"From Genes to Symptoms: Predicting Huntington’s Disease Progression with DNN"

Huntington’s Disease (HD) is a progressive neurodegenerative disorder which has a global prevalence of 5 to 10 per 100,000, with a markedly higher prevalence in North America and Europe than in Asia and Africa. Its autosomal dominant inheritance puts over 200,000 people globally at risk. HD is a neurodegenerative disorder that is defined by the motor impairment, cognitive decline, and neurodegeneration, and there is no available treatment that cures the disease— hence the need for early-stage classification for successful disease management and intervention strategies. We propose a DNN-based classification model to predict the three stages of HD progression, which are Early, Middle, and Late. It is trained on multi-factorial data, including genetic, clinical and neuroimaging parameters such as HTT CAG repeat length, motor impairment severity, cognitive decline, functional capacity, mutation type, and protein aggregation levels. The proposed model attains 92.10% accuracy, which is better than traditional classification approaches. The cohort of patients was divided into three groups, showing that 68% of participants were categorized in the Middle Stage, 20% in the Early Stage and 12% in the Late Stage. The superior predictive performance of the DNN model highlights its potential as a powerful computational tool that can assist in clinical decision-making, promoting timely interventions, personalized treatment options, and improved disease management. Further improvements in HD classification accuracy and prognostic modelling will be investigated with the incorporation of biomarker-based predictions/labels (e.g., genomics) as well as multimodal imaging and longitudinal patient tracking in future research.