Predicting Alzheimer's Disease from Mild Cognitive Impairment
Author Information
Author(s): Cui Yue, Liu Bing, Luo Suhuai, Zhen Xiantong, Fan Ming, Liu Tao, Zhu Wanlin, Park Mira, Jiang Tianzi, Jin Jesse S.
Primary Institution: University of Newcastle
Hypothesis
The combination of different modes of data would achieve better results in predicting conversion from MCI to AD.
Conclusion
The study established meaningful multivariate predictors that may be useful for clinical diagnosis of Alzheimer's disease.
Supporting Evidence
- The combination of selected neuropsychological, MRI, and CSF features achieved an accuracy of 67.13%.
- Neuropsychological measures outperformed individual MRI and CSF features in predicting conversion.
- Significant differences were found between MCI converters and non-converters in terms of selected features.
- Predictive values of MCI converters varied significantly based on conversion time.
Takeaway
Researchers looked at different types of brain data to figure out who with mild memory problems might get Alzheimer's disease. They found that using a mix of data types helps make better predictions.
Methodology
The study used structural MRI, CSF biomarkers, and neuropsychological assessments to predict conversion from MCI to AD, employing support vector machine classifiers for analysis.
Potential Biases
Potential bias in neuropsychological measures compared to MRI and CSF measures due to reliance on clinical evaluations.
Limitations
The follow-up period of 24 months may not be sufficient to obtain reliable ground truth labels for MCI converters.
Participant Demographics
Participants included 87 MCI non-converters, 56 MCI converters, 111 normal controls, and 96 Alzheimer's disease patients, with no significant differences in age and sex between groups.
Statistical Information
P-Value
p<0.01
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website