Deep Learning Pose Estimation for Phenotyping of Co‐Occurring Hyperkinetic Movement Disorders
Abstract
OBJECTIVE: To explore whether routine outpatient video combined with deep learning-based pose estimation and clinically interpretable kinematic features can support multi-label phenotyping of co-occurring hyperkinetic movement disorders (HMDs). METHODS: In this exploratory single-centre proof-of-concept study, videos from 21 patients with HMDs and 4 healthy controls were processed with markerless pose estimation (YOLOv8) and 2-dimensional keypoint trajectories transformed into kinematic descriptors spanning statistical, temporal, spectral, and complexity domains. Ten-second windows were aligned to expert annotations for eight hyperkinetic phenomenologies. Conventional supervised classifiers were trained on these tabular descriptors. Window-level predictions were aggregated to the patient level, and label-specific thresholds were tuned on training participants only. RESULTS: In patient-level multi-label performance reporting, (i) the best single pipeline selected by discrimination (StandardScaler + MLP) achieved a macro-AUPRC of 0.821 ± 0.019 and a macro-receiver operating characteristic area under the curve of 0.830 ± 0.029. (ii) The best single pipeline selected by Hamming accuracy (MinMaxScaler + SVM) reached 0.764 ± 0.041. (iii) Under prespecified nested cross-validation with per-label model selection within training folds (primary analysis), macro-AUPRC was 0.717 ± 0.030, macro-AUROC was 0.767 ± 0.069, Hamming accuracy was 0.764 ± 0.014 and patient-label agreement was 153/200 (76.5%). (iv) Post hoc per-label selection of the best-performing pipeline defined an exploratory upper bound of 172/200 (86.0%). INTERPRETATION: In this exploratory study, a hybrid pipeline combining deep learning pose estimation with feature-engineered supervised classification produced encouraging patient-level multi-label performance for co-occurring HMDs. These findings are proof-of-concept; external, multicentre, prospective validation is required before clinical or trial use.
Related
- Simultaneous hyperkinetic movement disorders phenotyping: a cross-cohort pediatric transfer study using routine videos, markerless pose estimation and a tabular foundation model
- Foundation-model-based multi-label phenotyping of combined hyperkinetic movement disorders
- Non-destructive estimation of individual buckwheat plant biomass using high-throughput phenotyping-derived canopy traits
- Aerial Phenotyping of Coffee Flowering Using Deep Learning for Early Yield Prediction
- Multi-temporal UAV multispectral phenotyping and machine learning for modeling cotton defoliation dynamics across diverse accessions
BibTeX
@article{cif2026deep,
title = {{Deep Learning Pose Estimation for Phenotyping of Co‐Occurring Hyperkinetic Movement Disorders}},
author = {Laura Cif and Diane Demailly and Gabriella Horvàth and Juan Dario Ortigoza Escobar and Nathalie Dorison and Mayté Castro Jiménez and Cécile Hubsch and Thomas Wirth and Gun‐marie Hariz and Sophie Huby and Morgan Dornadic and Zohra Souei and Muhammad Mushhood Ur Rehman and Simone Hemm and Mehdi Boulaymen and Eduardo Moraud and Jocelyne Bloch and Xavier Vasques},
journal = {Annals of Clinical and Translational Neurology},
year = {2026},
month = jul,
doi = {10.1002/acn3.70474},
eprint = {2602.00163},
archivePrefix = {arXiv},
url = {https://hal.science/hal-05795909v1},
note = {CC BY},
}
RIS
TY - JOUR
TI - Deep Learning Pose Estimation for Phenotyping of Co‐Occurring Hyperkinetic Movement Disorders
AU - Laura Cif
AU - Diane Demailly
AU - Gabriella Horvàth
AU - Juan Dario Ortigoza Escobar
AU - Nathalie Dorison
AU - Mayté Castro Jiménez
AU - Cécile Hubsch
AU - Thomas Wirth
AU - Gun‐marie Hariz
AU - Sophie Huby
AU - Morgan Dornadic
AU - Zohra Souei
AU - Muhammad Mushhood Ur Rehman
AU - Simone Hemm
AU - Mehdi Boulaymen
AU - Eduardo Moraud
AU - Jocelyne Bloch
AU - Xavier Vasques
PY - 2026
DA - 2026/07/25
JO - Annals of Clinical and Translational Neurology
DO - 10.1002/acn3.70474
UR - https://hal.science/hal-05795909v1
AB - OBJECTIVE: To explore whether routine outpatient video combined with deep learning-based pose estimation and clinically interpretable kinematic features can support multi-label phenotyping of co-occurring hyperkinetic movement disorders (HMDs). METHODS: In this exploratory single-centre proof-of-concept study, videos from 21 patients with HMDs and 4 healthy controls were processed with markerless pose estimation (YOLOv8) and 2-dimensional keypoint trajectories transformed into kinematic descriptors spanning statistical, temporal, spectral, and complexity domains. Ten-second windows were aligned to expert annotations for eight hyperkinetic phenomenologies. Conventional supervised classifiers were trained on these tabular descriptors. Window-level predictions were aggregated to the patient level, and label-specific thresholds were tuned on training participants only. RESULTS: In patient-level multi-label performance reporting, (i) the best single pipeline selected by discrimination (StandardScaler + MLP) achieved a macro-AUPRC of 0.821 ± 0.019 and a macro-receiver operating characteristic area under the curve of 0.830 ± 0.029. (ii) The best single pipeline selected by Hamming accuracy (MinMaxScaler + SVM) reached 0.764 ± 0.041. (iii) Under prespecified nested cross-validation with per-label model selection within training folds (primary analysis), macro-AUPRC was 0.717 ± 0.030, macro-AUROC was 0.767 ± 0.069, Hamming accuracy was 0.764 ± 0.014 and patient-label agreement was 153/200 (76.5%). (iv) Post hoc per-label selection of the best-performing pipeline defined an exploratory upper bound of 172/200 (86.0%). INTERPRETATION: In this exploratory study, a hybrid pipeline combining deep learning pose estimation with feature-engineered supervised classification produced encouraging patient-level multi-label performance for co-occurring HMDs. These findings are proof-of-concept; external, multicentre, prospective validation is required before clinical or trial use.
SN - 2328-9503
SN - 2328-9503
ER -