The PD patients I have known never smiled and showed little emotion on the face. Hence I came up with this hypothesis. If the answer is positive, it would be of clinical value.
Srinivasan Ramani I am not into sciences but from the technology end, I can shed some light:
I recently researched video data stream analysis using AI and read about PD use case implemented at the University of Florida. This approach involves analyzing a series of videos taken over several weeks to perform key point detection (You can try Google MediaPipe). Additionally, applying the K-nearest neighbor (KNN) algorithm helps identify classification boundaries. By combining these techniques, it's possible to quantify and predict the likelihood of Parkinson's Disease.
There is plenty of literature about gait analysis, just check on googlescholar.
Regarding emotions, I think It is possible for AI systems to analyze videos of a person in conversation and look for patterns of movement and facial expressions that could suggest PD. PD often leads to motor symptoms such as tremors, rigidity, and bradykinesia (slowness of movement), as well as changes in facial expressions, sometimes referred to as "masked facies," where a person shows reduced facial mobility or a diminished range of emotional expression, like not smiling or showing little emotion. Also this idea has been explored a bit, check "An integrated biometric voice and facial features for early detection of Parkinson’s disease" by Lim et al. Parkinson disease 2022, going back to the idea if this is of clinical value.... it depends. Maybe if you can show that improvements can correlate with treatments like Levadopa, I think anything that shows improvements of rigidity of muscles (movements or empathy) is welcome.
As a Movement Disorders specialist can say that facial expression is the point we assess within UPDRS scale, but it is not the only one feature. The diagnosis is made clinically on minimum 2 of 3 clinical signs (bradykinesia + rigidity and/or rest tremor). It seems that the error occurrence can be huge if only apply on AI based facial expression assessment. But may be when the physicians society will come to consensus about digital biomarkers on PD, we will have these features as the diagnostic one. But it seems to be a long process IMHO
- Badges
- Science topic