Research Papers

Cerebral Palsy (CP) is a prevalent motor disability in children, for which early detection can significantly improve treatment outcomes. While skeleton-based Graph Convolutional Network (GCN) models have shown promise in automatically predicting CP risk from infant videos, their "black-box" nature raises concerns about clinical explainability. To address this, we introduce a perturbation framework tailored for infant movement features and use it to compare two explainable AI (XAI) methods: Class Activation Mapping (CAM) and Gradient-weighted Class Activation Mapping (Grad-CAM). First, we identify significant and non-significant body keypoints in very low- and very high-risk infant video snippets based on the XAI attribution scores. We then conduct targeted velocity and angular perturbations, both individually and in combination, on these keypoints to assess how the GCN model's risk predictions change. Our results indicate that velocity-driven features of the arms, hips, and legs have a dominant influence on CP risk predictions, while angular perturbations have a more modest impact. Furthermore, CAM and Grad-CAM show partial convergence in their explanations for both low- and high-risk CP groups. Our findings demonstrate the use of XAI-driven movement analysis for early CP prediction and offer insights into potential movement-based biomarker discovery that warrant further clinical validation.

Deep learning-based medical image analysis faces a significant barrier due to the lack of interpretability. Conventional explainable AI (XAI) techniques, such as Grad-CAM and SHAP, often highlight regions outside clinical interests. To address this issue, we propose Segmentation-based Explanation (SegX), a plug-and-play approach that enhances interpretability by aligning the model's explanation map with clinically relevant areas leveraging the power of segmentation models. Furthermore, we introduce Segmentation-based Uncertainty Assessment (SegU), a method to quantify the uncertainty of the prediction model by measuring the 'distance' between interpretation maps and clinically significant regions. Our experiments on dermoscopic and chest X-ray datasets show that SegX improves interpretability consistently across mortalities, and the certainty score provided by SegU reliably reflects the correctness of the model's predictions. Our approach offers a model-agnostic enhancement to medical image diagnosis towards reliable and interpretable AI in clinical decision-making.

The growing interest in eXplainable Artificial Intelligence (XAI) has stimulated research on models with built-in interpretability, among which part-prototype models are particularly prominent. Part-Prototype Models (PPMs) classify inputs by comparing them to learned prototypes and provide human-understandable explanations of the form "this looks like that". Despite this intrinsic interpretability, PPMs have not yet emerged as a competitive alternative to post-hoc explanation methods. This survey reviews work published between 2019 and 2025 and derives a taxonomy of the challenges faced by current PPMs. The analysis reveals a diverse set of open problems. The main issue concerns the quality and number of learned prototypes. Further challenges include limited generalization across tasks and contexts, as well as methodological shortcomings such as non-standardized evaluation. Five broad research directions are identified: improving predictive performance, developing theoretically grounded architectures, establishing frameworks for human-AI collaboration, aligning models with human concepts, and defining robust metrics and benchmarks for evaluation. The survey aims to stimulate further research and promote intrinsically interpretable models for practical applications. A curated list of the surveyed papers is available at https://github.com/aix-group/ppm-survey.

Multimodal Large Language Models (MLLMs) represent the cutting edge of AI technology, with DeepSeek models emerging as a leading open-source alternative offering competitive performance to closed-source systems. While these models demonstrate remarkable capabilities, their vision-language integration mechanisms introduce specific vulnerabilities. We implement an adapted embedding manipulation attack on DeepSeek Janus that induces targeted visual hallucinations through systematic optimization of image embeddings. Through extensive experimentation across COCO, DALL-E 3, and SVIT datasets, we achieve hallucination rates of up to 98.0% while maintaining high visual fidelity (SSIM > 0.88) of the manipulated images on open-ended questions. Our analysis demonstrates that both 1B and 7B variants of DeepSeek Janus are susceptible to these attacks, with closed-form evaluation showing consistently higher hallucination rates compared to open-ended questioning. We introduce a novel multi-prompt hallucination detection framework using LLaMA-3.1 8B Instruct for robust evaluation. The implications of these findings are particularly concerning given DeepSeek's open-source nature and widespread deployment potential. This research emphasizes the critical need for embedding-level security measures in MLLM deployment pipelines and contributes to the broader discussion of responsible AI implementation.

The burgeoning growth of open-sourced vision-language models (VLMs) has catalyzed a plethora of applications across diverse domains. Ensuring the transparency and interpretability of these models is critical for fostering trustworthy and responsible AI systems. In this study, our objective is to delve into the internals of VLMs to interpret the functions of individual neurons. We observe the activations of neurons with respects to the input visual tokens and text tokens, and reveal some interesting findings. Particularly, we found that there are neurons responsible for only visual or text information, or both, respectively, which we refer to them as visual neurons, text neurons, and multi-modal neurons, respectively. We build a framework that automates the explanation of neurons with the assistant of GPT-4o. Meanwhile, for visual neurons, we propose an activation simulator to assess the reliability of the explanations for visual neurons. System statistical analyses on top of one representative VLM of LLaVA, uncover the behaviors/characteristics of different categories of neurons.

RECOVER: Designing a Large Language Model-based Remote Patient Monitoring System for Postoperative Gastrointestinal Cancer Care

Ziqi Yang, Yuxuan Lu, Jennifer Bagdasarian +9

Cancer surgery is a key treatment for gastrointestinal (GI) cancers, a group of cancers that account for more than 35% of cancer-related deaths worldwide, but postoperative complications are unpredictable and can be life-threatening. In this paper, we investigate how recent advancements in large language models (LLMs) can benefit remote patient monitoring (RPM) systems through clinical integration by designing RECOVER, an LLM-powered RPM system for postoperative GI cancer care. To closely engage stakeholders in the design process, we first conducted seven participatory design sessions with five clinical staff and interviewed five cancer patients to derive six major design strategies for integrating clinical guidelines and information needs into LLM-based RPM systems. We then designed and implemented RECOVER, which features an LLM-powered conversational agent for cancer patients and an interactive dashboard for clinical staff to enable efficient postoperative RPM. Finally, we used RECOVER as a pilot system to assess the implementation of our design strategies with four clinical staff and five patients, providing design implications by identifying crucial design elements, offering insights on responsible AI, and outlining opportunities for future LLM-powered RPM systems.

The automatic detection of disinformation presents a significant challenge in the field of natural language processing. This task addresses a multifaceted societal and communication issue, which needs approaches that extend beyond the identification of general linguistic patterns through data-driven algorithms. In this research work, we hypothesise that text classification methods are not able to capture the nuances of disinformation and they often ground their decision in superfluous features. Hence, we apply a post-hoc explainability method (SHAP, SHapley Additive exPlanations) to identify spurious elements with high impact on the classification models. Our findings show that non-informative elements (e.g., URLs and emoticons) should be removed and named entities (e.g., Rwanda) should be pseudo-anonymized before training to avoid models' bias and increase their generalization capabilities. We evaluate this methodology with internal dataset and external dataset before and after applying extended data preprocessing and named entity replacement. The results show that our proposal enhances on average the performance of a disinformation classification method with external test data in 65.78% without a significant decrease of the internal test performance.

AI ethics narratives have the potential to shape the public accurate understanding of AI technologies and promote communication among different stakeholders. However, AI ethics narratives are largely lacking. Existing limited narratives tend to center on works of science fiction or corporate marketing campaigns of large technology companies. Misuse of "socio-technical imaginary" can blur the line between speculation and reality for the public, undermining the responsibility and regulation of technology development. Therefore, constructing authentic AI ethics narratives is an urgent task. The emergence of generative AI offers new possibilities for building narrative systems. This study is dedicated to data-driven visual storytelling about AI ethics relying on the human-AI collaboration. Based on the five key elements of story models, we proposed a conceptual framework for human-AI collaboration, explored the roles of generative AI and humans in the creation of visual stories. We implemented the conceptual framework in a real AI news case. This research leveraged advanced generative AI technologies to provide a reference for constructing genuine AI ethics narratives. Our goal is to promote active public engagement and discussions through authentic AI ethics narratives, thereby contributing to the development of better AI policies.

Ensuring that AI systems make strategic decisions aligned with the specified preferences in adversarial sequential interactions is a critical challenge for developing trustworthy AI systems, especially when the environment is stochastic and players' incomplete preferences leave some outcomes unranked. We study the problem of synthesizing preference-satisfying strategies in two-player stochastic games on graphs where players have opposite (possibly incomplete) preferences over a set of temporal goals. We represent these goals using linear temporal logic over finite traces (LTLf), which enables modeling the nuances of human preferences where temporal goals need not be mutually exclusive and comparison between some goals may be unspecified. We introduce a solution concept of non-dominated almost-sure winning, which guarantees to achieve a most preferred outcome aligned with specified preferences while maintaining robustness against the adversarial behaviors of the opponent. Our results show that strategy profiles based on this concept are Nash equilibria in the game where players are risk-averse, thus providing a practical framework for evaluating and ensuring stable, preference-aligned outcomes in the game. Using a drone delivery example, we demonstrate that our contributions offer valuable insights not only for synthesizing rational behavior under incomplete preferences but also for designing games that motivate the desired behavior from the players in adversarial conditions.

What is the role of product managers in the responsible use of generative AI (genAI) in products and everyday work -- and what enables or constrains their ability to take action? Past literature has examined the ways in which organizational policies can become decoupled from practices when incentives for responsible action are misaligned or impeded by profit motives. While the role of engineers and professional ethicists in the context of AI has been examined in detail, the role of product managers -- who are frequently portrayed as "gatekeepers" or critical decision-makers in product teams -- remains unclear. In this paper, we examine what organizational conditions promote responsible use of genAI by product managers by drawing on twenty-five interviews and a global survey of over three hundred respondents in product management-related roles. We find that uncertainty around responsible AI and a sense of diffused responsibility constrain ethical action, while leadership commitment and organizational principles enable ethical action -- making some responsible practices up to fourteen times more likely. Further, we find two sets of actions product managers take to "recouple" ethical commitments and practices. The first includes low-resource, individual actions product managers can implement without explicit organizational incentives. The second includes high-resource, collective actions that require organizational incentives. Our research suggests recoupling ethical policies and practices at the level of product teams requires institutional buy-in and higher level leadership commitment. Nevertheless, we show that individual actors are able to exhibit agency through some meaningful, low resource actions, even in the absence of organizational incentives, though this alone is insufficient to operationalize responsible AI at scale.

Few-shot learning (FSL) enables machine learning models to generalize effectively with minimal labeled data, making it crucial for data-scarce domains such as healthcare, robotics, and natural language processing. Despite its potential, FSL faces challenges including sensitivity to initialization, difficulty in adapting to diverse domains, and vulnerability to noisy datasets. To address these issues, this paper introduces Adaptive Few-Shot Learning (AFSL), a framework that integrates advancements in meta-learning, domain alignment, noise resilience, and multi-modal integration. AFSL consists of four key modules: a Dynamic Stability Module for performance consistency, a Contextual Domain Alignment Module for domain adaptation, a Noise-Adaptive Resilience Module for handling noisy data, and a Multi-Modal Fusion Module for integrating diverse modalities. This work also explores strategies such as task-aware data augmentation, semi-supervised learning, and explainable AI techniques to enhance the applicability and robustness of FSL. AFSL provides scalable, reliable, and impactful solutions for real-world, high-stakes domains.

Importance: Emergency department (ED) returns for mental health conditions pose a major healthcare burden, with 24-27% of patients returning within 30 days. Traditional machine learning models for predicting these returns often lack interpretability for clinical use. Objective: To assess whether integrating large language models (LLMs) with machine learning improves predictive accuracy and clinical interpretability of ED mental health return risk models. Methods: This retrospective cohort study analyzed 42,464 ED visits for 27,904 unique mental health patients at an academic medical center in the Deep South from January 2018 to December 2022. Main Outcomes and Measures: Two primary outcomes were evaluated: (1) 30-day ED return prediction accuracy and (2) model interpretability using a novel LLM-enhanced framework integrating SHAP (SHapley Additive exPlanations) values with clinical knowledge. Results: For chief complaint classification, LLaMA 3 (8B) with 10-shot learning outperformed traditional models (accuracy: 0.882, F1-score: 0.86). In SDoH classification, LLM-based models achieved 0.95 accuracy and 0.96 F1-score, with Alcohol, Tobacco, and Substance Abuse performing best (F1: 0.96-0.89), while Exercise and Home Environment showed lower performance (F1: 0.70-0.67). The LLM-based interpretability framework achieved 99% accuracy in translating model predictions into clinically relevant explanations. LLM-extracted features improved XGBoost AUC from 0.74 to 0.76 and AUC-PR from 0.58 to 0.61. Conclusions and Relevance: Integrating LLMs with machine learning models yielded modest but consistent accuracy gains while significantly enhancing interpretability through automated, clinically relevant explanations. This approach provides a framework for translating predictive analytics into actionable clinical insights.