Abstract
DL-Based Prediction of MCI-to-AD Conversion: Systematic review and meta-analysis, Model Applications, and Future Directions
Alzheimer’s disease (AD) is an irreversible neurodegenerative disease characterized by progressive cognitive decline leading to dementia.(1) For effective intervention, it is essential to detect patients with mild cognitive impairment (MCI) at high risk of progression to AD.(2) Physicians first assess cognitive status and diagnose MCI or AD, then evaluate biomarkers to estimate risk of progression.(2)
Artificial intelligence (AI) has emerged as an important tool for modeling disease progression in AD, including predicting conversion from MCI to AD.(3) However, although AI has been increasingly used in Alzheimer’s research, applications specifically focused on predicting MCI-to-AD conversion remain limited. Only about 25% of 97 studies addressed this task, and most emphasized classification rather than progression, indicating a notable research gap.(4)
Deep learning (DL), an AI method and a subtype of machine learning (ML) that uses artificial neural networks to learn from data, has demonstrated superior predictive performance in MCI-to-AD prediction compared with traditional ML.(3,5) DL is particularly effective at capturing complex patterns in high-dimensional biomedical data.(5)
My dissertation will offer a comprehensive assessment of DL-based prediction of MCI-to-AD conversion. In Aim 1, I’ll conduct a systematic review and meta-analysis of existing literature to compare performance of existing AI models used in MCI-to-AD conversion in terms of model type, input features (e.g., imaging, biomarkers, clinical factors), and interpretability. In Aim 2, I will use the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset to apply the best-performing DL model identified in Aim 1 for MCI-to-AD prediction and evaluate its performance in the ADNI cohort. If necessary, I will extend the model by incorporating additional input features or developing a new predictive model to further improve accuracy and generalizability. Aim 3 will conduct an economic evaluation of the best-performing or refined DL model identified in Aim 2, quantifying the economic value of improved MCI-to-AD conversion prediction.
Current stage: database searches completed (Scopus, Web of Science, PubMed, Embase, CINAHL), with 3,790 records imported and 1,750 duplicates removed via Covidence. Title/abstract screening will begin next.
-
Fabrizio C, Termine A, Caltagirone C, Sancesario G. Artificial Intelligence for Alzheimer’s Disease: Promise or Challenge? Diagnostics. 2021;11(8):1473. doi:10.3390/diagnostics11081473
-
El-Sappagh S, Alonso JM, Islam SMR, Sultan AM, Kwak KS. A multilayer multimodal detection and prediction model based on explainable artificial intelligence for Alzheimer’s disease. Sci Rep. 2021;11(1):2660. doi:10.1038/s41598-021-82098-3
-
Sarker IH. Deep Learning: A Comprehensive Overview on Techniques, Taxonomy, Applications and Research Directions. SN Comput Sci. 2021;2(6):420. doi:10.1007/s42979-021-00815-1
-
Frizzell TO, Glashutter M, Liu CC, et al. Artificial intelligence in brain MRI analysis of Alzheimer’s disease over the past 12 years: A systematic review. Ageing Res Rev. 2022;77:101614. doi:10.1016/j.arr.2022.101614
-
Jo T, Nho K, Saykin AJ. Deep Learning in Alzheimer’s Disease: Diagnostic Classification and Prognostic Prediction Using Neuroimaging Data. Front Aging Neurosci. 2019;11. doi:10.3389/fnagi.2019.00220
