Balancing artificial intelligence (AI) support with appropriate human oversight is challenging, with associated risks such as algorithm aversion and technology dominance. Research areas like eXplainable AI (XAI) and Frictional AI aim to address these challenges. Studies have shown that presenting XAI explanations as “juxtaposed evidence” supporting contrasting classifications, rather than just providing predictions, can be beneficial.This study aimed to design and compare multiple pipelines for generating juxtaposed evidence in the form of class activation maps (CAMs) that highlight areas of interest in a fracture detection task with X-ray images.We designed three pipelines to generate such evidence. The pipelines are based on a fracture detection task from 630 thoraco-lumbar X-ray images (48% of which contained fractures). The first, a single-model approach, uses an algorithm of the Grad-CAM family applied to a ResNeXt-50 network trained through transfer learning. The second, a dual-model approach, employs two networks—one optimized for sensitivity and the other for specificity—providing targeted explanations for positive and negative cases. The third, a generative approach, leverages autoencoders to create activation maps from feature tensors, extracted from the raw images. Each approach produced two versions of activation maps: AM3—as we termed it—which captures fine-grained, low-level features, and AM4, highlighting high-level, aggregated features. We conducted a validation study by comparing the generated maps with binary ground-truth masks derived from a consensus of four clinician annotators, identifying the actual locations of fractures in a subset of positive cases.HiResCAM proved to be the best performing Grad-CAM variant and was used in both the single- and dual-model strategies. The generative approach demonstrated the greatest overlap with the clinicians' assessments, indicating its ability to align with human expertise.The results highlight the potential of Judicial AI to enhance diagnostic decision-making and foster a synergistic collaboration between humans and AI.

Human Activity Recognition (HAR) and exercise assessment models are increasingly used in healthcare to support clinical evaluation, rehabilitation, and remote monitoring. However, their real-world applicability critically depends on the ability to generalize across unseen subjects, whose movement patterns may differ substantially due to inter-individual variability. Despite this, many studies adopt random noncross-subject (NCS) data splits, where samples from the same individual appear in both training and test sets, potentially leading to overly optimistic and clinically misleading performance estimates. We investigate (i) how NCS and cross-subject (CS) splits affect performance estimation across machine learning and deep learning models under tasks of increasing complexity, (ii) how data splitting and differences between training and test sets contribute to predictive variance and stability. Experiments were performed using a large-scale HAR benchmark dataset (NTU RGB+D 120) and a rehabilitation-specific dataset (IntelliRehabDS). A total of 12 machine learning and deep learning models were trained across both tasks, and their performance was estimated and compared using a simulation-based approach. Predictive variance decomposition, via Generalized Linear Mixed-Effects models, was applied to link the split strategy and differences in training and test instances to model output stability. NCS splits consistently overestimated model performances, with discrepancies increasing alongside task and model complexity. DL architectures, in particular, showed markedly higher NCS performance compared to CS splits, generally with statistical significance. Variance decomposition revealed that greater subject difference between training and test sets often enhances predictive instability, while CS splitting reduces variance by promoting more generalizable representations. Improper dataset splits can mislead model evaluation, exaggerate generalization capabilities, and undermine clinical trust. Our study provides empirical evidence for computer vision-based rehabilitation models and offers methodological guidance for robust evaluation practices, supporting reproducible and trustworthy AI deployment in rehabilitation and broader healthcare applications.

In recent years, generative models have shown remarkable capabilities in synthesizing realistic human motion, with applications ranging from animation to virtual reality. However, their potential in clinical and rehabilitation settings remains underexplored. In this work, we introduce a conditional diffusion-based generative framework for rehabilitation-oriented motion synthesis, which directly operates on joint-angle representations of full-body movement. Unlike most existing approaches that rely on joint positions, our method generates motion in a clinically meaningful space that explicitly encodes joint range of motion, aligning the generation process with how motor performance is assessed in rehabilitation practice. This design enables subject-independent modeling while improving the interpretability of the generated movements from a clinical perspective. We propose a comprehensive evaluation protocol by combining qualitative and quantitative metrics, including simulation visualizations, similarity analysis, and automated assessment of simulations adherence to users input. Experiments based on cross-subject and leave-one-combination-out settings demonstrate the model's ability to generate plausible, contextually accurate motion sequences, with improved generalization when using joint angle representations, achieving superior performance compared to a position-based approach. Despite limitations due to dataset size and gesture diversity, results support the feasibility of generating rehabilitation-oriented motion simulations, motivating future investigation in personalized rehabilitation scenarios.