Mena Acosta Byron Hernán

Background: Multivariable early-warning models can integrate physiological signals and missingness patterns, but added complexity may not provide meaningful value over simple predictors. We evaluated whether a single prespecified L2-regularized logistic-regression model using physiological and missingness predictors improved the prediction of sustained hypotension beyond recent mean arterial pressure (MAP). Methods: We conducted a retrospective pilot prognostic-model study using the MIMIC-IV Clinical Database Demo on FHIR 2.1.0. Predictors were derived from the preceding 12 h, and sustained hypotension was defined as MAP < 65 mmHg for at least two consecutive hours during the following six hours. A single prespecified L2-regularized logistic-regression model using eight physiological or derived predictors and ten missingness-related predictors was compared with six-hour mean MAP. Performance was estimated using patient-grouped nested cross-validation with five outer and three inner folds. The primary comparison was the difference in area under the precision–recall curve (AUPRC); advancement required an absolute gain of ≥0.020. Paired uncertainty was estimated using 1000 patient-cluster bootstrap replicates. A prespecified sensitivity analysis recalculated AUPRC after assigning equal total weight to each patient, irrespective of the number of eligible windows contributed. Results: Ninety-two patients contributed 8970 eligible windows, of which 1774 (19.8%) were positive. The candidate achieved an AUPRC of 0.629 versus 0.622 for the comparator and an AUROC of 0.845 versus 0.842. The AUPRC difference was +0.0075 (95% bootstrap interval, +0.0005 to +0.0171). The prespecified patient-equal sensitivity analysis yielded a larger delta AUPRC point estimate of +0.0156; both the patient-equal and window-weighted estimates remained below the +0.020 advancement margin. In a post hoc analysis, 81.6% of positive windows had observed MAP below 65 mmHg during t − 5 through t. Among windows without observed recent hypotension, the delta AUPRC was +0.0030 (95% interval, −0.0288 to +0.0238). The 8970 prediction windows were overlapping repeated observations nested within 92 patients and should not be interpreted as independent samples. Conclusions: In this pilot internal-validation study, the single prespecified L2-regularized logistic-regression model using eight physiological or derived predictors and ten missingness-related predictors produced a small improvement in discrimination over six-hour mean MAP but did not meet the prespecified development-advancement margin. These findings apply only to this model specification, dataset, and outcome and do not establish the performance of multivariable vital-sign models generally. Neither strategy was validated for clinical use.

Introducción: Los modelos de lenguaje grande (LLM) se investigan como apoyo al diagnóstico, la comunicación y la toma de decisiones clínicas. Su rendimiento depende de la tarea, el modelo y el entorno de evaluación, lo que limita la extrapolación a países de ingresos medios. Objetivo: Analizar la evidencia sobre el rendimiento y la seguridad clínica de los LLM y otros sistemas de inteligencia artificial, y discutir su aplicabilidad en países de ingresos medios, con énfasis en Ecuador. Métodos: Revisión narrativa con búsqueda estructurada en PubMed y Google Scholar realizada en mayo de 2026, orientada a publicaciones de 2022 hasta esa fecha. Un autor seleccionó las fuentes y extrajo los datos. Se sintetizaron 12 publicaciones sobre rendimiento o implementación, diferenciando evaluaciones de LLM, otros sistemas de IA y documentos sin resultados clínicos comparativos. No se efectuaron metaanálisis ni evaluación formal del riesgo de sesgo. Resultados: En escenarios de obstetricia y ginecología, el conjunto de LLM evaluados alcanzó una exactitud global del 73,75 %, frente al 65,35 % de residentes. En contraste, una evaluación de dolor abdominal encontró un rendimiento inferior al de médicos. Las valoraciones de daño potencial de respuestas de Med-PaLM 2 no demostraron seguridad equivalente en la atención de pacientes. Los estudios de comunicación, pronóstico y soporte automatizado midieron desenlaces distintos y no permiten una estimación conjunta. La conectividad, la interoperabilidad, la validación local y los costos condicionan la aplicabilidad en Ecuador. Conclusiones: Los LLM muestran potencial como apoyo supervisado en tareas específicas, pero la evidencia revisada no establece superioridad clínica general ni seguridad equivalente a especialistas. Su implementación requiere evaluación prospectiva local, protección de datos, capacitación y vigilancia del desempeño.

Liquid biopsy has rapidly emerged as one of the most promising innovations in precision oncology by providing minimally invasive access to tumor-derived biomarkers throughout the clinical course of cancer. This review analyzes the current evidence regarding the clinical applications of liquid biopsy beyond early cancer detection, emphasizing its role in molecular profiling, therapeutic selection, treatment monitoring, minimal residual disease assessment, identification of resistance mechanisms, and evaluation of tumor heterogeneity in patients with solid tumors. A comprehensive literature review was conducted using major biomedical databases, selecting high-quality publications focused on circulating tumor DNA, circulating tumor cells, extracellular vesicles, circulating RNA, and other emerging biomarkers. The available evidence demonstrates that circulating tumor DNA currently represents the most clinically validated biomarker because of its high analytical performance and increasing incorporation into precision oncology protocols. Liquid biopsy has shown particular utility in lung, colorectal, breast, and prostate cancers, where molecular information directly contributes to personalized treatment decisions. Nevertheless, several barriers continue to limit universal implementation, including high costs, limited molecular infrastructure, methodological variability, and unequal access to advanced genomic technologies, particularly in developing countries. Ongoing technological improvements, international collaboration, and methodological standardization are expected to further expand the clinical applications of liquid biopsy and facilitate its incorporation into routine oncology practice. Overall, liquid biopsy represents a transformative strategy that complements conventional tissue biopsy and supports more personalized, dynamic, and evidence-based management of solid tumors.

Brain-computer interfaces (BCIs) have emerged as one of the most significant technological innovations in modern neurosurgery, integrating neuroscience, biomedical engineering, artificial intelligence, and clinical neurology to establish direct communication between the brain and external devices. This review provides a comprehensive analysis of the current scientific evidence regarding the neurophysiological foundations, technological evolution, clinical applications, limitations, and future perspectives of BCIs in neurosurgical practice. A qualitative documentary review was conducted following the Scientific Method, using peer-reviewed literature retrieved from internationally recognized databases, including PubMed, Scopus, Web of Science, ScienceDirect, IEEE Xplore, and Google Scholar. The analyzed evidence indicates that current clinical applications are primarily focused on motor restoration, neurorehabilitation, communication recovery, speech neuroprostheses, and adaptive neuromodulation for neurological disorders such as spinal cord injury, stroke, Parkinson's disease, amyotrophic lateral sclerosis, and drug-resistant epilepsy. Recent advances in machine learning, implantable electrodes, wireless technologies, and closed-loop therapeutic systems have substantially improved neural decoding performance and expanded the potential clinical utility of BCIs. However, important challenges remain regarding long-term signal stability, biocompatibility, surgical safety, economic accessibility, regulatory frameworks, and equitable implementation across different healthcare systems. The available evidence suggests that continued multidisciplinary collaboration and technological innovation will facilitate the progressive integration of BCIs into routine neurosurgical care, supporting personalized therapeutic strategies aimed at improving functional recovery, communication, and quality of life in patients with complex neurological disorders.

Photon-Counting Computed Tomography (PCCT) has emerged as one of the most significant innovations in diagnostic imaging by introducing photon-counting detector technology capable of overcoming several intrinsic limitations of conventional energy-integrating detector computed tomography. This review analyzes the technological principles underlying PCCT, its integration with artificial intelligence, current clinical applications, implementation challenges, and future perspectives in precision medicine. A structured review of the contemporary scientific literature was conducted using internationally recognized biomedical databases and peer-reviewed publications focusing on detector technology, spectral imaging, quantitative imaging biomarkers, radiation dose optimization, and multidisciplinary clinical applications. The reviewed evidence consistently demonstrated that PCCT provides superior spatial resolution, improved contrast-to-noise ratio, enhanced spectral imaging, quantitative tissue characterization, and substantial reduction of electronic noise and beam-hardening artifacts while maintaining lower radiation exposure than conventional CT systems. The greatest clinical benefits were observed in cardiovascular imaging, oncology, neurology, pediatric radiology, pulmonary imaging, abdominal imaging, and musculoskeletal applications. The integration of artificial intelligence further enhances image reconstruction, automated segmentation, lesion detection, radiomics, workflow optimization, and quantitative image analysis, facilitating increasingly individualized diagnostic assessment. Despite important challenges related to acquisition costs, infrastructure requirements, protocol standardization, and broader clinical implementation, the available evidence supports Photon-Counting Computed Tomography as a transformative technology with the potential to redefine diagnostic imaging and strengthen the development of precision medicine across diverse healthcare systems.