The Climber's Grip -- Personalized Deep Learning Models for Fear and Muscle Activity in Climbing

Matthias Boeker; Dana Swarbrick; Ulysse T. A. Côté-Allard; Marc T. P. Adam; Hugo L. Hammer; Pål Halvorsen

cs.LGMar 27, 2026

The Climber's Grip -- Personalized Deep Learning Models for Fear and Muscle Activity in Climbing

Authors: Matthias Boeker, Dana Swarbrick, Ulysse T. A. Côté-Allard, Marc T. P. Adam, Hugo L. Hammer, Pål Halvorsen

Abstract

Climbing is a multifaceted sport that combines physical demands and emotional and cognitive challenges. Ascent styles differ in fall distance with lead climbing involving larger falls than top rope climbing, which may result in different perceived risk and fear. In this study, we investigated the psychophysiological relationship between perceived fear and muscle activity in climbers using a combination of statistical modeling and deep learning techniques. We conducted an experiment with 19 climbers, collecting electromyography (EMG), electrocardiography (ECG) and arm motion data during lead and top rope climbing. Perceived fear ratings were collected for the different phases of the climb. Using a linear mixed-effects model, we analyzed the relationships between perceived fear and physiological measures. To capture the non-linear dynamics of this relationship, we extended our analysis to deep learning models and integrated random effects for a personalized modeling approach. Our results showed that random effects improved model performance of the mean squared error (MSE), mean absolute error (MAE) and root mean squared error (RMSE). The results showed that muscle fatigue correlates significantly with increased fear during lead climbing. This study highlights the potential of combining statistical and deep learning approaches for modeling the interplay between psychological and physiological states during climbing.

Figures & tables

Figure 1: The correlation matrix illustrates pairwise repeated-measures correlation coefficients and false discovery rate corrected p-values between the subjective and physiological variables. Each cell in the heatmap represents the correlation coefficient and significance of that specific pair. Only significant correlations are visualized. * p < .05, ** p < .01, * p < .001..

Figure 2: The relationship between MNF and self-reported fear, conditioned by climbing ascent style. The figure illustrates a steeper negative slope for lead climbing compared to top rope climbing, indicating a stronger association between MNF and fear in lead climbing.

Figure 3: The figure shows the relationship between iEMG and self-reported fear across two climbing styles. Although lead climbing shows a slightly steeper increase in self-reported fear with iEMG compared to top rope climbing, the interaction term was not statistically significant.

Model	Fixed Effect	Estimate	Std. Error	p-value
MNF	MNF	-0.0008	0.033	.980
	Intercept	5.236	2.595	.046 *
	Lead	13.650	2.822	<.001 *
	Interval2	1.054	0.620	.092
	Interval3	3.516	0.649	<.001 *
	Interval4	4.628	0.686	<.001 *

Table 1: Summary of linear mixed model analyses of the effect of muscle activity on fear. The table shows the estimates, standard errors, and p-values for the regression models. Significant parameters (

\alpha<.05

) are annotated with a single asterisk (*).

Panel of Figure 4 — Figure 4: The figure compares the four subjective scores measured at each interval and for lead and top rope climbing , displayed as line plots with standard error bars. We observe that participant consistently reported higher scores during lead climbing than top rope climbing .

Metric	GRUNN	GRUNNRE	Linear	LinearRE	NN	NNRE
MSE	$11.19\pm 0.99$	$6.25\pm 1.39$	$27.04\pm 0.46$	$19.24\pm 4.08$	$11.52\pm 1.07$	$8.96\pm 2.42$
MAE	$1.91\pm 0.11$	$1.42\pm 0.18$	$4.23\pm 0.01$	$2.94\pm 0.4$	$2.09\pm 0.08$	$1.78\pm 0.21$
RMSE	$3.34\pm 0.15$	$2.49\pm 0.27$	$5.2\pm 0.04$	$4.37\pm 0.47$	$3.39\pm 0.16$	$2.97\pm 0.39$

Table 2: Comparison of prediction performance between random effect models and the standard models across different error metrics. The results show a consistent improvement in all three evaluation metrics for the random effects compared to the standard models.

Metric	GRUNNRE	NNRE	LinearRE
RMSE	2.49	2.97	4.37
RMSE/SD Ratio	0.57	0.68	1.0
RMSE/IQR Ratio	0.42	0.50	0.73
Median	SD	Q1	Q3
7	4.37	5	11

Table 3: Model Performance Comparison for Total Fear Score Prediction. The table presents RMSE values and the RMSE / standard deviation ratio as well as the RMSE / interquartile range ratios) for our three random effects models, alongside summary statistics of the target variable.

Appendix figures & tables23 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 5: Participants responded to the post-climb questionnaire as they watched the video recording of the climb that they just completed to report their i) anxiety level, ii) fear of falling attributable to fatigue, iii) fear of heights, and iv) fatigue in their forearms

Figure 6: The Q-Q diagram shows how well the residuals of the linear mixed model correspond to a normal distribution. Points that are close to the reference line indicate that the residuals follow a normal distribution. We see that the residuals are mostly aligned with the reference line, which proves that the normality assumption is reasonably fulfilled.

Figure 7: Residuals plotted against fitted values of the regression model. The random dispersion of residuals around zero indicates homoscedasticity and proper model specification.

Figure 8: MLP training and validation loss.

Figure 9: Random effect MLP training and validation loss.

Figure 10: GRU training and validation loss.

Figure 11: Random effect GRU training and validation loss.

Emotion Recognition

Emotion recognition uses language, voice, face, body movement, and physiological signals to infer emotional categories or dimensions such as valence and arousal.

See all 129 papers

Estimating Human Muscular Fatigue in Dynamic Collaborative Robotic Tasks with Learning-Based Models
Feb 17, 2026Feras Kiki, Pouya P. Niaz, Alireza Madani +1Human-Robot Collaboration
F. Kiki, P. P. Niaz, A. Madani, and C. Basdogan (corresponding author) are with the Robotics and Mechatronics Laboratory (RML) and the KUIS AI Center, Koc University, Sariyer, Istanbul 34450, Turkey.
Training-Free Hold-Usage Detection in Sport Climbing with Foundation Pose Models
Sep 24, 2026Abu Bakar, Abdullah Aftab, Amir Hamza
Virtual University of Pakistan Islamabad, Pakistan · Army Public School and Colleges Rawalpindi, Pakistan · University of Trento, Italy Trento, Italy
Gait2Hip-60: A Unified Deep Learning Benchmark for Predicting Hip Muscle Forces and Joint Moments from Multi-Cadence Gait Kinematics
May 24, 2026Jiaqi Zhang, Ji Hou, Qing Sun +2Gait Analysis Sequence Modeling
Capital University of Physical Education and Sports. · Beijing Institute of Technology. · 3Beijing Key Laboratory of Interdisciplinary Intelligent Technologies of Sports, Medicine and Engineering.

Feb 17, 2026J/K move · Enter open · S save

Estimating Human Muscular Fatigue in Dynamic Collaborative Robotic Tasks with Learning-Based Models

Feras Kiki, Pouya P. Niaz, Alireza Madani, Cagatay Basdogan

F. Kiki, P. P. Niaz, A. Madani, and C. Basdogan (corresponding author) are with the Robotics and Mechatronics Laboratory (RML) and the KUIS AI Center, Koc University, Sariyer, Istanbul 34450, Turkey.

Human-Robot Collaboration

Open paper AskPDF arXiv

Assessing human muscle fatigue is critical for optimizing performance and safety in physical human-robot interaction(pHRI). This work presents a data-driven framework to estimate fatigue in dynamic, cyclic pHRI using arm-mounted surface electromyography(sEMG). Subject-specific machine-learning regression models(Random Forest, XGBoost, and Linear Regression predict the fraction of cycles to fatigue(FCF) from three frequency-domain and one time-domain EMG features, and are benchmarked against a convolutional neural network(CNN) that ingests spectrograms of filtered EMG. Framing fatigue estimation as regression (rather than classification) captures continuous progression toward fatigue, supporting earlier detection, timely intervention, and adaptive robot control. In experiments with ten participants, a collaborative robot under admittance control guided repetitive lateral (left-right) end-effector motions until muscular fatigue. Average FCF RMSE across participants was 20.8+/-4.3% for the CNN, 23.3+/-3.8% for Random Forest, 24.8+/-4.5% for XGBoost, and 26.9+/-6.1% for Linear Regression. To probe cross-task generalization, one participant additionally performed unseen vertical (up-down) and circular repetitions; models trained only on lateral data were tested directly and largely retained accuracy, indicating robustness to changes in movement direction, arm kinematics, and muscle recruitment, while Linear Regression deteriorated. Overall, the study shows that both feature-based ML and spectrogram-based DL can estimate remaining work capacity during repetitive pHRI, with the CNN delivering the lowest error and the tree-based models close behind. The reported transfer to new motion patterns suggests potential for practical fatigue monitoring without retraining for every task, improving operator protection and enabling fatigue-aware shared autonomy, for safer fatigue-adaptive pHRI control.

Abstract

Emotion Recognition

Explore similar work

Estimating Human Muscular Fatigue in Dynamic Collaborative Robotic Tasks with Learning-Based Models

Training-Free Hold-Usage Detection in Sport Climbing with Foundation Pose Models

Gait2Hip-60: A Unified Deep Learning Benchmark for Predicting Hip Muscle Forces and Joint Moments from Multi-Cadence Gait Kinematics