Establishing a machine learning dementia progression prediction model with multiple integrated data

Area of Science:

• Neurology
• Artificial Intelligence
• Data Science

Background:

• Dementia presents a significant global health challenge, necessitating effective tools for predicting disease progression.
• Machine learning (ML) offers a powerful approach to developing predictive models from complex, real-world clinical data.

Purpose of the Study:

• To develop and validate a machine learning model for predicting the progression of degenerative dementia.
• To identify key clinical and demographic variables that are most predictive of dementia progression.

Main Methods:

• Retrospective analysis of 679 patients with degenerative dementia, followed for over two years.
• Utilized the extreme gradient boosting (XGB) model to analyze demographic, clinical dementia rating (CDR), mini-mental state examination (MMSE), and laboratory data (LV) variables.
• Employed a step-wise approach to identify optimal feature combinations and variable importance.

Main Results:

• The integrated D-CDR-MMSE-LV model achieved a high area under the receiver operating characteristic curve (AUC) of 85.12.
• The XGBoost model identified eight critical variables from the integrated datasets.
• The model demonstrated robust performance with high sensitivity (84.66).

Conclusions:

• A machine learning model was successfully developed to monitor dementia progression using real-world clinical data.
• The identified eight critical variables provide valuable insights for clinicians in guiding dementia patient management.