Font Size: a A A

Research About The Modeling And Treatment Strategies Analysis Based On Menopause

Posted on:2017-06-18Degree:MasterType:Thesis
Country:ChinaCandidate:X LiFull Text:PDF
GTID:2370330590491387Subject:Mechanical engineering
Abstract/Summary:PDF Full Text Request
Menopausal problem has already become a serious threat to the quality of Middle-aged woman's life,and also caused heavy economic burden to the society.During the menopause period,women will suffer from some physical and psychological symptoms.Depression is the most harmful symptom in those symptoms and it will cost a lot.Based on menopausal problems,we carried out three studies,including the relationship between the severity of menopausal symptoms influencing factors,inverse optimization and the treatment strategies analysis.First of all,we established an artificial neural network model between the severity of menopausal symptoms(measured by KMI score)and 9 influencing factors.Through iteration,the fitting precision reached 0.95.The results were satisfying and effective compared with that based on statistical analysis.Furthermore,the network was optimized by changing the parameters.Based on the optimal network,we carried out the sensitivity analysis and then significant factors and how they affected the KMI score were identified.And then the problem of the medical data missing was discussed in this paper.Specifically,we combined the idea of regression analysis in imputation and artificial neural network method.Firstly,the reverse network was established,where the variable parity(with missing data)was considered as the output and all other variables were the inputs of the network.The missing data were completed from the outputs.Based on the new completed data,a forward neural network was established and it was compared with the one with original data in fitting accuracy.As a result,after the inverse optimization method was used,the prediction accuracy was improved by 4%.Finally,we developed a finite-horizon Markov decision process model for treatment strategies of menopausal syndrome.State transition probabilities are calculated with the quadratic programming method under treatment and no-treatment behaviors.The losses caused by depression are considered as the major economic losses of menopause.Finally,the method of value iteration is used to obtain the menopause treatment strategy under different periods and states,followed by annual cost sensitivity analysis to discuss the threshold price of the drug.
Keywords/Search Tags:Menopause, Artificial neural network, Sensitivity analysis, Markov decision process
PDF Full Text Request
Related items