Deep learning and Machine learning predictive models for neurodegenerative disease detection: A Systematic Review of the Literature

Frank C.-P.; Juan S.-Q.; Carlos Y.-C. · Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology (2025) · DOI: 10.18687/laccei2025.1.1.547

Resumen

In recent years, artificial intelligence has revolutionized various fields of medicine, particularly in the early detection of neurodegenerative diseases (ND). This study analyzes Machine Learning (ML) and Deep Learning (DL) models applied to the detection of neurodegenerative diseases. To achieve this, a systematic literature review (SLR) was conducted following the PICO method to guide the search for relevant studies in the SCOPUS and Web of Science databases, and the PRISMA statement was used for the final screening. Performance metrics such as accuracy, sensitivity, and area under the curve (AUC) were evaluated. Additionally, the Bibliometrix tool in R was used for an in-depth bibliometric analysis of the selected studies. Among the 30 articles finally included in the review, the most frequent approaches included Support Vector Machines (SVM) in ML, with 6 main studies, and Convolutional Neural Networks (CNN) in DL, with 11 prominent works. Notably, one SVM model achieved 100% accuracy, while the CNN-InceptionV4 architecture stood out in DL with 99% accuracy. DL models such as Graph Convolutional Networks (GCN), and hybrid approaches like CNN-GCN, proved more robust in managing complex data, whereas ML models offer advantages in terms of lower computational requirements. In conclusion, DL- and ML-based models represent a promising tool for the early detection of ND. However, their adoption in clinical practice requires further optimization to overcome technical limitations and ensure real-world applicability.

Instituciones: Universidad Tecnológica del Perú