Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Deep Learning approach for Depressive Symptoms assessment in Parkinson's disease patients using facial videos

Ioannis Kyprakis, Vasileios Skaramagkas, Iro Boura, Georgios Karamanis, Dimitrios I. Fotiadis, Zinovia Kefalopoulou, Cleanthe Spanaki, Manolis Tsiknakis

TL;DR

This work tackles the detection and severity assessment of depressive symptoms in Parkinson's disease (PD) via facial video analysis using deep learning. It evaluates three architectures—ViViT, Video Swin Tiny, and a 3D CNN–LSTM with attention—on a large, in-clinic video dataset of PD patients captured under ON- and OFF-medication conditions and labeled with the Geriatric Depression Scale (GDS). The Video Swin Tiny model achieved the best performance, with binary classification accuracy up to $0.94$ and F1-score $0.937$, and multiclass accuracy up to $0.871$ with F1-score $0.854$, indicating strong capability to detect depressive symptoms and their severity from dynamic facial cues despite hypomimia. The results also showed generally higher performance in the ON-medication state, suggesting pharmacological stabilization of facial expressions improves signal quality. Overall, the study demonstrates a robust, noninvasive digital biomarker potential for PD-related depression, with implications for remote monitoring and personalized care; future work will aim at broader populations, multimodal data, and real-time deployment.

Abstract

Parkinson's disease (PD) is a neurodegenerative disorder, manifesting with motor and non-motor symptoms. Depressive symptoms are prevalent in PD, affecting up to 45% of patients. They are often underdiagnosed due to overlapping motor features, such as hypomimia. This study explores deep learning (DL) models-ViViT, Video Swin Tiny, and 3D CNN-LSTM with attention layers-to assess the presence and severity of depressive symptoms, as detected by the Geriatric Depression Scale (GDS), in PD patients through facial video analysis. The same parameters were assessed in a secondary analysis taking into account whether patients were one hour after (ON-medication state) or 12 hours without (OFF-medication state) dopaminergic medication. Using a dataset of 1,875 videos from 178 patients, the Video Swin Tiny model achieved the highest performance, with up to 94% accuracy and 93.7% F1-score in binary classification (presence of absence of depressive symptoms), and 87.1% accuracy with an 85.4% F1-score in multiclass tasks (absence or mild or severe depressive symptoms).

A Deep Learning approach for Depressive Symptoms assessment in Parkinson's disease patients using facial videos

TL;DR

This work tackles the detection and severity assessment of depressive symptoms in Parkinson's disease (PD) via facial video analysis using deep learning. It evaluates three architectures—ViViT, Video Swin Tiny, and a 3D CNN–LSTM with attention—on a large, in-clinic video dataset of PD patients captured under ON- and OFF-medication conditions and labeled with the Geriatric Depression Scale (GDS). The Video Swin Tiny model achieved the best performance, with binary classification accuracy up to and F1-score , and multiclass accuracy up to with F1-score , indicating strong capability to detect depressive symptoms and their severity from dynamic facial cues despite hypomimia. The results also showed generally higher performance in the ON-medication state, suggesting pharmacological stabilization of facial expressions improves signal quality. Overall, the study demonstrates a robust, noninvasive digital biomarker potential for PD-related depression, with implications for remote monitoring and personalized care; future work will aim at broader populations, multimodal data, and real-time deployment.

Abstract

Parkinson's disease (PD) is a neurodegenerative disorder, manifesting with motor and non-motor symptoms. Depressive symptoms are prevalent in PD, affecting up to 45% of patients. They are often underdiagnosed due to overlapping motor features, such as hypomimia. This study explores deep learning (DL) models-ViViT, Video Swin Tiny, and 3D CNN-LSTM with attention layers-to assess the presence and severity of depressive symptoms, as detected by the Geriatric Depression Scale (GDS), in PD patients through facial video analysis. The same parameters were assessed in a secondary analysis taking into account whether patients were one hour after (ON-medication state) or 12 hours without (OFF-medication state) dopaminergic medication. Using a dataset of 1,875 videos from 178 patients, the Video Swin Tiny model achieved the highest performance, with up to 94% accuracy and 93.7% F1-score in binary classification (presence of absence of depressive symptoms), and 87.1% accuracy with an 85.4% F1-score in multiclass tasks (absence or mild or severe depressive symptoms).
Paper Structure (11 sections, 11 equations, 2 figures, 3 tables)

This paper contains 11 sections, 11 equations, 2 figures, 3 tables.

Table of Contents

  1. INTRODUCTION
  2. Related Work
  3. Methodology
  4. Dataset
  5. Preprocessing
  6. Model Architectures
  7. Experiments and Results
  8. Experiments
  9. Results
  10. Discussion
  11. Conclusions

Figures (2)

  • Figure 1: The six tasks performed by the PD participants; i) open and close mouth, ii) still face, iii) smile and kiss, iv) tongue up and down, v) tongue in and out, vi) tongue right and left.
  • Figure 2: 3D CNN-LSTM with attention architecture.