News

June 3rd 2020 - Press release

New data mining tool to identify characteristics in patients with covid-19

This is a pioneering Data Analytics project in Spain, which will enable us to increase our knowledge of the evolution of COVID-19 and its interaction with patients with other pathologies based on the analysis of discharge reports. The initiative is the result of collaboration between Hospital del Mar, the technology company Bismart and Ferrer Laboratory and uses the artificial intelligence tool Bismart Folksonomy. In less than a month, a project has been launched with more than 34,000 hospital and emergency room discharges, 2,700 of which correspond to COVID-19 patients. This will allow an in-depth analysis of their evolution and the relationship of the infection with other basic pathologies.

Hospital del Mar, in collaboration with the technology company Bismart and Ferrer Laboratory, is leading a pioneering project in the country that uses data mining tools in conjunction with artificial intelligence in natural language processing to analyze and detect common characteristics in patients with COVID-19. To do this, the Bismart Folksonomy tool has been used, based on Microsoft Azure's cognitive services and which has already produced outstanding results in other collaborative projects.

Dr. Laia Sans, assistant doctor of the Nephrology Service of the Hospital del Mar and coordinator of the project, together with Dr. Julio Pascual, Director of Healthcare and Medical Area of the centre, and Dr. Jordi Martínez, Director of Innovation and Digital Transformation, has pointed out that the initiative allows to obtain an image of the characteristics of the patients treated in the hospital, without the need to create and feed a database. A work that can help to better understand the disease. "The study can help identify common characteristics in relation to history, previous treatments or treatments received during the disease, and relate them to its evolution", explains Dr. Sans. It is a tool that allows us to automatically obtain more information from the patients' medical records (more than 34,000 hospital and emergency room discharges have been included, 2,700 of which correspond to COVID-19 patients) and thus allows to better adapt the treatments to their reality. This type of information can be used much more quickly.

The project, which is in its initial phase, will make it possible to establish relations between the basic characteristics of patients and the resolution of the disease, as well as to see the evolution of patients with other pathologies. "We will be able to obtain results that allow us to relate basic characteristics with the resolution of the disease or the treatments administered. We will also be able to select those patients we have treated with COVID-19 and other pathologies to analyze their evolution in depth", explains Dr. Pascual. Many of the patients attended had an underlying pathology. For this reason, "this tool is not only of interest to professionals linked to the area of infectious diseases, but also to those in other specialties, such as cardiology, oncology and nephrology," he adds.

Digital transformation in health

"With this project we continue to advance in the use of textminig tools that help us to form a complete set of data, very useful in the approach of this disease," analyzed Dr. Jordi Martinez from the Direction of Innovation and Digital Transformation of the center. On the other hand, the general director of Ferrer in Spain, Ricardo Castrillo, assures that this project responds to Ferrer's mission to bring greater well-being to society through innovation and research. "The project that we are promoting with Hospital del Mar and Bismart Folksonomy reinforces our commitment to the digital transformation of health, to contribute to improving clinical results, reducing time in decision making and sharing aspects that represent a positive change in patients and society. For Ferrer, and even more so in the context of the current pandemic, it is essential to collaborate with health centres and to do so in an agile way".

For Jaume Santacana, CEO of Bismart, "the use of Folksonomy in this type of project allows us to reduce information analysis times and discover patterns and relationships between existing data, helping to improve the work of health professionals". The consulting firm Bismart maintains its commitment to the development of innovative solutions with special attention to the health sector. The collaboration with Hospital del Mar and Ferrer opens up a scenario of new opportunities in which the digital transformation will be key to improving healthcare.

The Bismart Folksonomy tool is based on the cognitive services of Microsoft Azure and incorporates the technique of folksonomy, a way of classifying terms according to how they are reproduced in documents, without the need to previously generate a master key to do so. It is an automatic system for managing large quantities of documents in real time, based on the labels and their frequency of appearance.