Loading...
Publication
Estudo estatístico do impacto da pandemia nas manutenções curativas de dispositivos médicos e predição de stocks baseada em aprendizagem automática
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 3.12 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
A interligação entre a engenharia biomédica, desenvolvimento tecnológico e
cuidados de saúde, permite que os dispositivos médicos desempenhem um papel
crucial na área da saúde. Este estudo aborda a gestão da manutenção de dispositivos
médicos e dos materiais utilizados nas intervenções, com especial foco na análise
estatística de dados, referentes aos anos entre 2019 e 2022, pertencentes a dois tipos
de clientes, Cliente A (centros de saúde) e Cliente B (centros hospitalares).
A pesquisa visa compreender o impacto da pandemia por COVID-19 na realização
de manutenções, perceber que materiais são mais utilizados e realizar uma previsão de
stocks para os próximos 48 meses recorrendo a técnicas de Machine Learning
(Aprendizagem Automática). Foi realizado uma análise estatística dos tipos de
equipamentos com mais manutenções nos quatro anos em estudo, e dos tipos de
material mais utilizados. Já para a realização da previsão de stocks foram utilizados os
métodos de previsão Suavização Exponencial Holt-Winters, SARIMA, ARIMA e SVR,
que não demonstraram resultados tão fidedignos com os outros dois modelos.
Os principais resultados revelam informações importantes sobre os dispositivos
médicos com maior número de manutenções, onde foram selecionados os 4
equipamentos e os 4 materiais com maiores percentagens de intervenções/ utilizações.
Através destes resultados percebeu-se que, para o Cliente A o esfigmomanómetro foi
o equipamento com mais intervenções nos quatro anos e para o Cliente B foi o monitor
de sinais vitais. Numa fase posterior, foi selecionado o material mais utilizado no Cliente
A (braçadeiras de adulto) e no Cliente B (Sensor O2) e foram realizadas as previsões
de stocks destes dois materiais para os próximos 48 meses, onde se verificou uma
tendência crescente da quantidade dos dois materiais ao longo dos próximos quatro
anos.
A interligação entre a análise estatística, a seleção criteriosa de dispositivos
médicos e materiais e a aplicação de técnicas preditivas demonstra um caminho
promissor para aprimorar a eficiência do setor dos dispositivos médicos, tornando estes
equipamentos mais eficientes e com um menor tempo de paragem. Esta dissertação
contribui de forma significativa para a compreensão da dinâmica da manutenção de
dispositivos médicos, fornecendo dados para a aplicação de estratégias futuras de
forma informada, como por exemplo a diminuição de stocks, antecipação ou diminuição
de manutenções
The interconnection between biomedical engineering, technological development and healthcare allows medical devices to play a crucial role in healthcare. This study addresses the maintenance management of medical devices and the materials used in interventions, with a special focus on the statistical analysis of data, referring to the years between 2019 and 2022, belonging to two types of clients, Client A (health centres) and Client B (hospital centres). The research aims to understand the impact of the COVID-19 pandemic on maintenance, to understand which materials are most used and to carry out a stock forecast for the next 48 months using Machine Learning techniques. A statistical analysis was made of the types of equipment with the most maintenance in the four years under study, and the types of material most used. For stock forecasting, the Holt-Winters Exponential Smoothing, SARIMA, ARIMA and SVR forecasting methods were used, which did not show such reliable results as the other two models. The main results reveal important information about the medical devices with the most maintenance, where the 4 pieces of equipment and 4 materials with the highest percentages of interventions/uses were selected. These results showed that for Client A the sphygmomanometer was the device with the most interventions over the four years and for Client B it was the vital signs monitor. At a later stage, the material most used at Client A (adult armbands) and Client B (O2 sensor) was selected and stock forecasts were made for these two materials for the next 48 months, which showed an upward trend in the quantity of both materials over the next four years. The interconnection between statistical analysis, the careful selection of medical devices and materials, and the application of predictive techniques shows a promising path to improving the efficiency of the medical device sector, making this equipment more efficient and with less downtime. This dissertation makes a significant contribution to understanding the dynamics of medical device maintenance, providing data for the application of future strategies in an informed manner, such as reducing stocks, anticipating, or reducing maintenance.
The interconnection between biomedical engineering, technological development and healthcare allows medical devices to play a crucial role in healthcare. This study addresses the maintenance management of medical devices and the materials used in interventions, with a special focus on the statistical analysis of data, referring to the years between 2019 and 2022, belonging to two types of clients, Client A (health centres) and Client B (hospital centres). The research aims to understand the impact of the COVID-19 pandemic on maintenance, to understand which materials are most used and to carry out a stock forecast for the next 48 months using Machine Learning techniques. A statistical analysis was made of the types of equipment with the most maintenance in the four years under study, and the types of material most used. For stock forecasting, the Holt-Winters Exponential Smoothing, SARIMA, ARIMA and SVR forecasting methods were used, which did not show such reliable results as the other two models. The main results reveal important information about the medical devices with the most maintenance, where the 4 pieces of equipment and 4 materials with the highest percentages of interventions/uses were selected. These results showed that for Client A the sphygmomanometer was the device with the most interventions over the four years and for Client B it was the vital signs monitor. At a later stage, the material most used at Client A (adult armbands) and Client B (O2 sensor) was selected and stock forecasts were made for these two materials for the next 48 months, which showed an upward trend in the quantity of both materials over the next four years. The interconnection between statistical analysis, the careful selection of medical devices and materials, and the application of predictive techniques shows a promising path to improving the efficiency of the medical device sector, making this equipment more efficient and with less downtime. This dissertation makes a significant contribution to understanding the dynamics of medical device maintenance, providing data for the application of future strategies in an informed manner, such as reducing stocks, anticipating, or reducing maintenance.
Description
Keywords
Manutenção Hospitalar Dispositivos Médicos Análise Estatística Machine Learning (Aprendizagem Automática) Biomédica Hospital Maintenance Medical Devices Statistical Analysis Machine Learning Biomedical