A5039641842- Text and Document Classification Technologies
- Machine Learning and Data Classification
- Multiple Sclerosis Research Studies
- Face and Expression Recognition
- Machine Learning and Algorithms
- Evolutionary Algorithms and Applications
- Algorithms and Data Compression
- Gene expression and cancer classification
- E-Learning and Knowledge Management
- Cutaneous Melanoma Detection and Management
- AI in cancer detection
- Neural Networks and Applications
- Metaheuristic Optimization Algorithms Research
- Web Data Mining and Analysis
- Machine Learning in Bioinformatics
- Bioinformatics and Genomic Networks
- Image Processing Techniques and Applications
- Telemedicine and Telehealth Implementation
- Knowledge Societies in the 21st Century
- Atrial Fibrillation Management and Outcomes
- Cell Image Analysis Techniques
- Music and Audio Processing
- Image Retrieval and Classification Techniques
- Autoimmune and Inflammatory Disorders Research
- Engineering and Information Technology
University of Córdoba
2017-2020
Instituto Maimónides de Investigación Biomédica de Córdoba
2018-2020
Asociación Colombiana de Gastroenterología
2019
Asociación Colombiana de Hematología y Oncología
2019
University of Holguín
2012-2016
Multi-target regression (MTR) comprises the prediction of multiple continuous target variables from a common set input variables. There are two major challenges when addressing MTR problem: exploration inter-target dependencies and modeling complex input-output relationships. This paper proposes neural network model that is able to simultaneously address these in flexible way. A deep architecture well suited for learning outputs designed, providing some flexibility relationships by sharing...
In the last decade several modern applications where examples belong to more than one label at a time have attracted attention of research into machine learning. Several derivatives k-nearest neighbours classifier deal with multi-la
Multi-label learning has become an important area of research owing to the increasing number real-world problems that contain multi-label data. Data labeling is expensive process requires expert handling. The annotation data laborious since a human needs consider presence/absence each possible label. Consequently, numerous modern may involve small labeled examples and plentiful unlabeled simultaneously. Active methods allow us induce better classifiers by selecting most useful data, thus...
There is an unmet need for reliable and sensitive measures better monitoring people with multiple sclerosis (PwMS) to detect disease progression early adapt therapeutic accordingly.
Cognitive impairment occurs in up to 70% of people with MS (pwMS) and has a large impact on quality life working capacity. As part the development smartphone-app (dreaMS) for monitoring disease activity progression, we assessed feasibility acceptance using cognitive games as assessment tools domains.We integrated ten dreaMS app. Participants were asked play these twice week 5 weeks. All subjects underwent battery established neuropsychological tests. User feedback was obtained via five-point...
Locally weighted regression allows to adjust the models nearby data of a query example.In this paper, locally method for multi-target problem is proposed.A novel way weighting based on gravitation-based approach presented.The process does not need decompose into several single-target problems.This can be used with any regressor as local provide target vector example.The proposed was assessed largest collection datasets publicly available.The experimental stage showed that performance...
Passive remote monitoring of patients with MS (PwMS) sensor-based wearable technologies promises near-continuous evaluation high ecological validity. Step counts correlate strongly traditional measures severity. We hypothesized that sleep and heart rate will yield complementary information.We recruited 31 PwMS age- sex-matched healthy volunteers (HV) as part the dreaMS feasibility study (NCT04413032). Fitbit Versa 2 smartwatches were worn for 6 weeks provided a total 25 features activity,...
The multi-target regression problem comprises the prediction of multiple continuous variables at same time using a common set input variables, and in last few years, this has gained an increasing attention due to broad range real-world applications that can be analyzed und er framework. complexity is higher than single-target one since target often have statistical dependencies, these dependencies should correctly exploited order effectively solve problem. Consequently, additional...
The definition of similarity metrics is one the most important tasks in development nearest neighbours and instance based learning methods. Furthermore, performance lazy algorithms can be significantly improved with use an appropriate weight vector. In last years, from multi-label data has attracted significant attention a lot researchers, motivated increasing number modern applications that contain this type data. This paper presents new method for feature weighting, defining metric as...
The study of problems that involve data examples associated with multiple targets at the same time has gained a lot attention in past few years. In this work, method based on gene expression programming for multi-target regression problem is proposed. This solves symbolic contexts, allowing construction model, without previous knowledge any its elements, fits set cases. Our proposal directly handles data, encoding individuals chromosome several genes, where each constructs mathematical...
The known set of genetic factors involved in the development several types cancer has considerably been expanded, thus easing to devise and implement better therapeutic strategies. automatic diagnosis cancer, however, remains as a complex task because high heterogeneity tumors biological variability between samples. In this work, long short-term memory network-based model is proposed for diagnosing from transcript-base data. An efficient method that transforms data into gene/isoform...
Introducción: el dabigatrán es un anticoagulante directo que actúa inhibiendo la trombina vinculada al coágulo. Aunque no necesario hacer monitoreo estricto de las concentraciones este, debido a su predictibilidad farmacocinética y farmacodinámica, existen métodos cualitativos cuantitativos para realizarlo; entre los cuantitativos, se encuentran plasmáticas fármacos medidas con tiempo coagulación ecarina diluido. El objetivo este trabajo fue medir parcial en pacientes anticoagulados por...