Deep learning and Machine learning predictive models for neurodegenerative disease detection: A Systematic Review of the Literature (#547)

Authors

Cataño Portocarrero, Frank Josue

Soria Quijaite, Juan Jesus

Yapo Cáceres, Carlos Iván

Abstract

In recent years, artificial intelligence has revolutionized various fields of medicine, highlighting its impact on the early detection of neurodegenerative diseases (ND). This study analyzes Machine Learning (ML) and Deep Learning (DL) models applied in the detection of neurodegenerative diseases. For this purpose, a systematic literature review (SLR) was performed following the PICO method in the search of information from SCOPUS, Web of Science and the PRISMA statement for the final screening, evaluating metrics such as accuracy, sensitivity and area under the curve (AUC). Likewise, the Bibliometrix of R study was used to evaluate in depth the studies collected from the 2 databases analyzed. Among the most prominent studies, the most frequent approaches include support vector machines (SVM) in ML, with 6 main investigations, and convolutional neural networks (CNN) in DL, with 11 outstanding studies. In addition, one SVM model achieved 100% accuracy, while CNN-InceptionV4 stands out in DL with 99% accuracy. DL models, such as GCN and advanced combinations such as CNN-GCN, have proven to be more robust in handling complex data, while ML approaches present advantages in terms of lower computational demand. In conclusion, DL- and ML-based models represent a promising tool for early detection of ND. However, their adoption in clinical practice requires further optimizations to overcome technical barriers and ensure their applicability in real-world scenarios.