels were further crosslinked with genipin (5 mM) and further tested with AFM at 0% stretch. Macromechanical properties of uncrosslinked hydrogels were assessed by rheometry using a HAAKE RheoStress 1 rheometer (ThermoFisher, MA) with 35 mm serrated parallel plates. 1 mL of hydrogel solution was loaded onto the rheometer Peltier plate (4 °C) with a plate gap of 200 µm. The storage modulus (G'), loss modulus (G''), and viscosity (η) were recorded at a constant frequency of 0.1 Hz and strain of 0.5%. The plate temperature was maintained at 4 °C for 15 min, then increased to 37 °C (to allow for thermal crosslinking of the hydrogels) and held constant until the end of the experiment.

4. Results and Discussion

The AFM-compatible stretching device (Figs. 1A,B), fabricated using 3D-printed polylactic acid, allowed successful micromechanical measurement of hydrogel Young’s modulus under stretch. Collagen I hydrogels exhibited a baseline stiffness of ~100 Pa (uncrosslinked, control), ~900 Pa (crosslinked, 5 min), and ~1200 Pa (crosslinked, 30 min) (Fig. 1C). The Young’s modulus displayed significant nonlinearity under strain, increasing ~3 to 5 fold at 30% strain (Fig. 1C), independently of the crosslinking degree. Uncrosslinked collagen hydrogels were softer than myocardial ECM hydrogels, but crosslinking increased collagen I stiffness beyond that of myocardial ECM (Fig. 1D). Finally, the rheometry results showed that thermally crosslinked myocardial ECM hydrogels are also stiffer than the collagen I hydrogels following the same tendency as the micromechanical data.

5. Conclusion

This study presents the first report of multiscale mechanical comparison between collagen I hydrogels and myocardial ECM hydrogels with different crosslinking degrees and stretches applied. The findings offer insights into developing novel in vitro platforms with tunable mechanical properties.

Figure 1. A) Schematics of the stretching device compatible with AFM. B) Photograph of the stretching device with collagen I hydrogel. C) Micromechanical properties of collagen I hydrogels under strain with different crosslinking degrees. **, * statistical significance compared to 0% stretch. $ statistical significance compared to different crosslinking degrees. D) Micromechanical properties of the collagen I and myocardial ECM hydrogels. E) Rheometry of the uncrosslinked collagen I and myocardial ECM hydrogels.

6. References

1. Jorba I, Gussenhoven S, van der Pol A, et al. Steering cell orientation through light-based spatiotemporal modulation of the mechanical environment. Biofabrication. 2024;16(3):10.1088/1758-5090/ad3aa6. Published 2024 Apr 16. doi:10.1088/1758-5090/ad3aa6.
2. Sanz-Fraile H, Herranz-Diez C, Ulldemolins A, et al. Characterization of Bioinks Prepared via Gelifying Extracellular Matrix from Decellularized Porcine Myocardia. Gels. 2023;9(9):745. Published 2023 Sep 13. doi:10.3390/gels9090745.

This work reports the development of the first volumetric 3D printer in Spain dedicated to tissue engineering applications. The system, based on tomographic reconstruction, enables the rapid fabrication of complex biological structures in a single step, avoiding the limitations of traditional layer-by-layer methods that often reduce cell viability. Built with a 60x40x40 cm configuration, the printer was validated using GelMA solutions with LAP as photoinitiator, confirming their suitability for light-based volumetric printing. By offering an accessible and customizable alternative to costly commercial equipment, this achievement opens new opportunities for research and innovation in regenerative medicine.

Engineering functional livers is critical to alleviate the global shortage of donor organs. Reconstituting an organized biliary tree within a bioengineered liver remains a major challenge, requiring a cell source capable of large-scale expansion and precise differentiation. Here, we demonstrate that human intrahepatic cholangiocyte organoids (ICOs) can reconstruct a functional biliary network within decellularized rat liver scaffolds. ICOs derived from fresh or cryopreserved hepatocytes were expanded in spinner flasks, dissociated, and delivered via the bile duct using an alternating high- and low-pressure injection strategy to ensure coverage of large and small intrahepatic ducts. Following 7 days of perfusion culture, ICOs adhered to the scaffold, displayed reduced Ki67 expression, and maintained ductal compartmentalization without parenchymal invasion. High-pressure pulses promoted continuous biliary-like structures spanning multiple ductal levels. Immunofluorescence confirmed cholangiocyte-specific markers, and mature gene expression (CK19, CK7) was upregulated. This approach establishes ICOs as a scalable cell source for biliary reconstruction and introduces a pressure-controlled seeding method that preserves anatomical fidelity. Our findings represent a critical step toward generating bioengineered livers with functional biliary networks and support translation to large-animal models for transplantation applications.

La mecanotransducción es el fenómeno mediante el cual las células transforman estímulos mecánicos en respuestas bioquímicas a través de mecanosensores, lo que regula procesos como la proliferación y diferenciación celular. Las cadherinas son proteínas de adhesión celular que actúan como mecanosensores, y su estimulación puede influir en la regulación de vías de señalización como la vía Wnt y la vía Hippo, siendo uno de los efectores clave de esta última las proteínas YAP/TAZ. La estimulación de las cadherinas y otros factores, como la dureza de la matriz extracelular, juegan un papel fundamental en la activación de YAP y por tanto en esta vía de proliferación.

Se han desarrollado diversas herramientas necesarias para el estudio de la activación transitoria y controlada de YAP mediante magnetogenética a través de micropartículas magnéticas funcionalizadas con fragmentos de E-cadherina y campos magnéticos. Así mismo, se ha optimizado la producción de hidrogeles de poliacrilamida como sustrato blando para el estudio de la actividad de YAP y se ha generado una línea celular reportera de su actividad mediante expresión de Luciferasa.

Se han obtenido distintos tipos de micropartículas magnéticas funcionalizadas con fragmentos de E-cadherina que presentan diversas afinidades a la E-cadherina presente en la membrana celular. También se demostró que la siembra en los hidrogeles de las células reporteras generadas producían una disminución de la actividad basal de YAP debido a la dureza del sustrato, concluyéndose así que este factor deberá tenerse en cuenta en el futuro estudio de la modulación de esta vía de proliferación.

Trabajo comparativo entre tres diferentes metodos de medicion del desgaste, en cabezas acetabulares de UHMWPE, en implantes de cadera. Se busca cuantificar si la medicion mediante los tres metodos (micrometro, software y codigo python) tienen correlacion y arrojan resultados similares, de forma que puedan ayudar a una mejor cuantfiicacion del desgaste, asi como proporcionar resultados mas fiables y rapidos que de la manera convencional.

The digital transformation of healthcare has led to an unprecedented growth in the volume and heterogeneity of clinical data. In maternal-fetal medicine, this wealth of information offers new opportunities to improve outcomes for mothers and babies, but it also brings major challenges in terms of organization, integration, and quality. Rapid advancement in this field requires large-scale, well-structured datasets and tools that allow all actors to collaborate efficiently.
This work presents the development of a comprehensive platform that integrates multimodal data management and curation along with a wide range of tools for data analysis, thereby providing an agile environment that accelerates the achievement of improved research outcomes.
This platform was built with a modular architecture, enabling flexibility and scalability across different research domains. It includes a data pipeline that supports the ingestion, validation, anonymization, and secure storage of various types of clinical data —ranging from medical images (e.g. US, MRI), biosignals (e.g. ECG), and structured tables (e.g. sociodemographic or laboratory data). It supports interoperability through the adoption of domain standards (file formats, ontologies), and facilitates collaboration through role-based access and intuitive web-based interfaces. The system supports both on-premise and cloud deployments and ensures compliance with data protection regulations by design.
Our platform is currently in active use in 12 ongoing maternal-fetal research projects by over 40 users, including clinicians and researchers. Clinicians can easily explore the data, perform quality control and take measurements through intuitive web-based interfaces, while researchers benefit from standardized data inputs and can deploy, test, and refine their solutions directly within the environment.
The system already includes 6 specialized user interfaces with different components that support the mentioned key tasks, plus others such as graphical exploration and interpretation of research outcomes, patient browsing and follow-up, and risk assessment for specific clinical conditions.
The complete platform infrastructure has been successfully deployed in two major cloud services as well as on-premise at two leading maternal-fetal medicine centers in Barcelona (with further deployments in progress) and has already processed more than 400,000 clinical data samples, demonstrating its robustness and scalability in real-world research environments.
In conclusion, the proposed platform is tailored to the needs of maternal-fetal medicine research and tackles the challenges of managing multimodal, multi-source clinical data. By fostering collaborative development of new research tools, aligned with current regulatory requirements, this integrated environment accelerates the delivery of reliable innovations in maternal-fetal healthcare.

Wall Shear Stress (WSS) is a key biomechanical factor in the progression of aortic diseases such as aneurysms and dissections. While Computational Fluid Dynamics (CFD) is the standard for WSS estimation, its high computational cost and complexity limit its routine use in clinical practice. The objective of this work is to develop and validate a deep learning (DL) framework capable of predicting high-resolution WSS maps directly from aortic geometry, providing a fast and accurate surrogate for CFD. Predicted WSS maps exhibit strong visual agreement with CFD references, accurately reproducing both the location and magnitude of key hemodynamic features. Good linear correlation is observed between predicted and ground-truth values, with a slight tendency to underpredict extreme WSS peaks.

1. Introduction

Hypoplastic left heart syndrome (HLHS) is a cyanotic congenital heart disease characterized by the underdevelopment of the left ventricle, aorta, and aortic arch [1]. This structural deficiency results in obstruction to blood flow from the left ventricular outflow tract, altering oxygen delivery throughout the fetal body and potentially impairing organ maturation. We studied this pathology using a computational model of fetal circulation.

2. Materials and Methods

We used our previously published 0D computational model of the fetal cardiovascular system [2], which described the vasculature using Windkessel components, and the ventricles and atria with a single-fiber model [3], with all parameters scaled to represent a 36-week fetus [4].

From its simulated mean blood flows, we estimated oxygen content assuming perfect mixing of blood, computing saturation and partial pressure using Hill’s equation [5]. To simulate placental diffusion, we used Erlich et al.’s [6] model, which estimates oxygen flux through a single villus using geometric features extracted from 3D imaging of the fetoplacental vasculature, assuming 3.5·10⁵ villi in the placenta [7]. Oxygen consumption in fetal organs was derived from the literature, with 57% assigned to the brain [8] and the rest distributed by organ volume [9].

To simulate HLHS, we minimized the left ventricle and mitral and aortic valve dimensions to near-zero, halved the aortic arch diameter, doubled the right ventricle size, enlarged the ductus arteriosus and the main pulmonary artery to 140% of their healthy diameters, and reduced the combined cardiac output to 90% of the healthy value [10]. Additionally, the number of villi was halved since the placenta in HLHS is usually inefficient [11], and organ consumption was reduced proportionally to the cardiac output variation. We compared our blood flows and oxygen saturations with literature MRI data [10].

3. Results

Our simulated blood flows and oxygen saturations are reported in Table 1. In healthy cases, part of the oxygen-rich blood from the left heart perfuses the brain, while the rest mixes with the lower-oxygen blood from the right heart in the descending aorta to supply the lower body. In HLHS, in contrast, the right ventricle supplies the entire body. The combination of complete blood mixing in the right heart and the reduced placental efficiency leads to globally lower oxygen saturation, especially in the brain.

Table 1. Blood flow and oxygen saturation in healthy and HLHS cases simulated (bold, left) compared to [10] (right). CO: cardiac output. PA: pulmonary artery. AA: ascending aorta. DA: descending aorta. UV: umbilical vein. SVC: superior vena cava.

4. Conclusions

Our fetal circulation model simulated blood flow and oxygen distribution in both healthy and HLHS fetuses. It could be used for studying other cyanotic congenital heart defects and birth transition, and as a basis for creating a growth model.

References

[1] J. A. Connor et al., Orphanet J Rare Di., 2:23, 2007

[2] I. Villanueva-Baxarias et al., PLoS Comput Biol, 21:5, 2025

[3] T. Arts et al., Am J Physiol Heart Circ Physiol, 288:4, 2005

[4] G. Pennati et al., Ann Biomed Eng, 28, 2000

[5] E. P. Hill et al., Am J Physiol, 222:3, 1972

[6] A. Erlich et al., Sci Adv, 5:4, 2019

[7] A. R. Clark et al., Interface Focus, 5:2, 2015

[8] L. Sun et al., Circulation, 131:15, 2015

[9] R. H. Luecke et al., Int J Bio-Med Comput, 27:2, 1995

[10] L. Sun et al., Ultrasound Obstet Gynecol, 5:4, 2021

[11] H. N. Jones et al., Placenta, 36:10, 2015

Background.

Left atrial appendage occlusion (LAAO) prevents thromboembolic events in atrial fibrillation patients with bleeding risk, but device-related thrombosis (DRT) remains a challenge.

Case summary.

A 78-year-old man with atrial fibrillation (CHA₂DS₂-VASc = 3) and prior intracranial hemorrhage underwent successful LAAO with a 22-mm Amplatzer Amulet. After stepping down from apixaban to aspirin, DRT developed. Thrombi resolved with apixaban but recurred upon its discontinuation. A multiscale model integrating hemodynamics, thrombus formation events and apixaban revealed fibrin buildup beneath the pulmonary ridge, matching clinical thrombus. Optimized placement alone prevented fibrin formation in the virtual model.

Discussion.

Covering the pulmonary ridge and low-dose DOAC have emerged as effective strategies to reduce DRT. This case links both mechanisms, showing recurrence due to incomplete ridge coverage and antithrombotic decision.

Take Home Messages.

Our high-fidelity model offers mechanistic insight about antithrombotic therapy, which cannot be assessed with flow-based simulations alone. The results presented provide proof-of-concept evidence of the model’s capabilities.

Left atrial appendage occlusion (LAAO) devices are effective in reducing stroke risk in atrial fibrillation (AF) patients, but device-related thrombosis (DRT) remains a serious complication. In-silico modeling provides a valuable tool for studying the underlying hemodynamic mechanisms, although the accuracy of predictions depends strongly on modeling assumptions. This study presents the first large-scale uncertainty quantification framework to systematically assess the sensitivity of computational fluid dynamics (CFD) simulations to variations in modeling settings for addressing DRT risk, including LAAO devices. A dataset of 50 AF patients was used, with two implanted occluders (Amplatzer Amulet and Watchman FLX) virtually positioned in two configurations relative to the pulmonary ridge. For each patient-specific case, seven modeling conditions were tested, including Newtonian versus non-Newtonian blood rheology, inclusion of the A-wave in the mitral profile, scaled outlet velocities, increased pulmonary vein pressure, and rigid versus dynamic wall motion. In total, 1,400 CFD simulations were performed. Preliminary analysis of a representative case showed that AF conditions led to reduced velocities near the device, while inclusion of the A-wave increased velocity magnitude and complexity. Outlet velocity scaling had the strongest influence on flow dynamics, particularly in recirculating regions around the device. In contrast, non-Newtonian rheology and pressure changes had minimal impact on global velocity fields, though localized effects were noted. These findings highlight the critical role of boundary condition assumptions in in-silico DRT risk prediction and underscore the importance of systematic sensitivity analysis for reliable clinical translation.

Critically ill patients often experience pain, which can disrupt cardiorespiratory function, cause emotional distress, and lead to long-term complications. Although several analgesic and sedative medications are available, they may have profound, even life-threatening effects. Optimal strategies must therefore balance pain relief with patient safety. Reinforcement learning (RL) provides a framework for formalizing therapy as a sequence of decisions and identifying optimal policies. Small datasets, simplified actions, and omission of patient mortality have limited prior applications to sedation and analgesia. Our research addresses these gaps by jointly considering patient well-being and safety.

We modeled the task as a partially observable Markov decision process (POMDP) defined over observations of patient experience, cardiorespiratory function, metabolism, and one-year survival. The action space included four continuous dosing signals: opioids, propofol, benzodiazepines, and dexmedetomidine. We defined 34 reward functions and analyzed their asymptotic properties. We derived a cohort of 42591 intensive care stays from the MIMIC-IV database, and latent representations of these stays were learned to approximate the POMDP. Finally, the TD3-BC algorithm was applied to train our policies (data were split 80/20 for training and evaluation).

Both theoretical and empirical analyses highlighted a strategy that improved patient safety and comfort. The proposed actions were associated with a reduction of mortality by up to 59% and mean pain levels by up to 45% per stay, restricting opioids and propofol while relying on dexmedetomidine. Consequently, our results extend previous work through a richer model, a more comprehensive dataset, and policies that jointly minimize pain and one-year mortality.

La Ingeniería Biomédica es un sector en plena expansión, pero todavía existe cierto desconocimiento de este sector y de sus profesionales, tanto entre los propios actores implicados como en la sociedad en general. Asimismo, el propio alumnado del Grado de Ingeniería Biomédica (GIBIO), podría beneficiarse de un mayor conocimiento del sector para cimentar su incipiente carrera profesional. En este artículo se presenta el planteamiento de un Proyecto de Innovación Educativa (PINNE) que tiene como objetivo el desarrollo de un sitio web de referencia en Ingeniería Biomédica en Navarra, que será diseñado, creado y mantenido por los alumnos del GIBIO. La metodología del proyecto es “aprender haciendo” (learning by doing), mediante varias etapas: (1) entrega de documentación, (2) reunión informativa, (3) reparto de tareas, (4) comienzo de las tareas, (5) reunión intermedia, (6) finalización de tareas, (7) exposición y (8) evaluación. Con este proyecto, se espera contribuir a la divulgación de la profesión de la Ingeniería Biomédica, difundir los productos y proyectos de empresas del sector, así como visibilizar los proyectos de Ingeniería Biomédica desarrollados en la Universidad y en los Institutos de investigación afines. De esta manera, se creará un espacio de referencia en el sector biomédico de Navarra, que permitirá crear sinergias para futuros proyectos en el marco de la Ingeniería Biomédica. Por otro lado, a los alumnos del GIBIO les permitirá conocer de primera mano el sector mediante un aprendizaje activo y experiencial, recabando el contenido necesario y desarrollando el sitio web en su totalidad.

The implementation of 3D surgical planning in hospital environments faces significant workflow challenges that limit its clinical adoption and effectiveness. This study addresses the critical need for workflow optimization in 3D surgical planning for hepatobiliary surgeries at Hospital Germans Trias i Pujol through systematic analysis and digital solution development. Using Lean Management principles and the Plan-Do-Check-Act cycle, a four-phase methodology was employed: analysis, design, implementation, and validation. Current inefficiencies were identified through direct observation, stakeholder interviews, and quantitative error tracking over one month. A comprehensive digital platform prototype was developed using React.js, Node.js, and MariaDB to address identified bottlenecks. Quantitative analysis revealed systematic problems affecting 100% of cases, with critical information gaps requiring an average of 23.75 minutes of additional search time per case. Biomedical engineers spent 27% of their time on administrative tasks. User acceptance evaluation of the developed web-based platform yielded exceptional results, with System Usability Scale scores averaging 92.0. This research demonstrates that successful 3D surgical planning integration requires not only advanced technologies but equally robust workflow management and collaborative digital tools.

It has been shown that the continuous delivery of low-magnitude (1–3 V/cm) alternating electric fields of moderate frequency (100–300 kHz) hinders cell proliferation [1]. In particular, these fields, referred to as Tumor Treating Fields (TTFs), have demonstrated clinical efficacy against glioblastoma, slowing disease progression when combined with chemotherapy. The mechanism of action of TTFs has long been proposed to involve dielectrophoretic forces that disrupt tubulin polymerization, thereby arresting cells in mitosis. However, this hypothesis has been challenged [2]. We hypothesize that, instead, their mechanism of action is mediated by electroporation.

Because electroporation outcomes depend more on electric field amplitude than on exposure duration [3], we propose that delivering slightly stronger—but still mild—fields in short AC bursts, which we term Tumor Treatment Alternating Currents (TTACs), may inhibit proliferation more effectively than conventional continuous TTFs while maintaining thermal safety.

1. Materials and Methods

U87 human glioblastoma cells were cultured and treated in rectangular chambers with platinum electrodes at opposite ends. TTF exposure consisted of a continuous sinusoidal field of 2 V/cm and 200 kHz. TTACs were generated as sinusoidal bursts using a function generator and high‑voltage amplifier. Importantly, TTACs and TTFs were matched for Specific Absorption Rate (SAR = 30 W/kg) to equalize energy deposition.

Phase‑contrast images of fixed fields of view were acquired immediately before initiating treatment, that is, 24 hours after seeding cells (day 1), and 48 h later (day 3). Cells were segmented with Cellpose, a generalist deep learning–based algorithm for cellular segmentation, and a Proliferation Index was computed as the percent change from Day 1 to Day 3. Data from six independent repetitions were collected and analyzed statistically. Group differences were assessed with a Kruskal–Wallis test followed by pairwise two‑sided Mann–Whitney U tests.

2. Results

Across a 48-h window, the Proliferation Index (percent change from day 1 to day 3) differed among groups (Kruskal–Wallis p = 1.63 × 10⁻⁶) (Figure 1). Both treatments reduced proliferation relative to Control: TTAC showed the largest reduction (median 107% [IQR 52–129]), TTF a moderate reduction (median 133% [IQR 126–141]), versus Control (median 161% [IQR 140–178]). Pairwise Mann–Whitney U tests confirmed lower proliferation for **TTAC vs Control (p = 4.2 × 10⁻⁶, **) and **TTF vs Control (p = 9.34 × 10⁻⁴, *), and indicated a difference between TTAC and TTF (p = 1.85 × 10⁻², *). Collectively, both exposures suppressed 48-h proliferation versus Control, with TTAC exerting the stronger effect.

Figure 1. 48‑h Proliferation Index (median and IQR) for Control, TTAC (i.e., pulsed alternating field therapy), and TTF (i.e., continuous alternating electric field therapy).

3. Conclusions

In these 72 hours in vitro assays, pulsed alternating fields of higher magnitude but with the same eneregy as conventional TTFs produced stronger antiproliferative effects in U87 cells. These preliminary results suggest that delivering prolonged, mild-intensity electric fields in short bursts may be more effective than continuous TTFs at inhibiting cell proliferation, highlighting a potential strategy for cancer therapy.

References

[1] E. D. Kirson et al., ‘Disruption of Cancer Cell Replication by Alternating Electric Fields’, Cancer Research, vol. 64, no. 9, pp. 3288–3295, May 2004, doi: 10.1158/0008-5472.CAN-04-0083.

[2] X. Li, K. Liu, H. Fang, Z. Liu, Y. Tang, and P. Dai, ‘Electrodynamic interaction between tumor treating fields and microtubule electrophysiological activities’, APL Bioengineering, vol. 8, no. 2, p. 026118, Jun. 2024, doi: 10.1063/5.0197900.

[3] S. N. Campelo, P.-H. Huang, C. R. Buie, and R. V. Davalos, ‘Recent Advancements in Electroporation Technologies: From Bench to Clinic’, Annu. Rev. Biomed. Eng., vol. 25, no. 1, pp. 77–100, Jun. 2023, doi: 10.1146/annurev-bioeng-110220-023800.

La evaluación precisa de la madurez cerebral en neonatos prematuros es esencial para anticipar complicaciones neurológicas y diseñar intervenciones adecuadas. Los métodos convencionales, como el aEEG y la escala de Burdjalov, presentan limitaciones por su subjetividad y susceptibilidad a artefactos. Este estudio propone la Entropía Multiescala (MSE-10) como marcador cuantitativo y objetivo. Se analizaron 470 horas de EEG de neonatos entre 28 y 42 semanas de edad gestacional, identificando periodos de sueño tranquilo anotados por un neurofisiólogo. A partir de segmentos filtrados de 100 segundos, se calculó la MSE-10 mediante Fast-SampEn en MATLAB. Los resultados evidenciaron un aumento consistente de la MSE-10 con la edad posconcepcional, con alta capacidad discriminativa (AUC > 0.9). Además, permitió detectar cuatro casos subclínicos de alteraciones neurológicas que no eran visibles en el aEEG. En conclusión, la MSE-10 constituye un marcador robusto y objetivo para monitorizar el desarrollo cerebral neonatal, con gran potencial clínico.

La médula espinal humana puede verse afectada por diversas patologías que generan alteraciones en las funciones motoras y sensoriales, dificultando considerablemente su diagnóstico debido a que muchas veces no existen cambios estructurales visibles mediante las técnicas de imagenología tradicionales, como la resonancia magnética y la tomografía computacional. En años recientes, la espectroscopia funcional de infrarrojo cercano (fNIRS) ha demostrado ser una herramienta prometedora para estudiar el acoplamiento neurovascular de la médula espinal, ofreciendo información dinámica sobre cambios en las concentraciones de oxihemoglobina (O2Hb) y desoxihemoglobina (HHb). En este contexto, se utilizó una base de datos conformada por 241 registros obtenidos durante la aplicación de estímulos en un nervio periférico, no nocivo e indoloro, en distintos individuos.

Con el objetivo de facilitar y optimizar el análisis de estos datos, se diseñó y desarrolló una interfaz gráfica interactiva, accesible de manera remota, para explorar patrones hemodinámicos que puedan reflejar alteraciones funcionales espinales. La interfaz permite filtrar las señales según variables demográficas como edad, índice de masa corporal (IMC) y sexo, así como visualizar respuestas individuales por canal y comparar señales ante diferentes estímulos. Además, ofrece la posibilidad de exportar métricas relevantes para análisis clínicos e investigaciones posteriores. Gracias a esta herramienta, se logró identificar con eficacia registros con artefactos o señales atípicas que podrían tener relevancia clínica, demostrando así su potencial para mejorar el diagnóstico precoz y el seguimiento funcional de pacientes con alteraciones medulares.

En condiciones normales el impulso eléctrico que recorre el axón de una unidad motora llega a sus ramas terminales y genera potenciales de acción en las fibras musculares inervadas por el axón. En algunas situaciones patológicas la trasmisión puede fallar y no dispararse el potencial de acción en alguna de las fibras. Es el bloqueo de transmisión neuromuscular, fenómeno que se consigna y recuenta manualmente en los llamados estudios de “jitter”, que analizan el estado de la unión neuromuscular a partir de trenes de potenciales de unidad motora extraídos de señales de electromiografía de aguja.

En este trabajo se presenta un método para la detección automática de bloqueos en trenes de PUMs, basado en el agrupamiento (“clustering”) de picos de PUMs. El método ha sido probado con un conjunto de 36 señales, 24 de las cuales presentaban bloqueo. La detección resultó correcta en todos los casos excepto en una de las señales, en las que se produjo un falso positivo (sensibilidad de 1.0 y especificidad de 0.917). También se analizaron bloqueos en los 2102 PUMs individuales que conformaban los anteriores trenes, obteniéndose asimismo resultados de detección muy satisfactorios: sensibilidad de 0.873 y especificidad de 0.973. Pese a que el conjunto de trenes analizados ha sido reducido y debe ser ampliado en futuros estudios, los resultados permiten apreciar la fiabilidad del método automático y su utilidad clínica.

Diabetic foot ulcers (DFUs) are among the most frequent complications of diabetes mellitus (DM), presenting as chronic wounds with challenging healing processes. Conventional treatments, typically performed in outpatient clinics, often have limited effectiveness and require prolonged periods. As an innovative alternative, the use of latex sheets combined with phototherapy—via the Rapha® device—has shown promising results in tissue regeneration. Accurate assessment of DFUs is crucial for appropriate treatment but is hindered by variability in clinical evaluations. To address this limitation and the lack of robust databases in the literature, this study developed a dedicated database comprising clinical information, measurement data, and manually segmented images from specialized Brazilian diabetic foot outpatient clinics. An adapted invasive measurement protocol was initially tested on a mannequin to standardize weekly procedures. Essential data such as the Texas University classification, wound dimensions, exudate, and borders were collected by experienced nurses. The database integrates both tabular and visual information, supporting future research in automatic classification using machine learning and image processing techniques. Additionally, quantitative analyses of wound progression demonstrate the effectiveness of the Rapha® device and the metrological relevance of the database for clinical applications.

Light propagation and scattering through diffusive media such as

oral fluids, enables the implementation of Spatial Light Interference

Microscopy (SLIM) advanced optical technique to examine

the sample’s surface morphology, bulk composition, i.e., dry mass,

and optical properties, among others. Interference microscopy, aid

by sample’s contrast enhancement and beams’ phase matching, is

an appealing possibility for Artificial Intelligence (AI) improved

accuracy determinations when quantitatively imaging microscopic

objects. This investigation is intended to provide insight through

Quantitative Phase Imaging (QPI) for bacterial infections diagnosis

inclinical trials. The spectroscopic test results show the inhibition

of the UV spectral portion, which is important for damage

avoidance of the Spatial Light Modulator (SLM). Preliminary results

also show the automation process implementation of static

systems. Phase-contrast images are also presented at three different

magnifications with two different phase ring sizes.

Este trabajo mide los efectos del error en la señal de variabilidad del ritmo cardíaco (HRV) sobre el índice pRRx como detector de fibrilación auricular (FA). Este índice mide la proporción de intervalos RR en los que la variabilidad respecto al segmento anterior es menor que una duración x. Para hacerlo parte de señales de electrocardiograma de tres bases de datos públicas que contienen señales de pacientes con FA, ritmo sinusal normal y otros ritmos que no son FA. Después de obtener resultados similares a los de otros trabajos previos en las señales de HRV originales, limpias, se les han añadido distintos niveles de errores aleatorios para cuantificar la degradación del rendimiento de la detección de FA.

Los resultados en el índice pRR50 muestran como errores en la señal HRV de hasta un 5% del ritmo cardíaco provoca una reducción de la especificidad del 85% al 75%, y de exactitud del 90% al 88%, y errores del 10% llegan a degradaciones del 26% en exactitud. Los errores hasta un máximo constante, independiente del ritmo cardíaco, han mostrado degradaciones muy pequeñas hasta los 25 milisegundos de error, y del 10% de especificidad con 50 milisegundos. El estudio ha abarcado diversos valores de la duración x del pRRx, y se ha observado que el punto óptimo del valor x y del umbral de clasificación cambia con el nivel de ruido.

Desconocemos estudios previos de los efectos de la calidad de la señal en el índice pRRx como detector de FA.

En el estudio de modelos biomecánicos, el equilibrio representa un problema complejo debido a la indeterminación de las fuerzas durante la bipedestación. Para resolver este problema, es esencial disponer de las fuerzas de reacción del suelo (GRF) generadas en cada pie. En el presente trabajo se propone una metodología para determinar estas fuerzas durante la bipedestación cuando solo se dispone de una plataforma dinamométrica para realizar mediciones. Para ello, se propone un método analítico para determinar la distribución de las GRF verticales y la localización del centro de presión (CoP) independiente en cada pie y se han entrenado varios modelos de redes neuronales (NN) con el fin de mejorar los resultados obtenidos por ésta. Para entrenar y validar experimentalmente estos métodos, se ha construido un dispositivo de desarrollo propio que permite obtener simultáneamente las GRF verticales y las localizaciones de los CoP de cada pie al mismo tiempo que la GRF y la localización global del CoP obtenido a partir de una única plataforma dinamométrica. Los resultados obtenidos alcanzan un error de posición de los CoP inferior al 8% y de las GRF del 2% para los casos contemplados en los ensayos.

El estudio y monitorización de las bioseñales asociadas a las funciones del sistema nervioso contribuyen a la mejora de diagnósticos precoces y el tratamiento de diversas enfermedades mediante herramientas y técnicas de soporte y ayuda a la decisión. Las patologías que afectan al sistema nervioso autónomo (SNA) destacan por tener el mayor impacto sobre la calidad de vida del paciente por su influencia sobre el sueño o la capacidad respiratoria entre otras \cite{health-neurological}. El análisis comprensivo de la respuesta del sistema nervioso a diferentes estímulos y situaciones permite la identificación de anomalías sobre los distintos procesos fisiológicos que subyacen. Es por ello que el modelado de estas bioseñales contribuye al desarrollo de tecnologías médicas aplicables a terapias sanitarias mediante el acercamiento e interpretación de su comportamiento

Integrating domain knowledge from physics, biology, and medical expertise has the potential to significantly enhance diagnostic accuracy and contribute to more personalized clinical diagnoses and decision-making. A successful clinical translation of artificial intelligence systems in medical imaging relies on the development of trustworthy algorithms capable of capturing the complexity and variability inherent in real-world data. Incorporating expert medical knowledge, the physical principles underlying imaging modalities, and biological insights into artificial intelligence models plays a crucial role in embedding prior knowledge that complements purely data-driven approaches. This synergy can improve learning robustness, generalization, and interpretability. In this work, we review the forms of domain knowledge that can be incorporated into artificial intelligence for medical imaging applications and discuss their impact on advancing the field toward clinically robust solutions.

La silicosis, una grave enfermedad pulmonar ocupacional que resulta de la inhalación de sílice cristalina, continúa siendo un problema de salud global significativo, exacerbado por el creciente uso de piedra artificial de alto contenido de sílice. Los métodos de diagnóstico radiológico actuales para la silicosis a menudo están limitados por su baja sensibilidad en las etapas tempranas y la interpretación subjetiva. Este estudio investiga la aplicación del aprendizaje profundo para la detección y estadificación automatizada de la silicosis usando imágenes de rayos X de tórax. Usando un conjunto de datos completo de radiografías de tórax de trabajadores expuestos a la sílice, se ha desarrollado un enfoque basado en aprendizaje profundo. La detección de anomalías se realizó usando un modelo Mamba-YOLOvX, seguida por la clasificación con YOLOv8. La fase de segmentación inicial demostró un buen rendimiento, permitiendo un proceso de cribado efectivo con una precisión del 81,9%, una sensibilidad del 88,2% y una métrica F1 del 85,0% en la detección de fibrosis masiva progresiva. Las fortalezas inherentes de la arquitectura Mamba, como su capacidad para modelar dependencias de largo alcance y aplicar atención espacial eficientemente, mejoraron significativamente la robustez del análisis de la imagen. Estos hallazgos subrayan el potencial del modelo Mamba-YOLOX para crear herramientas precisas y escalables, que podrían contribuir a la detección temprana y al manejo clínico de la silicosis.

Los productos sanitarios (PS) desempeñan un papel fundamental en los sistemas de salud, al contribuir al bienestar de los pacientes y a la calidad de la atención. Asegurar su seguridad refuerza la eficacia y confianza en los servicios sanitarios, impactando positivamente en pacientes y profesionales.

En España, la vigilancia de la seguridad de los PS implica a múltiples agentes: fabricantes, la AEMPS, los Puntos de Red autonómicos y la organización interna de los centros sanitarios. Los hospitales deben contar con sistemas eficaces para gestionar alertas e incidentes de forma oportuna, activando las medidas correctivas necesarias. Los protocolos actuales de gestión del hospital son ineficientes y dependientes del envío manual de correos electrónicos. En este trabajo de investigación se desarrolla una aplicación web para la gestión de alertas e incidentes relacionados con la seguridad de los PS.

La herramienta, AlertaPS, ha sido diseñada siguiendo principios de usabilidad y diseño centrado en el usuario. Su arquitectura modular, basada en Django, permite una gestión completa y mejora la comunicación entre servicios. Tras su despliegue, la usabilidad y funcionalidad han sido validadas y se ha lanzado un piloto para su integración en rutina clínica.

El manejo diario de la diabetes tipo 1 representa un desafío complejo, donde se requiere un equilibrio constante entre la administración de insulina, la ingesta de alimentos o la actividad física, entre otros factores. Los pacientes deben ajustar frecuentemente las dosis de insulina en función de estos parámetros para mantener sus niveles de glucosa en un rango objetivo. Para ello, se utilizan actualmente protocolos estandarizados y ajustes empíricos, basados en la experiencia del equipo médico y en el seguimiento continuo de los datos del paciente. Sin embargo, estos métodos no siempre logran una personalización óptima, lo que puede traducirse en una adaptación lenta a los cambios en las necesidades individuales. El objetivo de este trabajo es desarrollar un modelo de aprendizaje automático capaz de predecir de forma personalizada los parámetros clave para la toma de decisiones en el contexto de la ingesta en pacientes con páncreas artificial. El modelo ha obtenido un error porcentual absoluto medio de 3,89 % para el subgrupo con mejor desempeño, mientras que en el peor caso el valor registrado fue de 7,67 %. Este trabajo responde a una necesidad clínica de personalización del tratamiento, una tendencia creciente que busca adaptar los tratamientos a las características específicas de cada paciente.

El síndrome de insuficiencia límbica es una patología de la superficie ocular que provoca ceguera corneal por pérdida de células madre limbares. El trasplante de células madre mesenquimales (MSCs) ha demostrado una eficacia parcial en su tratamiento gracias a su capacidad inmunomoduladora, migratoria y regenerativa. Dicha eficacia se ha tratado de potenciar en este trabajo mediante la modificación genética de MSCs humanas de tejido adiposo con CXCR4 o TSG-6. La transducción de las MSCs no alteró su inmunofenotipo ni su capacidad de diferenciación multilinaje. Ensayos de migración y regeneración epitelial en modelos in vitro de inflamación demostraron la mejora en la capacidad regenerativa y migratoria de las MSCs gracias a la sobreexpresión de TSG-6 o CXCR4. El análisis de sobrenadantes mediante tecnología Luminex reveló que IL-6, MCP-1, RANTES y VEGF-A podrían estar implicados en la modulación de la efectividad del tratamiento.

En el contexto de la ablación cardíaca por radiofrecuencia (RF), se ha propuesto que la colocación del parche dispersivo (PD) de forma concordante con la orientación del electrodo de ablación puede aumentar el tamaño de la lesión térmica. El objetivo de este estudio es investigar cómo el tamaño corporal individual puede modular la magnitud de este efecto. Se desarrollaron tres modelos computacionales que representan diferentes tamaños corporales. Se simuló una ablación con catéter irrigado aplicando un pulso de 30 W durante 30 segundos sobre la superficie endocárdica de la pared anterior. Se compararon los tamaños de la lesión entre dos configuraciones del PD: posición anterior (concordante) vs. posición posterior (discordante). El tamaño de la lesión fue significativamente mayor con la colocación concordante. Esta diferencia disminuyó significativamente con el aumento del tamaño corporal: desde 0.65 ± 0.08 mm en un modelo porcino de 35 kg hasta 0.51 ± 0.06 mm en el modelo humano. En conclusión, el tamaño corporal tiene una influencia modesta en el efecto de la posición del PD sobre el tamaño de la lesión de RF. La posible ventaja de una posición concordante podría ser más significativa en individuos con menor volumen corporal.

La diabetes mellitus tipo 1 (DM1) requiere un control estricto con insulina. En mujeres, las fluctuaciones hormonales del ciclo menstrual modifican la sensibilidad a la insulina (SI), generando variaciones en el control glucémico que los protocolos actuales no consideran. Este trabajo plantea integrar esa variabilidad en un simulador de pacientes virtuales para valorar si ajustar la terapia insulínica según la fase menstrual mejora el control glucémico.

La hiperglucemia es frecuente en pacientes hospitalizados tratados con glucocorticoides y representa un reto clínico por su variabilidad metabólica y riesgo de complicaciones. Los protocolos basados en mediciones capilares puntuales (POC) presentan limitaciones importantes. La monitorización continua de glucosa (CGM) surge como una alternativa prometedora. Este estudio, en el marco del proyecto MANGLAI (UdG-Dexcom y HUGTiP), tiene como objetivo diseñar y validar un protocolo computacional con CGM y un sistema de soporte a la decisión clínica (DSS) para pacientes hospitalizados no críticos bajo tratamiento con glucocorticoides.

La adherencia a estilos de vida saludables es crucial para la promoción de la salud cerebral, pero su heterogeneidad entre individuos limita la efectividad de las intervenciones generalizadas. Este estudio ha identificado y caracterizado perfiles de adherencia multidominio en 1.545 participantes de la cohorte Barcelona Brain Health Initiative mediante análisis de clustering mediante K-Medoids y distancia de Gower. El análisis reveló cuatro arquetipos conductuales estables (índice Jaccard = 0.926): adherencia general baja, adherencia selectiva con foco en dieta, adherencia selectiva con foco en actividades estructuradas, y adherencia holística con énfasis social. Estos perfiles mostraron asociaciones significativas con variables sociodemográficas y psicológicas: el perfil de baja adherencia se asoció con menor edad, nivel educativo, inestabilidad familiar y un perfil psicológico de vulnerabilidad (alto neuroticismo, baja autoeficacia y propósito vital), mientras que los perfiles de alta adherencia se relacionaron con mayores recursos psicológicos protectores. Los resultados demuestran que la adherencia sigue patrones conductuales ligados a factores contextuales y psicológicos, proporcionando una base empírica para el diseño de intervenciones personalizadas que aborden las barreras específicas de cada perfil de usuario y optimicen así la efectividad de las estrategias de promoción de la salud cerebral