Unmasking Parkinson's Disease with Smile: An AI-enabled Screening Framework

Parkinson's disease (PD) diagnosis remains challenging due to lacking a reliable biomarker and limited access to clinical care. In this study, we present an analysis of the largest video dataset containing micro-expressions to screen for PD. We collected 3,871 videos from 1,059 unique participants, including 256 self-reported PD patients. The recordings are from diverse sources encompassing participants' homes across multiple countries, a clinic, and a PD care facility in the US. Leveraging facial landmarks and action units, we extracted features relevant to Hypomimia, a prominent symptom of PD characterized by reduced facial expressions. An ensemble of AI models trained on these features achieved an accuracy of 89.7% and an Area Under the Receiver Operating Characteristic (AUROC) of 89.3% while being free from detectable bias across population subgroups based on sex and ethnicity on held-out data. Further analysis reveals that features from the smiling videos alone lead to comparable performance, even on two external test sets the model has never seen during training, suggesting the potential for PD risk assessment from smiling selfie videos.

Real-time Seismic Intensity Prediction using Self-supervised Contrastive GNN for Earthquake Early Warning

Seismic intensity prediction in a geographical area from early or initial seismic waves received by a few seismic stations is a critical component of an effective Earthquake Early Warning (EEW) system. State-of-the-art deep learning-based techniques for this task suffer from limited accuracy in the prediction and, more importantly, require input waveforms of a large time window from a handful number of seismic stations, which is not practical for EEW systems. To overcome the above limitations, in this paper, we propose a novel deep learning approach, Seismic Contrastive Graph Neural Network (SC-GNN) for highly accurate seismic intensity prediction using a small portion of initial seismic waveforms received by a few seismic stations. The SC-GNN comprises two key components: (i) a graph neural network (GNN) to propagate spatiotemporal information through the nodes of a graph-like structure of seismic station distribution and wave propagation, and (ii) a self-supervised contrastive learning component to train the model with larger time windows and make predictions using shorter initial waveforms. The efficacy of our proposed model is thoroughly evaluated through experiments on three real-world seismic datasets, showing superior performance over existing state-of-the-art techniques. In particular, the SC-GNN model demonstrates a substantial reduction in mean squared error (MSE) and the lowest standard deviation of the error, indicating its robustness, reliability, and a strong positive relationship between predicted and actual values. More importantly, the model maintains superior performance even with 5s input waveforms, making it particularly efficient for EEW systems.