Classification accuracy of TMS for the diagnosis of mild cognitive impairment

ObjectiveTo evaluate the performance of a Random Forest (RF) classifier on Transcranial Magnetic Stimulation (TMS) measures in patients with Mild Cognitive Impairment (MCI).MethodsWe applied RF TMS obtained from multicenter cohort MCI, including MCI-Alzheimer’s Disease (MCI-AD), MCI-frontotemporal dementia (MCI-FTD), MCI-dementia Lewy bodies (MCI-DLB), and healthy controls (HC). All underwent assessment at recruitment (index test), application reference clinical criteria, to predict different neurodegenerative disorders. The primary outcome were classification accuracy, precision, recall F1-score differentiating each disorder.Results160 participants included, namely 64 diagnosed as MCI-AD, 28 MCI-FTD, 14 MCI-DLB, 47 A series 3 binary classifiers was employed, prediction model exhibited high accuracy (ranging 0.72 0.86), precision (0.72–0.90), (0.75–0.98), F1-scores (0.78–0.92), disorder. By computing new classifier, trained validated current MCI patients, indices showed even higher 0.83 0.93), (0.87–0.89), (0.83–1.00), (0.85–0.94).ConclusionsTMS may be considered useful additional screening tool used practice prodromal stages dementias.