Applications of machine learning and multimodal integration for the early diagnosis of neurodegenerative diseases (Review)

Neurodegenerative disorders (NDDs) such as Alzheimer's disease, Parkinson's disease and amyotrophic lateral sclerosis are critical worldwide health issues. Recent diagnostic methods primarily rely on biomarkers and clinical evaluations, often exhibiting insufficient specificity and sensitivity during the initial stages of illness. The present review discusses the machine learning (ML) techniques used to enhance the early prediction and detection of NDDs. The use of ML in analyzing many data modalities, including genetic biomarkers, molecular and cellular biomarkers, neuroimaging data, and cognitive/behavioral evaluations is also discussed. Research with ML techniques, including convolutional neural networks, support vector machines and recurrent neural networks has demonstrated substantial improvements in diagnostic precision for numerous NDDs, often exceeding conventional methodologies. Moreover, multimodal integration techniques that integrate various types of data further enhance prediction power. However, despite the positive results, challenges such as data standardization, privacy concerns and the requirement for robust validation across numerous populations persist. Addressing these challenges will be crucial for translating the potential of ML into clinically impactful tools for the early diagnosis, personalized treatment and improved management of NDDs.