Neurodegenerative diseases (NDs), including Parkinson's (PD) and Alzheimer's (AD) disease are devastating conditions that affect millions worldwide, with the number of cases expected to rise significantly in the coming years. Despite considerable advancements in understanding their pathophysiology, etiology, and treatment, there is still a lack of effective disease-modifying interventions. Currently, no cure exists and there is an urgent need for modern tools that allow precise detection and objective severity scoring for the development of new therapeutic targets and approaches. Therefore, this study evaluates the effectiveness of an online version of the Trail Making Test Part A and B (TMT A and TMT B), incorporating time-based measures, to recognize cognitive and motor manifestations of Parkinson's disease severity. For validation, this research was conducted with 15 Parkinson’s patients under care at UMass Chan Medical School. This study applied the TMT sensitivity to executive function impairments by measuring response and reaction times, to correlate these with stages of PD severity. Machine learning models (Naïve Bayes, Logistic Regression, Support Vector Machine, and Random Forest) were used to predict the disease severity based on TMT performance. Among these, Random Forest was the most effective, achieving scores with an Area Under the Curve (AUC) of 0.92 (80% accuracy), indicating good performance in distinguishing between mild and advanced stages of PD. Although limited by a small sample size, this preliminary study highlights the role of digital tools in enhancing PD diagnostics and monitoring. Future research with larger cohorts and longitudinal designs is essential to validate these preliminary findings and further develop digital diagnostics as crucial in the fight against neurodegenerative diseases.