Desarrollo de un sistema inteligente para predecir los consumos de medicamentos genéricos de mayor demanda en el distrito de salud 06d05 guano-penipe, aplicando técnicas de regresión de machine learning.

Abstract

In the current curriculum integration work, an intelligent system was developed to predict consumption of generic medicine of greater demand in the Health District 06D05 Guano - Penipe using machine learning regression techniques. In the projection process, 5 sub-processes were identified, which are: request for projections, compilation of information by the cellar, comparison of information, preparation of the projection of generic medicine with the highest demand and preparation of the report; the projection process lasts 25 days. For the development of the product, 3 phases were defined: analysis of the solution, design of the solution and development of the solution. In the analysis of the solution, the data were analyzed in order to obtain the model using the CRISP-ML methodology, where the multiple linear regression technique was chosen with a coefficient of determination of 0.78 that exceeded the established success criteria. Then, the solution was designed using a diagram that indicates the flow of information until the prediction. In the development of the solution, the project was managed with the SCRUM methodology where 8 sprints were developed with a duration of 1120 hours using the Python programming language with the Django framework. To evaluate the time, the ISO / IEC 25010 standard was applied, the time was measured in the current process and the automated process. The analysis determined that in terms of performance efficiency, the system qualifies as very good, obtaining a weighted score of 87.5% that was obtained by evaluating the time behavior and use of system resources. The intelligent system improves the projection process time by 35.48% of the current time by saving 433.92 seconds.