Research On The Influencing Factors And Prediction Of The Fundraising Results Of Online Medical Crowdfunding Projects

The improvement of Internet finance has made online fund payment more and more convenient,and the rapid development of network technology has also spawned various public welfare crowdfunding platforms with standardized operations,providing great convenience for the public to participate in public welfare undertakings.On the other hand,China’s social security system is not yet perfect,and families of patients with major diseases still need to bear huge financial pressure.Online medical crowdfunding provides a new way of raising funds.It publishes and raises funds through the Internet to help patients solve the dilemma of medical fund shortage.This financing channel connects the three parties of public offering institutions,donated individuals and the public,so as to achieve two-way communication and realize the transparent mechanism of the whole chain.It has stimulated the public’s enthusiasm for public welfare and promoted the continuous development of domestic philanthropy.Various online donation platforms are becoming more and more mature,and social attention and support are increasing.Online crowdfunding has also developed rapidly around the world.However,among various crowdfunding platforms,the success rate of medical crowdfunding projects is generally low.Therefore,it is meaningful to explore its influencing factors from different perspectives.This research is oriented to the field of online medical crowdfunding,taking real medical crowdfunding projects on Tencent’s public welfare platform as the research object,and using the framework of empirical and prediction to carry out research.The empirical part uses the hierarchical multiple regression method for analysis,and explores the impact of project text characteristics on the fundraising results of medical crowdfunding projects from the perspectives of project title and project details,and provides a series of text narrative strategies for sponsors.The prediction part uses a variety of machine learning methods to perform predictive analysis on the fundraising results of the project,and provide the sponsors with feedback on the predicted fundraising results.The results of this study show that:（1）whether there is a description of the patient’s age,gender,disease,occupation and negative emotions in the item title is a significant factor affecting the success rate of fundraising.Among them,the title contains the patient’s disease and occupation.（2）When there are narratives of patient occupation,monetary evidence and negative emotions in the project details,it has a positive impact on the success rate of fundraising,while the existence of age and positive emotions show a negative impact.Further research found that,（3）in the project texts,high frequency of monetary evidence and negative words and short detailed narrative length can promote the success of medical crowdfunding.Compared with older patients,younger patients have a higher success rate of fundraising.In addition,medical projects that raise funds for patient groups of young children,students or public-spirited person have higher fundraising success rate than groups in other occupations.Finally,based on the optimal model obtained from the above empirical research,this study uses four machine learning methods including Logistic Regression,Support Vector Machine,Random Forest and Boosted Tree to predict the fundraising ability of online medical crowdfunding projects.The results show that: The final prediction accuracy of all models reached more than 84.5%,indicating that the models obtained in this study have high accuracy in predicting the results of medical crowdfunding.Among them,the lifting tree algorithm based on Light GBM framework has the highest prediction accuracy and the best F1 value,indicating that compared with the other three models,the boosted tree algorithm has the best prediction effect.