Models for the prevention of arthritis, natural language processing and wireless virtual sensing, first projects in the European 3-i ICT programme for attracting international talent
- The CITIC of the UDC hosted the presentation of the first three thesis projects, of the eight planned under this call for proposals.
- The 3-i ICT is the first programme awarded to a Galician institution within the European call MSCA-COFUND of the European framework programme H2020.
The ICT Research Centre (CITIC) of the University of A Coruña hosted on Tuesday 23 May the presentation of the international doctoral theses of the 3-i ICT (International, Interdisciplinary and Intersectoral Information and Communications Technology PhD Programme), an innovative programme for attracting pre-doctoral talent awarded to the UDC centre last year.
Specifically, three of the eight research projects planned under this plan, which offers extensive training in research and transfer skills and will provide a solid competitive advantage in the professional development of the participants, were presented. In addition to the doctoral students and researchers, the event was attended by the head of European projects at CITIC, Cristina Villar, and the director of CITIC, Manuel Penedo, who recalled that this is the first programme awarded to a Galician institution within the European call MSCA- COFUND of the European framework programme H2020.
Natural language processing
“Sequence Labelling Parsing for Applied Natural Language Processing” is the title of the thesis project presented by Muhammad Imran with Carlos Gómez and Margarita Alonso Ramos as director and co-director, respectively.
The aim of this research is to exploit the potential of sequence labelling analysis to improve practical Natural Language Processing (NLP) tasks in large configurations. To achieve this, different ways of integrating information from analysis tasks such as named entity recognition, aspect-based sentiment analysis and automatic text summarisation are explored. Thus, analytics information is used as label features and multi-task learning architectures are designed to perform both analytics and downstream tasks together. The ultimate goal is to improve the accuracy and quality of downstream PLN tasks using linguistic information without compromising efficiency.
Wireless virtual sensing
Dariel Pereira Ruisánchez is the author of the thesis project “Wireless virtual sensing fuere control applications” directed by Luis Castedo and co-directed by Francisco Javier Cuadrado.
This PhD project aims to overcome the current limitations of virtual sensing by introducing a new concept: wireless virtual sensing. This innovative idea seeks to take advantage of recent advances in wireless connectivity and edge computing to develop a network infrastructure suitable for virtual sensing.
The main focus of this thesis is on the design of a specific wireless network infrastructure and the optimisation of its transmission and computing resources. The design of the wireless network infrastructure and the optimisation of its resources are key aspects to achieve efficient and reliable virtual sensing.
Rheumatoid arthritis prediction models
Finally, PhD student Blanca Estela Monroy Castillo presented the thesis project “Flexible cure models in data science to predict sustained remission in rheumatoid arthritis”, with Ricardo Cao as director and Francisco Javier Blanco and Amalia Jácome as co-directors.
In this thesis project, the aim is to develop efficient statistical data analysis techniques to predict the remission of rheumatoid arthritis (AIRE) after treatment. For this purpose, cure models, non-parametric and semi-parametric methods for survival analysis, as well as functional data analysis techniques will be used. Cure models will be applied to determine the probability of point remission and sustained remission in patients with AIRE. In addition, methods will be employed in survival analysis to predict the latency time in those patients who do not achieve sustained remission.
To improve the accuracy of predictions, the use of functional data analysis techniques to define biomarker classifications is proposed. These will allow a more accurate assessment of the likelihood of timely and sustained remission in patients with AIRE.
The cure modelling approach, together with functional data analysis techniques, facilitates a comprehensive and efficient analysis of data to predict AIRE remission and time to recurrence in patients who do not achieve sustained remission.
< Return to news