JMIR Publications

ABSTRACT

Background:

Nowadays, the number of patients with neurological diseases grows up year by year, which presents tremendous challenges to 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 get consultation, track their diseases and receive rehabilitation training in more convenient and efficient ways. In this paper, we explore to diagnose a typical neurological system disease, the Parkinson’s disease through facial expression recognition via artificial intelligence.

Objective:

This study proposed some methods to diagnose the Parkinson’s disease (PD) through facial expression recognition.

Methods:

We collected some videos recording the facial expressions of people with Parkinson’s disease and matched controls respectively. We used relative coordinates and positional Jitter to extract facial expression features (the facial expression amplitude features and the facial small muscle group shaking features) from the key points returned by Face++. Some algorithms in traditional machine learning and advanced deep learning area were utilized for diagnosis of PD.

Results:

Experimental results showed our models can achieve outstanding facial expression recognition ability for PD diagnosis. Applying Long Short-term Model (LSTM) neural network to the positions of key points features, the Precision and F1-value can reach 86% and 75% respectively. Further, utilizing Support Vector Machine (SVM) algorithm on the facial expression amplitude features, i.e. the facial small muscle group shaking features, the F1-value can be 99%.

Conclusions:

This study contributed to digital diagnosis of PD based on facial expression recognition. The disease diagnosis model was validated by our experiment. It can help doctors understand the real-time disease dynamics and even conduct remote diagnosis.