Artificial Intelligence for Early Detection of Motor Neuron Disease Using Gait Analysis and Speech Patterns in Pekanbaru, Indonesia

Abstract

Introduction: Motor neuron disease (MND) is a devastating neurodegenerative disorder characterized by progressive muscle weakness, atrophy, and ultimately, paralysis. This study investigated the potential of artificial intelligence (AI) to detect MND in its early stages using gait analysis and speech pattern recognition in a population in Pekanbaru, Indonesia.

Methods: A cross-sectional study was conducted at the Neurology Department of a tertiary referral hospital in Pekanbaru, Indonesia. A total of 150 participants aged 40-75 years were recruited and categorized into three groups. Gait analysis was performed using wearable sensors to collect data on stride length, cadence, swing time, stance time, and gait variability. Machine learning algorithms, including support vector machines (SVM), random forest (RF), and deep learning models like convolutional neural networks (CNN), were trained on the combined gait and speech data to classify participants into the three groups.

Results: Significant differences were observed in gait parameters between the MND group and the other two groups. Individuals with MND exhibited shorter stride length (p<0.001), slower cadence (p<0.001), increased swing time variability (p=0.002), and reduced stance time (p=0.003). Speech analysis revealed distinct patterns in the MND group, including reduced speech rate (p<0.001), increased pause duration (p=0.004), and decreased vocal intensity (p=0.001). The AI models, particularly the CNN model, demonstrated high accuracy in differentiating individuals with MND from healthy controls and those with other neurological conditions. The CNN model achieved an accuracy of 94.7%, sensitivity of 92%, specificity of 96%, and an area under the receiver operating characteristic curve (AUC) of 0.98.

Conclusion: AI-powered gait analysis and speech pattern recognition show promise as a non-invasive and cost-effective tool for the early detection of MND in Pekanbaru, Indonesia. This technology has the potential to improve diagnostic accuracy and facilitate timely intervention, ultimately enhancing the quality of life for individuals with MND.