Research On Public Mental Health Early Warning System Based On Ensemble Learning

The accelerated pace of life has led to an increasing pressure on the Chinese people,which has greatly increased the incidence of psychological problems.Psychological problems are one of the main causes of self harm and suicide worldwide,and it is difficult to cure them in clinical practice.Therefore,timely prediction of the early psychological state of the public is particularly important.Based on the above situation,this article mainly conducts the following research:1.Adopting ensemble learning technology and using various classic machine learning algorithms as base classifiers,a mental health warning framework based on multi machine learning competition is proposed.The framework is mainly divided into two stages: multi machine learning competition and ensemble prediction.After preprocessing the original psychological data,an improved global chaotic bat algorithm is used to select the feature combination with the highest evaluation value and enter the mental health warning framework.In the competition stage of multi machine learning,multiple individual learners are trained on the training set,and multiple optimal individual learners are selected based on their prediction accuracy on the test set.In the ensemble prediction stage,an ensemble prediction algorithm is proposed,which uses the accuracy of each optimal learner as a weight for ensemble prediction.2.On the basis of the ensemble prediction algorithm proposed above,requirements analysis and overall design work were carried out,and a mental health warning system was implemented by combining relevant development technologies.Relevant workers can identify risk groups prone to psychological problems through functional analysis of visual display of risk groups,so as to intervene in advance.Currently,most mental health warning technologies use a single machine learning algorithm,which has weak predictive generalization performance.This dissertation proposes a mental health warning framework based on multi machine learning competition and corresponding ensemble prediction algorithms for prediction,and designs and implements a corresponding mental health warning system to alert high-risk populations.In addition,optimized evaluation indicators(psychological normal detection rate and abnormal detection rate)are also proposed to evaluate each algorithm.The experimental results demonstrate that the performance of the mental health warning framework proposed in this article is more powerful than a single prediction algorithm,and will play an important role in public mental health warning.