Autor: Naiara Muro Amuchastegui

Directores: Nekane Larburu Rubio (Vicomtech) Prof. Brigitte Séroussi (Universidad)

Universidad: Sorbonne Université

Fecha: 05.12.2019

Evidence-Based Medicine has been formalized as Clinical Practice Guidelines, which define workflows and recommendations to be followed for a given clinical domain. These documents were formalized aiming to standardize healthcare and seeking the best patient outcomes. Nevertheless, clinicians do not adhere as expected to these guidelines due to several clinical and implementation limitations. On one hand, clinicians do not feel familiar, agree with and or are unaware of guidelines, hence doubting their self-efficacy and outcome expectancy compared to previous or more common practices. On the other hand, maintaining these guidelines updated with the most recent evidence requires continuous versioning of these paper-based documents. Clinical Decision Support Systems are proposed to help during the clinical decision-making process with the computerized implementation of the guidelines to promote their easy consultation and increased compliance. Even if these systems help improving guideline compliance, there are still some barriers inherited from paper-based guidelines that are not solved, such as managing complex cases not defined within the guidelines or the lack of representation of other external factors that may influence the provided treatments, biasing from guidelines’ recommendations (i.e. patient preferences). Retrieving observational data and patients’ quality of life outcomes related to the provided healthcare during routine clinical practice could help to identify and overcome these limitations and would generate Real World Data representing the real population and going beyond the limitations of the knowledge reported in the Randomized Clinical Trials.
This thesis proposes an advanced Clinical Decision Support System for coping with the purely guideline-based support limitations and going beyond the formalized knowledge by analyzing the clinical data, outcomes, and performance of all the decisions made over time. To achieve these objectives, an approach for modeling the clinical knowledge and performance in a semantically validated and computerized way has been presented, leaning on an ontology and the formalization of the Decisional Event concept. Moreover, a domain-independent framework has been implemented for easing the process of computerizing, updating and implementing Clinical Practice Guidelines within a Clinical Decision Support System in order to provide clinical support for any queried patient. For addressing the reported guideline limitations, a methodology for augmenting the clinical knowledge using experience has been presented along with some clinical performance and quality evaluation over time, based on different studied clinical outcomes, such as the usability and the strength of the rules for evaluating the clinical reliability behind the formalized clinical knowledge. Finally, the accumulated Real World Data was explored to support future cases, promoting the study of new clinical hypotheses and helping in the detection of trends and patterns over the data using visual analytics tools.
The presented modules had been developed and implemented in their majority within the European Horizon 2020 project DESIREE, in which the use case was focused on supporting Breast Units during the decision-making process for Primary Breast Cancer patients management, performing a technical and clinical validation over the presented architecture, whose results are presented in this thesis. Nevertheless, some of the modules have been also used in other medical domains such as Gestational Diabetes guidelines development, highlighting the interoperability and flexibility of the presented work.


Parque Científico y Tecnológico de Gipuzkoa,
Paseo Mikeletegi 57,
20009 Donostia / San Sebastián (España)

+(34) 943 309 230

Zorrotzaurreko Erribera 2, Deusto,
48014 Bilbao (España)

close overlay

Las cookies de publicidad comportamental son necesarias para cargar el contenido

Aceptar cookies de publicidad comportamental