CO poisoning prediction method based on meteorological data

Abstract

The invention provides a CO poisoning prediction method based on meteorological data. The CO poisoning prediction method comprises the following steps: acquiring historical meteorological data and COpoisoning case number of a to-be-predicted region; dividing the CO poisoning case number into k levels; selecting significant correlation effective features by using parson correlation analysis and partial correlation analysis; constructing an ordered multi-classification Logistic regression model of the CO poisoning risk level y according to the selected effective feature set X; and obtaining andsubstituting meteorological data of a to-be-predicted region on the same day needing to be predicted into the regression model to calculate the probability Pj of each CO poisoning risk level, and thej value corresponding to the maximum value max (p) of Pj being the predicted CO poisoning risk level. The CO poisoning prediction method based on meteorological data can predict the CO poisoning risklevel, and the CO poisoning prediction is simple and convenient.

Description

Technical field [0001] The invention relates to a CO poisoning prediction method, in particular to a CO poisoning prediction method based on meteorological data. Background technique [0002] The Beijing Specialized Meteorological Observatory used weather forecast factors and derived factors to construct a prediction model for non-occupational CO (carbon monoxide) poisoning in Beijing. It collected the Beijing Emergency Center from February 13, 2002 to December 31, 2005. With 1,418 samples of CO poisoning cases, corresponding to local meteorological conditions, using quasi-multiple regression index probability classification technology, a CO poisoning index (4 classification) forecast and corresponding risk level assessment model were established. The model is designed as the following formula: Y i =Y Wi (B 0 +∑B j X j ), where Y i Is the regression fitting value of the meteorological index of CO poisoning, which is defined as the daily average number of CO poisonings per 10 mill...

AI Technical Summary

Problems solved by technology

The pattern is designed as follows: Y i =Y Wi (B 0 +∑B j x j ), where Y i is the regression fitting value of the meteorological index of CO poisoning, defined as the daily average CO poisoning person-times per 10 million population; Y Wi (i=1, 2,..., n) is the background harmonic value of CO poisoning people, X j is the predictor factor entering the equation, B j (i=0, 1, 2, ..., m) is the corresponding regression coefficient, m and n respectively identify the number of factors entering the equation and the maximum number of samples in the forecast year, but this forecasting model has its limitations: 1. The poisoning data only comes from 16 emergency centers in the urban area of ​​Beijing. In fact, many poisoned patients did not go through the emergency station, but went directly to the hospital. The research data may be significantly less than the actual data; 2. The weather forecast factors in this area are used And the prediction of derived factors, there are many derived factors, it is difficult to obtain them at the same time in actual forecasting, which increases the difficulty of forecasting; 3. The forecasting model is profound, not intuitive enough, and difficult to understand

Images