TY - JOUR
T1 - A novel approach to dementia prediction of DTI markers using BALI, LIBRA, and machine learning techniques
AU - Akbarifar, Ahmad
AU - Maghsoudpour, Adel
AU - Mohammadian, Fatemeh
AU - Mohammadzaheri, Morteza
AU - Ghaemi, Omid
PY - 2024/6/27
Y1 - 2024/6/27
N2 - Early prediction of dementia and disease progression remains challenging. This study presents a novel machine learning framework for dementia diagnosis by integrating multimodal neuroimaging biomarkers and inexpensive and readily available clinical factors. Fractional anisotropy (FA) measurements in diffusion tensor imaging (DTI) provide microstructural insight into white matter integrity disturbances in dementia. However, the acquisition of DTI is costly and time-consuming. We applied Recursive Feature Elimination (RFE) to identify predictors from structural measures of the 9 factors of Brain Atrophy and Lesion Index (BALI) factors and 42 factors of Clinical Lifestyle for Brain Health (LIBRA) factors to estimate FA in DTI. The 10 most effective features of BALI/LIBRA selected by RFE were used to train an interpretable decision tree model to predict the severity of dementia from DTI. A decision tree model based on biomarkers selected by RFE achieved an accuracy of 96.25% in predicting dementia in an independent test set. This integrated framework pioneers the prediction of white matter microstructural changes from available structural/clinical factors using machine learning. By avoiding DTI acquisition, our approach provides a practical and objective tool to improve dementia screening and progress monitoring. The Identification of key predictive markers of BALI/LIBRA will also provide information on the mechanisms of lifestyle-related disease mechanisms, neurodegeneration, and white matter dysfunction. This study aims to predict FA measures from DTI, which indicate white matter integrity and dementia severity, using inexpensive and readily available BALI and LIBRA factors through machine learning.
AB - Early prediction of dementia and disease progression remains challenging. This study presents a novel machine learning framework for dementia diagnosis by integrating multimodal neuroimaging biomarkers and inexpensive and readily available clinical factors. Fractional anisotropy (FA) measurements in diffusion tensor imaging (DTI) provide microstructural insight into white matter integrity disturbances in dementia. However, the acquisition of DTI is costly and time-consuming. We applied Recursive Feature Elimination (RFE) to identify predictors from structural measures of the 9 factors of Brain Atrophy and Lesion Index (BALI) factors and 42 factors of Clinical Lifestyle for Brain Health (LIBRA) factors to estimate FA in DTI. The 10 most effective features of BALI/LIBRA selected by RFE were used to train an interpretable decision tree model to predict the severity of dementia from DTI. A decision tree model based on biomarkers selected by RFE achieved an accuracy of 96.25% in predicting dementia in an independent test set. This integrated framework pioneers the prediction of white matter microstructural changes from available structural/clinical factors using machine learning. By avoiding DTI acquisition, our approach provides a practical and objective tool to improve dementia screening and progress monitoring. The Identification of key predictive markers of BALI/LIBRA will also provide information on the mechanisms of lifestyle-related disease mechanisms, neurodegeneration, and white matter dysfunction. This study aims to predict FA measures from DTI, which indicate white matter integrity and dementia severity, using inexpensive and readily available BALI and LIBRA factors through machine learning.
UR - https://www.open-access.bcu.ac.uk/15649/
U2 - 10.1140/epjp/s13360-024-05367
DO - 10.1140/epjp/s13360-024-05367
M3 - Article
SN - 2190-5444
VL - 139
JO - The European Physical Journal Plus
JF - The European Physical Journal Plus
ER -