Predicting Alzheimer's disease progression in healthy and MCI subjects using multi‐modal deep learning approach. (20th December 2022)

• Record Type: Journal Article
• Authors: Mirabnahrazam, Ghazal; Ma, Da; Beaulac, Cédric; Lee, Sieun; Popuri, Karteek; Lee, Hyunwoo; Cao, Jiguo; Wang, Lei; Galvin, James E; Beg, Mirza Faisal
• Is Part Of: Alzheimer's & dementia. Volume 18(2022)Supplement 2
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-12-20
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.060949
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: Alzheimer's disease (AD) is a complex disorder influenced by many factors, but it is unclear how each factor contributes to disease progression. An in‐depth examination of these factors may yield an accurate estimate of time‐to‐conversion to AD for patients at various disease stages. Recent advances in deep learning have enabled researchers to predict patient's disease onset time by exploring the influencing factors in AD progression. Method: We used 543 subjects with 63 features from 3 data modalities from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. The following modalities were used: 1) MRI, 2) genetic and 3) DTC (D emographic, cognitive T ests and C erebrospinal fluid). The 21 most important features were automatically selected for the three modalities. We used a Deep Learning‐based survival analysis model that extends the classic Cox regression model to predict the subjects' disease onset time. Here we re‐define the non‐AD‐progression as "survivor", and AD‐progression as "non‐survivor". The subjects were divided into two groups: progressive subjects (non‐survivor), who were either healthy or diagnosed with Mild Cognitive Impairment (MCI) at initial clinical visit and later developed AD, and non‐progressive subjects ("survivor"), who were either healthy or MCI at initial visit but did not develop AD later. We used 10 random sub‐samples, selecting 80% of the subjects for training and 20% for testing each time; 20% of training data was used for internal validation. Result: Figure 1 shows the estimated survival rates over 10 years. Both groups had a high survival chance at the start. The progressive group's survival chance dropped much faster and fell below 20% by the end of the period. The non‐progressive group's survival chance remained around 50%. Feature importance analysis is displayed in Figure 2. Eight of the top ten most important features are from the cognitive tests, demonstrating their importance in survival analysis. Amygdala and Hippocampus regions, as well as age, are also notable features. Conclusion: Our study demonstrated that using powerful predictive models on multi‐modal data can improve prediction of time‐to‐conversion. This not only leads to a better understanding of AD, but also provides essential tools for practitioners who wish to follow their patients' disease progression.