The Design and Assessment of Computer-Aided Multi-State Disease Prediction Model

Project: A - Government Institutionb - National Science and Technology Council

Project Details


Although number of methods has been developed for selecting cutoffs, early detection or diagnosis for disease using interval-scale biomarkers rarely assesses the effects of heterogeneity and utility varying with individual covariates on cutoffs related to the two error rates (false positive rate and false negative rate). Fuzzy set approach is proposed to cope with this question. In addition to genome influence, epigenetic mechanism, like DNA methylation, is an alterative genetic mechanism which may lead to heterogeneous genotype expression in the absence of DNA mutation. There are a number of predicted models for estimating the risk of disease. The drawback is that they are all classified as two-state models rather than multi-state models which can delineate multi-state disease progression in related to genetic and environmental factors. In the present study, we would like to construct a multi-state Markov model to depict cervical cancer nature history. Genetic, epigenetic, and personal attributes obtained from literature are incorporated into model to assess their influences on disease progression. Moreover, the application of multistate risk assessment to cervical cancer prevention at individual level is hardly addressed. By simulating empirical data, we design a hypothetical study to show how to achieve this object. Through multistate risk assessment model, we apply the parameters estimated from simulated dataset to build up individualized preventive strategy on which we are based to do economic appraisal between different preventive strategies. Moreover, we need the health informatics which can combine information technology and statistical models to develop more computer-aided tools in terms of disease prediction.
Effective start/end date8/1/117/31/12


  • Fuzzy set
  • Markov Model
  • Predictive Model


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.