Related Experiment Video
Updated: May 24, 2025
Quantified Assessment of Infant's Gross Motor Abilities Using a Multisensor Wearable
Published on: May 17, 2024
Clustering of Infant Poses as a Novel Metric of Data Diversity for Automated General Movement Assessments
Insights
This study introduces k-means clustering with silhouette analysis to assess infant pose diversity in videos. This method helps create better training datasets for automated General Movements (GM) tracking, improving early detection of neurological impairments.
Area of Science:
- Neuroscience
- Developmental Pediatrics
- Computer Science
Background:
- General Movements (GMs) monitoring in infants (6-20 weeks post-term) is crucial for assessing neurodevelopmental quality and identifying neurological impairment risks.
- Deep learning algorithms excel at markerless pose estimation for infant tracking in clinical videos.
- Algorithm performance on unseen data depends heavily on diverse training pose data.
Purpose of the Study:
- To introduce a novel, objective method for assessing pose diversity in infant movement data.
- To enhance the development of robust, generalized automated General Movement (GM) tracking classifiers.
- To improve the selection of diverse poses for machine learning training sets.
Main Methods:
- Utilized k-means clustering algorithm on infant pose data.
- Employed silhouette analysis to determine the optimal number of clusters (k) for pose data.
- Validated the clustering results against the cosine similarity index for pose similarity measurement.
Main Results:
- K-means clustering with silhouette analysis provides a convenient and objective metric for pose diversity assessment.
- The results demonstrated strong agreement with the cosine similarity index, confirming the method's validity.
- The approach successfully quantifies pose diversity within infant video datasets.
Conclusions:
- The k-means clustering method offers a reliable way to assess and ensure pose diversity in training datasets for infant GM analysis.
- This technique is vital for improving the generalization capabilities of automated GM tracking algorithms, especially for 'out-of-domain' data.
- Findings support the development of more accurate and reliable infant neurodevelopmental assessment tools.
Abstract:
Monitoring General Movements (GMs) in infants between 6-20 weeks post-term age serves as a reliable prognostic tool for assessing neurodevelopmental quality and determining the risk of neurological impairments in early infancy. Recent studies have demonstrated the power of deep-learning algorithms for markerless pose-estimation and tracking of infants' anatomical landmarks in clinically recorded videos from handheld devices. However, the performance of these algorithms on 'out-of-domain' unseen recordings is largely reliant on using training sets containing a diverse range of poses. This research demonstrates the use of an established clustering algorithm, k-means with silhouette analysis to determine the optimal number of clusters (k), on infant pose data. This method promises to be a convenient, objective metric to assess pose diversity within a set of video frames. We show that the results of this process agree well with another method of measuring pose similarity, the cosine similarity index.Clinical relevance- Findings hold promise for the identification and selection of a set of diverse poses for inclusion in the training set of machine-learning algorithms that are crucial for designing generalized automated General Movement (GM) tracking machine-learning classifiers to tackle challenges associated with 'out-of-domain' unseen poses.
