Artificial Intelligence-Driven Biomarker Discovery in Huntington’s Disease: Integrating Multi-Omics and Multimodal Data

Abstract

Huntington’s disease (HD) is an autosomal dominant neurodegenerative disorder caused by an expanded CAG repeat in the HTT gene. Although the causal mutation is established, disease onset and progression vary among individuals, creating a need for biomarkers that capture molecular injury, neurodegeneration, disease stage, and treatment response. Artificial intelligence (AI), particularly machine learning, can integrate high-dimensional clinical, neuroimaging, electrophysiological, and molecular datasets to identify multidimensional biomarker signatures. This review examines primary research on AI-assisted biomarker discovery in HD, with emphasis on the integration of genomics, transcriptomics, proteomics, metabolomics, epigenomics, neuroimaging, EEG, gait, speech, and circulating biofluid markers. Structural and functional MRI provide established measures of disease-related brain changes, while machine-learning approaches can combine distributed imaging features into composite biomarkers. Biofluid studies support neurofilament light (NfL), mutant huntingtin, microRNAs, proteins, and metabolites as complementary biomarker candidates. Transcriptomic, epigenomic, and multi-omic studies further identify molecular and cell-specific alterations that may improve biological characterization of disease progression. AI-based analysis of EEG, speech, gait, and wearable-sensor data offers additional opportunities for objective and potentially repeated disease monitoring. However, translation remains limited by small sample sizes, cohort heterogeneity, high-dimensional datasets, overfitting, limited external validation, and challenges in interpreting model-derived features biologically.