Facial Landmark Video Analysis for Remote TD Severity Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Tardive dyskinesia (TD) is often underdiagnosed due to the time-consuming and subjective nature of conventional Abnormal Involuntary Movement Scale (AIMS) tests, which require in-person visits to trained clinicians and can lead to inconsistent results, deterring patients from seeking timely diagnosis or treatment.
Innovation Solution
A method and system using machine learning models to analyze video data of facial movements, identifying facial landmarks, and generating severity scores for TD symptoms, enabling remote and standardized assessment of TD risk and severity through mobile devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional AIMS test is used, then TD diagnosis can be obtained, but the test is time-consuming and requires in-person clinician evaluation
Solution Approach 1:
The patent creates a digital copy of the AIMS test by capturing video footage of patient facial movements and analyzing them through machine learning models. This virtual copy replicates the diagnostic functionality of the in-person test while eliminating the need for physical clinician presence, thereby reducing time loss while maintaining diagnostic accuracy.
Solution Approach 2:
The patent replaces the mechanical system of in-person clinician evaluation with an automated computer vision system. The machine learning models process video data to detect facial movements and generate AIMS scores, substituting human mechanical assessment with algorithmic analysis, which significantly reduces time requirements while preserving diagnostic precision.
2Measurement precision
If conventional AIMS test is used, then TD diagnosis can be obtained, but the test relies on subjective judgments leading to inconsistent results
Solution Approach 1:
The patent replaces subjective human judgment with objective machine learning analysis. The system automatically detects facial landmarks and quantifies movements through standardized algorithms, eliminating the variability inherent in human subjective assessment and ensuring consistent diagnostic results across different patients and time points.
Solution Approach 2:
The patent transforms subjective clinical observations into objective quantitative parameters by measuring specific facial landmark coordinates and movement distances. This parameterization converts the assessment from qualitative subjective judgment to quantitative objective measurement, improving result consistency while reducing assessment complexity.
3Reliability
If frequent AIMS testing is conducted, then TD monitoring can be improved, but patient burden increases leading to treatment abandonment
Solution Approach 1:
The patent enables remote copying of the AIMS test through video capture, allowing patients to complete assessments from their own environments without traveling to clinical facilities. This dramatically reduces patient burden while maintaining monitoring reliability through standardized automated analysis of the video footage.
Solution Approach 2:
The patent implements a self-service assessment system where patients independently complete the AIMS test by recording their own facial movements at home. The automated machine learning system processes the video data without requiring clinician involvement, making frequent monitoring feasible and reducing patient burden while preserving diagnostic reliability.
Data Source
AI summary
The present disclosure relates to systems and methods for assessing severity of involuntary movement associated with tardive dyskinesia (TD). The method includes receiving video data of the patient. The video data may include facial movements of the patient. The method includes processing the video data to identify facial landmark data of the patient, which may include applying an image processing technique to the video data to identify facial landmarks on one or more frames of the video data to produce labeled frames. The method may include applying a plurality of trained machine learning models to the facial landmark data to determine one or more movement severity scores based on changes in the facial landmark data of the patient, where each movement severity score is representative of a category of facial movement, and a total severity score may be generated based on multiple movement severity scores.