Adaptive Machine Learning Head Model Across Different Head Impact Types Using Unsupervised Domain Adaptation and Generative Adversarial Networks

Xianghao Zhan,Jiawei Sun,Yuzhe Liu,Nicholas J. Cecchi,Enora Le Flao,Olivier Gevaert,Michael M. Zeineh,David B. Camarillo

DOI: https://doi.org/10.1109/jsen.2023.3349213

IF: 4.3

2024-03-02

IEEE Sensors Journal

Abstract:Machine learning head models (MLHMs) are developed to estimate brain deformation from sensor-based kinematics for early detection of traumatic brain injury (TBI). However, the overfitting to simulated impacts and the decreasing accuracy caused by distributional shift of different head impact datasets hinder the broad clinical applications of current MLHMs. We propose a new MLHM configuration that integrates unsupervised domain adaptation with a deep neural network (DNN) to predict whole-brain maximum principal strain (MPS) and MPS rate (MPSR). With 12780 simulated head impacts, we performed unsupervised domain adaptation on target head impacts from 302 college football (CF) impacts and 457 mixed martial arts (MMA) impacts using domain regularized component analysis (DRCA) and cycle-generative adversarial network (GAN)-based methods. The new model improved the MPS/MPSR estimation accuracy, with the DRCA method outperforming other domain adaptation methods in prediction accuracy: MPS mean absolute error (MAE): 0.017 (CF) and 0.020 (MMA); MPSR MAE: (CF) and (MMA). On another two hold-out test sets with 195 CF impacts and 260 boxing impacts, the DRCA model outperformed the baseline model without domain adaptation in MPS and MPSR estimation MAE. The DRCA domain adaptation approach reduces the error of MPS/MPSR estimation to be well below previously reported TBI thresholds, enabling accurate brain deformation estimation to detect TBI in future clinical applications.

engineering, electrical & electronic,instruments & instrumentation,physics, applied