Can bite detection algorithms substitute manual video annotations in elderly people with and without Parkinson's disease? An experimental study

Parkinson's disease, as a neurodegenerative disorder that affects movement, can significantly affect patients' nutrition, potentially leading to malnutrition and weight loss. Therefore, monitoring the eating behavior of Parkinsons' patients in real-time could better inform when interventions are needed to maintain or increase energy intake and improve the overall quality of life of patients, while reducing the disease severity. Traditional eating behavior analysis methods rely on self-reported measures, whose reliability is limited due to miss-reporting. This work aims to assess the ability of an automated algorithm to accurately idenitfy eating characteristics based on video inputs. Experimental results showthe proposed deep learning-based algorithm achieves a near-perfect agreement (correlation coefficient 0.95) with manual annotation of bite instances, thus paving the way for the creation of automated eating behavior monitoring systems with the potential to be integrated with current clinical practice for improved Parkinson's disease assessment and handling.