Abstract:

Parkinson's disease (PD) sufferers are increasing in number alongside the ageing populace. Due to a lack of training and information, PD is often not diagnosed in a timely manner in developing nations. On top of that, not every person with PD experiences the same symptoms, nor do those symptoms always become more noticeable at the same point in time. Thus, the goal of this work is to utilise a cloud-based machine learning system for tele monitoring PD patients in developing countries, combining more than one symptom (rest tremor and voice degradation) to make a diagnosis. Data on rest tremor and vowel phonationare taken from smartphones equipped with accelerometers and fed into the proposed system. For the purposes of developing and refining more effective machine learning models, the data are mainly gathered from individuals who have been identified with PD and from the general population. The accuracy of the taught algorithms in detecting PD is then assessed by collecting data from freshly suspected PD patients. Patients diagnosed with PD are referred to a local physician for evaluation based on the results of these methods

Description:

Vocal Recording; Tremor Recording; Machine Learning; Python – keras, Tensor flow; Accuracy