The objective of this paper is to propose a new unsupervised machine learning-based fall tracking model for the monitoring of patients with neurodegenerative disorders such as Parkinson's disease which is physician oriented. This framework hence incorporates next-gen 3-axis accelerometer sensor attached to the patient’s belly region for the purpose of on-going movement tracking. The algorithm teams up high-level data coping, cleaning and feature set extraction techniques which lead to precise differentiation of fall events and routine activity. The usage of advanced machine learning classifiers, i.e., C4.5, Random Forest, and Logistic Model Trees, helps attain high precision levels. Precisely, the C4.5 classification is able to obtain the highest accuracy which is 97.36%. Unlike the traditional system that offers fall detection only after the event, this new method, in addition to saving patients’ safety, facilitating timely interventions, and thus reducing the costs, can also identify the condition before the event thereby preventing the fall itself. The implication is getting the best care possible for patients with incurable neurodegenerative diseases to ensure that the treatment is done per best practices.