Markov transitional model
Posted: Fri Jun 22, 2007 6:37 pm
Never heard of this before. Anyone?
Markov model predicts short-term disability in multiple sclerosis
Source: Reuters
Author:
Date: Fri, 22 June 2007
The findings published in the June 12th issue of Neurology support the novel use of a Markov transitional model in predicting the short-term disability in patients with multiple sclerosis (MS).
The standard statistical methods of survival analysis used to create clinical predictive models do not include the ongoing changing nature of MS, Dr. H. L. Weiner, of Harvard Medical School, Boston, and colleagues point out.
However, "Markov transitional models incorporate the fluctuating nature of chronic diseases through the analysis of discrete states of progression and make use of all clinical information."
In the current study, the authors developed covariate-specific short-term disability curves to predict the likelihood of disease progression using the Expanded Disability Status Scale (EDSS) at semiannual visits. They prospectively collected EDSS scores for 218 MS patients with relapsing-remitting disease or clinically isolated syndrome.
The investigators applied their longitudinal data to a Markov transitional model and the patients' previous history of disability was used to predict subsequent short-term disability measured by EDSS. Once the model was fit, a probability matrix was generated for each subject based on specific clinical and MRI covariates.
The researchers found that patients in the lowest baseline brain parenchymal fraction quartile and those in highest T2 lesion volume quartile experienced progression according to EDSS.
In patients with a six-month EDSS of two, the probability of progressing to a sustained EDSS of three within three years was 0.277 for those with a low brain parenchymal fraction and a high T2 lesion volume.
Whereas, the corresponding probability was 0.055 for patients with a high brain parenchymal fraction and a low T2 lesion volume.
"Now that we have established this new method, we can further assess its value in the progressive stage of MS and its ability to evaluate the influence of additional covariates such as specific treatments, relapses, new MRI metrics, as well as immunologic and genetic markers on subsequent disability," Dr. Weiner's team concludes.
Markov model predicts short-term disability in multiple sclerosis
Source: Reuters
Author:
Date: Fri, 22 June 2007
The findings published in the June 12th issue of Neurology support the novel use of a Markov transitional model in predicting the short-term disability in patients with multiple sclerosis (MS).
The standard statistical methods of survival analysis used to create clinical predictive models do not include the ongoing changing nature of MS, Dr. H. L. Weiner, of Harvard Medical School, Boston, and colleagues point out.
However, "Markov transitional models incorporate the fluctuating nature of chronic diseases through the analysis of discrete states of progression and make use of all clinical information."
In the current study, the authors developed covariate-specific short-term disability curves to predict the likelihood of disease progression using the Expanded Disability Status Scale (EDSS) at semiannual visits. They prospectively collected EDSS scores for 218 MS patients with relapsing-remitting disease or clinically isolated syndrome.
The investigators applied their longitudinal data to a Markov transitional model and the patients' previous history of disability was used to predict subsequent short-term disability measured by EDSS. Once the model was fit, a probability matrix was generated for each subject based on specific clinical and MRI covariates.
The researchers found that patients in the lowest baseline brain parenchymal fraction quartile and those in highest T2 lesion volume quartile experienced progression according to EDSS.
In patients with a six-month EDSS of two, the probability of progressing to a sustained EDSS of three within three years was 0.277 for those with a low brain parenchymal fraction and a high T2 lesion volume.
Whereas, the corresponding probability was 0.055 for patients with a high brain parenchymal fraction and a low T2 lesion volume.
"Now that we have established this new method, we can further assess its value in the progressive stage of MS and its ability to evaluate the influence of additional covariates such as specific treatments, relapses, new MRI metrics, as well as immunologic and genetic markers on subsequent disability," Dr. Weiner's team concludes.