| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 9.72 MB | Adobe PDF |
Authors
Abstract(s)
Em diversos setores, há uma grande quantidade de dados recolhidos e armazenados, mas que
não são analisados. A área da saúde não é distinta e, pela importância que tem junto da
população, esta situação não é desejável. Os dados podem fornecer informações históricas ou
tendências que poderão ajudar a melhorar o desempenho das organizações no futuro.
Para contornar as dificuldades referidas, a mineração de processos (Process Mining) possibilita
a extração de conhecimento a partir de dados gerados e armazenados nos sistemas de
informação. Assim, este trabalho pretende mostrar os benefícios da utilização do Process
Mining na melhoria de processos de saúde, neste caso aplicada a dados de um serviço de
urgências.
O trabalho iniciou-se por uma fase de pesquisa e exploração, na qual os algoritmos e
ferramentas de Process Mining foram investigados, analisados e comparados. Nesta fase, as
ferramentas de Process Mining, PM4Py, ProM e Disco, foram testadas, bem como os
algoritmos disponíveis Alpha Miner, Directly-Follows Graphs, Fuzzy Miner, Heuristic Miner
e Inductive Miner, em cenários com dados que apresentavam desafios reais. Destes testes
resultaram comparações e conclusões que foram apresentadas de forma compreensível e
intuitiva. Esta fase foi de extrema importância, pois permitiu obter um conhecimento prático
necessário para que a posterior aplicação das técnicas fosse o mais eficiente possível.
O caso de estudo final permitiu perceber o real potencial das contribuições que o Process
Mining pode dar a uma área de tão relevante interesse geral. Com resultados muito positivos,
chegou-se a um modelo que permitiu a análise do real funcionamento de um serviço de
urgências. Este modelo foi examinado ao pormenor e, posteriormente, foi verificada a
conformidade de acordo com o modelo previsto. Esta comparação possibilitou retirar
conclusões em relação à gestão do processo, assim como dos seus recursos. Estas informações
são extremamente úteis para que os profissionais de cada unidade possam agir sobre as
ineficiências existentes e melhorar um serviço com tanta importância para a população.
ABSTRACT: In many industries, there is a large amount of data collected and stored but not analyzed. The health area is not distinct and, given its importance to the population, this situation is not desirable. Data can provide historical information or trends that can help improve organizations' performance in the future. To overcome the aforementioned difficulties, Process Mining allows extracting knowledge from data generated and stored in information systems. Thus, this work aims to show the benefits of using Process Mining to improve health processes, in this case applied to data of an emergency service. The work began with a research and exploration phase, where the Process Mining algorithms and tools were investigated, analyzed and compared. In this phase, the Process Mining tools, PM4Py, ProM and Disk, were tested, as well as the available algorithms, Alpha Miner, DirectlyFollows Graphs, Fuzzy Miner, Heuristic Miner and Inductive Miner, in scenarios with data that represented real challenges. These tests resulted in comparisons and conclusions that were presented in an understandable and intuitive way. This phase was extremely important because it allowed to obtain the practical and necessary knowledge for the subsequent application. The final case study allowed to realize the real potential of the contributions that Process Mining can make in an area of such relevant general interest. With very positive results, a model was generated in order to analyze the real functioning of an emergency service. This model was examined in detail and, subsequently, compliance was verified in accordance with the expected model. This comparison led to draw conclusions in relation to the management of the process, as well as its resources. This information is extremely useful for professionals in order to act on existing inefficiencies and improve a service that is so important to the population.
ABSTRACT: In many industries, there is a large amount of data collected and stored but not analyzed. The health area is not distinct and, given its importance to the population, this situation is not desirable. Data can provide historical information or trends that can help improve organizations' performance in the future. To overcome the aforementioned difficulties, Process Mining allows extracting knowledge from data generated and stored in information systems. Thus, this work aims to show the benefits of using Process Mining to improve health processes, in this case applied to data of an emergency service. The work began with a research and exploration phase, where the Process Mining algorithms and tools were investigated, analyzed and compared. In this phase, the Process Mining tools, PM4Py, ProM and Disk, were tested, as well as the available algorithms, Alpha Miner, DirectlyFollows Graphs, Fuzzy Miner, Heuristic Miner and Inductive Miner, in scenarios with data that represented real challenges. These tests resulted in comparisons and conclusions that were presented in an understandable and intuitive way. This phase was extremely important because it allowed to obtain the practical and necessary knowledge for the subsequent application. The final case study allowed to realize the real potential of the contributions that Process Mining can make in an area of such relevant general interest. With very positive results, a model was generated in order to analyze the real functioning of an emergency service. This model was examined in detail and, subsequently, compliance was verified in accordance with the expected model. This comparison led to draw conclusions in relation to the management of the process, as well as its resources. This information is extremely useful for professionals in order to act on existing inefficiencies and improve a service that is so important to the population.
Description
Keywords
Big Data de saúde Process Mining PM4Py ProM Disco Alpha Miner Directly-Follows Graph Fuzzy Miner Heuristic Miner Inductive Miner