Application of Concordance Probability Estimate to Predict Conversion from Mild Cognitive Impairment to Alzheimer's...

Area of Science:

• Neuroscience
• Gerontology
• Biostatistics

Background:

• Mild cognitive impairment (MCI) significantly increases the risk of developing Alzheimer's disease (AD) dementia.
• Accurate prediction of AD progression in MCI patients is vital for effective clinical management.
• Existing risk prediction models often rely on accuracy metrics like AUC or Harrell's C-statistic, which have limitations.

Purpose of the Study:

• To evaluate the prognostic accuracy of a Cox proportional hazards (PH) model for predicting MCI to AD conversion.
• To introduce and utilize the Gönen & Heller's K-index (concordance probability estimate - CPE) as a robust measure of predictive accuracy.
• To address limitations of existing metrics by employing a method independent of censoring distribution.

Main Methods:

• Development of a Cox proportional hazards (PH) model to predict the conversion from MCI to AD.
• Utilized the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset for model building and validation.
• Evaluated the prognostic accuracy of the model using the K-index (CPE).

Main Results:

• A Cox PH model was successfully constructed to predict Alzheimer's disease progression in subjects with mild cognitive impairment.
• The K-index was employed to assess the overall predictive accuracy of the developed model.
• The study demonstrated the utility of the K-index in evaluating prognostic models for neurodegenerative disease progression.

Conclusions:

• The K-index provides a reliable measure for assessing the predictive accuracy of Cox PH models in the context of Alzheimer's disease progression.
• This approach offers an alternative to AUC and Harrell's C-statistic, particularly in the presence of censoring.
• Accurate prognostic models are essential for identifying high-risk MCI patients, facilitating timely interventions and clinical trial enrollment.