Research and prediction of opioid crisis based on BP neural network and Markov chain

Nowadays, in the United States, opioid abuse is so serious that it has become a crisis, causing serious health impacts and huge losses to the US economy. To study the social and economic data and the relationship between opioid drug abuse situation, this paper uses Grey Relation Analysis to analyze the identification and counting of synthetic opioids and heroin provided by NFLIS and the data related to socioeconomic factors provided by the United States Census Bureau. After that, the orderly clustering is used to introduce the corresponding level. Then, BP neural network and Markov model are built to forecast the degree of the flood of opioid. The Proportion of Low Education Level People, Number of New Pregnant Women, The Proportion of The Elderly Living Alone and the other 7 factors were selected as the input nodes of the neural network. Based on the prediction results of the BP neural network, Markov Chain is used to correct the residual sequence. It is found that the deviation of prediction is reduced from[-10.99%, 22.33%] to[-8.29%, 2.81%], making the modified value closer to the measured value. This method combines the advantages of the BP neural network and the Markov network, which improves the accuracy of prediction and provides certain reference values for opioid prediction. Finally, we propose strategies to address the opioid crisis and test their effectiveness.