Complex networks to model and mine patient pathways

• Ano de defesa: 2024
• Autor(a) principal: Rosa, Caroline de Oliveira Costa Souza
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Laboratório Nacional de Computação Científica; Coordenação de Pós-Graduação e Aperfeiçoamento (COPGA); Brasil; LNCC; Programa de Pós-Graduação em Modelagem Computacional
• Palavras-chave em Português: Redes complexas; Exploração de dados (Computação); Mineração de processos; Registros médicos; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: https://tede.lncc.br/handle/tede/392

The use of healthcare data to rebuild the steps patients followed during their treatment— the pathway of patients—is a helpful tool in a variety of scenarios. Examples include inspecting whether clinical guidelines are working as expected, identifying whether there are groups of patients with similar disease patterns, and assessing whether health resources distribution is appropriate. The automatic discovery of patient pathways is a growing field of research that records approaches based on different techniques, such as sequence mining, stochastic modelling, and process mining. Despite the advancements in the area, there are still challenges, especially when modelling and mining specific types of pathways, such as those associated with chronic conditions and health maintenance. These types of pathways demand methods to deal with encounters that repeat themselves, to support multiple perspectives (interventions, diagnoses, speciality of the healthcare professionals, among others) influencing the results, and to keep time information between the encounters. This thesis proposes a framework to deal with such pathways. It comprises (i) a pathway model based on a multiaspect graph, (ii) a dissimilarity function to compare pathways while taking the elapsed time into account, and (iii) a mining method based on traditional centrality measures to discover the most relevant steps of the pathways. We evaluated the framework using the case studies of pregnancy and diabetes and comparing the framework’s results with those from two well-established and popular process mining tools. These results revealed the usefulness of the framework in finding clusters of similar pathways, representing them in an easy-to-interpret way and highlighting the most significant patterns according to multiple perspectives.