Almeida João Rafael, Monteiro Eriksson, Silva Luís Bastião, Pazos Alejandro, Oliveira José Luís
DETI/IEETA, University of Aveiro, Portugal.
Department of Computation, University of A Coruña, Spain.
Stud Health Technol Inform. 2020 Jun 16;270:1183-1184. doi: 10.3233/SHTI200353.
Aiming to better understand the genetic and environmental associations of Alzheimer's disease, many clinical trials and scientific studies have been conducted. However, these studies are often based on a small number of participants. To address this limitation, there is an increasing demand of multi-cohorts studies, which can provide higher statistical power and clinical evidence. However, this data integration implies dealing with the diversity of cohorts structures and the wide variability of concepts. Moreover, discovering similar cohorts to extend a running study is typically a demanding task. In this paper, we present a recommendation system to allow finding similar cohorts based on profile interests. The method uses collaborative filtering mixed with context-based retrieval techniques to find relevant cohorts on scientific literature about Alzheimer's diseases. The method was validated in a set of 62 cohorts.
为了更好地理解阿尔茨海默病的遗传和环境关联，已经开展了许多临床试验和科学研究。然而，这些研究通常基于少数参与者。为了解决这一局限性，对多队列研究的需求日益增加，多队列研究可以提供更高的统计效力和临床证据。然而，这种数据整合意味着要处理队列结构的多样性和概念的广泛可变性。此外，发现相似队列以扩展正在进行的研究通常是一项艰巨的任务。在本文中，我们提出了一种推荐系统，以允许基于概况兴趣找到相似队列。该方法将协同过滤与基于上下文的检索技术相结合，以在关于阿尔茨海默病的科学文献中找到相关队列。该方法在一组62个队列中得到了验证。
