Category: Technology
Objective: The study aims at detecting patients affected with Parkinson’s disease from their vocal measurements using Artificial Intelligence Techniques.
Background: The Parkinson’s Disease (PD) is a progressive nervous disorder that impairs the movement during disease progression. The symptoms include tremors, slowness of movement, poor balance and speech difficulty. Diagnosis of PD at earlier stages is of prime significance. The diagnosis is usually done by a neurologist who relays on general neurological examination, rather than a specific test. There is a dire need to develop a specific test for diagnosis of PD, to eliminate personal biases.
Method: A range of biomedical voice measurements are collected from 31 people, which includes 23 PD patients. A total of about 195 voice recordings are collected from these individuals [1]. These recordings are fed into the AI model. Various Machine learning models are used for this diagnosis. The Training : Testing dataset are in the ratio 80 : 20. Several vocal parameters like Average Vocal Fundamental Frequency, Maximum Vocal Fundamental Frequency, Minimum Vocal Fundamental Frequency, several measures of variations in Amplitude, Jitter and Shimmer are evaluated from these voice recordings. The score of 0 for healthy individual and 1 for an individual affected with PD is assigned. The performance measures like accuracy, sensitivity, specificity and F1-Score of the various models are evaluated.
Results: The comparison of performance measures of the various models is listed in Table 1 [table1]. From the table, it is evident that AdaBoost classifier has given the best outcome with an accuracy, sensitivity, specificity and F1-Score of 0.9719, 0.9627, 0.9788 and 0.9753 respectively.
Conclusion: From the results, it is evident that AdaBoost classifier tool serves as a suitable candidate for a PD-specific test. This tool if employed in patient care, could aid in the diagnosis of PD by neurologists. This could eliminate the false identification or mis-identification by neurologists in diagnosing PD.
References: [1] https://www.kaggle.com/nidaguler/parkinsons-data-set
To cite this abstract in AMA style:
SJS. Rajasekar, V. Narayanan, V. Perumal. ParkAI – An AI Based Tool for Detection of Parkinson’s Disease using Vocal Measurements [abstract]. Mov Disord. 2021; 36 (suppl 1). https://www.mdsabstracts.org/abstract/parkai-an-ai-based-tool-for-detection-of-parkinsons-disease-using-vocal-measurements/. Accessed December 12, 2024.« Back to MDS Virtual Congress 2021
MDS Abstracts - https://www.mdsabstracts.org/abstract/parkai-an-ai-based-tool-for-detection-of-parkinsons-disease-using-vocal-measurements/