The number of patients with neurological diseases is currently increasing annually, which presents tremendous challenges for both patients and doctors. With the advent of advanced information technology, digital medical care is gradually changing the medical ecology. Numerous people are exploring new ways to receive a consultation, track their diseases, and receive rehabilitation training in more convenient and efficient ways. In this paper, we explore the use of facial expression recognition via artificial intelligence to diagnose a typical neurological system disease, Parkinson disease (PD).

BACKGROUND Parkinson’s disease (PD) is one of the most common neurological diseases. At present, because the exact cause is still unclear, accurate diagnosis and progression monitoring remain challenging. In recent years, exploring the relationship between Parkinson's disease and speech disorder has attracted widespread attention in the academic world. Most of the work successfully validate the effectiveness of some vocal features. Moreover, the non-invasive nature of speech signal-based testing has pioneered a new way to realize telediagnosis and telemonitoring. In particular, there is an increasing demand for AI-powered tools in the digital health era. OBJECTIVE This study aimed to build a real-time speech signal analysis tool for PD diagnosis and severity assessment. Further, the underlying system should be scalable to integrate any traditional machine learning or advanced deep learning algorithms. METHODS At its core, the system consists of two parts: 1) Speech signals processing: both traditional and novel speech signal processing technologies were used for feature engineering, which can automatically extract linear and nonlinear dysphonia features. 2) Application of machine learning algorithms: some classical regression and classification algorithms from machine learning field were tested, and we then chose the most efficient algorithms and relevant features. RESULTS Experimental results showed our system’s outstanding ability in both PD diagnosis and severity assessment. If we used both linear and nonlinear dysphonia features, Support Vector Machine (SVM) achieved the best results with accuracy 88.74% and recall 97.03% in the diagnosis task. Meanwhile, in the assessment task, Support Vector Regression (SVR) performed best with mean absolute error (MAE) 3.7699. Moreover, the system has a well-designed architecture that is scalable to advanced algorithms and it has already been deployed into a mobile application called “No Pa”. CONCLUSIONS This study explored the diagnosis and severity assessment of PD from speech order detection perspective. The efficiency and effectiveness our tool contribute to telediagnosis and telemonitoring of the Parkinson’s disease. The real-time feedback can also optimize the decision-making process of the doctor's treatment. CLINICALTRIAL N/A

No abstract