Respiratory system disease outpatient quantity prediction method

Abstract

The invention provides a respiratory system disease outpatient quantity prediction method, which is based on outpatient quantity data, air quality information and weather information, constructs a time series hybrid prediction model, and realizes automatic and accurate prediction of respiratory system disease outpatient quantity. The method not only can describe the trend rule of the outpatient quantity index, but also can incorporate a plurality of related factors including air quality and weather information to describe the details of the fluctuation rule. Therefore, based on the method, important reference can be provided for hospital managers, managers are helped to judge the attack law and trend of respiratory system diseases, the managers are assisted to make decisions such as resource allocation and task planning, and a quantitative basis is provided for reasonable allocation of medical resources.

Description

technical field [0001] The invention relates to a HoltWinters-XGBoost-based method for predicting the number of outpatients of respiratory diseases, belonging to the field of time series prediction. Background technique [0002] Respiratory diseases are the most common frequently-occurring diseases in my country, ranking first in the cause of death of the total population in my country all the year round. There are many types of respiratory diseases with complex etiology, which are usually affected by factors such as air pollution, smoking, industrial granulation factors, inhalation and infection of biological factors, age, and personal constitution. Common respiratory diseases, such as acute upper respiratory infection, influenza and pneumonia, and chronic lower respiratory disease, have a large incidence in the population, and show obvious seasonal periodicity and time trend. Since the 1990s, time series analysis has been widely used in the study of the short-term inciden...

AI Technical Summary

Problems solved by technology

However, both have their own limitations. The traditional time series model is usually modeled in the form of a single indicator, which contains limited information and cannot describe holiday factors with drift (such as the Spring Festival, etc.), and cannot add weather and air pollution. The impact on the number of outpatients; the machine learning model can carry out multi-index modeling and fully incorporate all the factors that can be collected for modeling. However, when the time series has a monotonic trend (unstationary), the range of model prediction values ​​can only be limited to historical Therefore, the predicted value is always too small or too large; the deep learning model not only requires a large number of parameter tuning experiments, but also is easy to overfit when the sample size is small, so it is difficult Realize automated modeling predictions

Images