Table 5 Features selected from the ADNI dataset by the best models using data-driven thresholding
From: Ensemble feature selection with data-driven thresholding for Alzheimer's disease biomarker discovery
ADNI feature descriptions | AD biomarker evidence |
|---|---|
Mini Mental State Exam score (MMSE) | Designed to identify dementia [51] |
Instrumental activities of daily living (IADL) score | Often used to determine the cognitive function of an individual [60] |
Status of the epsilon 4 allele of the APOE gene | Increases the risk of late-onset AD [11] |
42-amino-acid-long beta amyloid peptide (abeta42) level in CSF | Thought to be a biomarker for AD [61] |
Plasma neurofilament light level | Increased concentrations of CSF and plasma levels of the neurofilament light chain have been identified as potential biomarkers of AD [62] [63] |
CSF neurofilament light level | |
CSF tau level | The presence of tau and phosphorylated tau in CSF accurately detects Alzheimer's disease pathology [64] |
CSF phosphorylated tau level | |
Neuropsychiatric Inventory – total score | Can be used to detect behavioural changes in AD [65] |
CSF neurogranin concentration | Patients with AD show an increase in CSF neurogranin concentration and this is not seen in other neuro- degenerative diseases [66] |
Geriatric depression scale – total score | Depression has been linked to AD and cognitive decline [67] |
Percentage of neutrophils | People with AD have a higher blood neutrophil–lymphocyte ratio (NLR), a marker of inflammation, than healthy controls [68] |