A Novel Diagnostic Model for Early Detection of Alzheimer’s Disease Based on Clinical and Neuroimaging Features

Alzheimer’s Disease (AD) is a dangerous disease that is known for its characteristics of eroding memory and destroying the brain. The classification of Alzheimer's disease is an important topic that has recently been addressed by many studies using Machine Learning (ML) and Deep Learning (DL) methods. Most research papers tackling early diagnosis of AD use these methods as a feature extractor for neuroimaging data. In our research paper, the proposed algorithm is to optimize the performance of the prediction of early diagnosis from the multimodal dataset by a multi-step framework that uses a Deep Neural Network (DNN) as an optimization technique to extract features and train these features by Random Forest (RF) classifier. The results of the proposed algorithm showed that using only demographic and clinical data results in a balanced accuracy of 88% and an area under the curve (AUC) of 94.6. Ultimately, combining clinical and neuroimaging features, prediction results improved further to a balanced accuracy of 92% and an AUC of 97%. This study successfully outperformed other studies for both clinical and the combination of clinical and neuroimaging data, proving that multimodal data is efficient in the early diagnosis of AD. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.