The accurate extraction of medical orders fromdoctor-patient conversations is a critical taskfor reducing clinical documentation burdensand ensuring patient safety. This paper detailsour team’s submission to the MEDIQA-OE-2025Shared Task. We investigate the performanceof MedGemma, a new domain-specific opensource language model, for structured order extraction. We systematically evaluate three distinct prompting paradigms: a straightforwardone-shot approach, a reasoning-focused ReActframework, and a multi-step agentic workflow.Our experiments reveal that while more complex frameworks like ReAct and agentic flowsare powerful, the simpler one-shot promptingmethod achieved the highest performance onthe official validation set. We posit that on manually annotated transcripts, complex reasoningchains can lead to “overthinking” and introduce noise, making a direct approach more robust and efficient. Our work provides valuableinsights into selecting appropriate promptingstrategies for clinical information extraction invaried data conditions.

The rapid expansion of asynchronous remote care has intensified provider workload, creating demand for AI systems that can assist clinicians in managing patient queries more efficiently. The MEDIQA-WV 2025 shared task addresses this challenge by focusing on generating free-text responses to wound care queries paired with images. In this work, we present two complementary approaches developed for the English track. The first leverages a mined prompting strategy, where training data is embedded, and the top-k most similar examples are retrieved to serve as few-shot demonstrations during generation. The second approach builds on a metadata ablation study, which identified four metadata attributes that consistently enhance response quality. We train classifiers to predict these attributes for test cases and incorporate them into the generation pipeline, dynamically adjusting outputs based on prediction confidence. Experimental results demonstrate that mined prompting improves response relevance, while metadata-guided generation further refines clinical precision. Together, these methods highlight promising directions for developing AI-driven tools that can provide reliable and efficient wound care support.

The ComMA@ICON 2021 Shared Task involved identifying the level of aggression and identifying gender bias and communal bias from texts in various languages from the domain of social media. In this paper, we present the description and analyses of systems we implemented towards these tasks. We built systems utilizing Transformer-based models, experimented by individually and jointly modelling these tasks, and investigated the performance of a feature engineering method in conjunction with a joint modelling approach. We demonstrate that the joint modelling approaches outperform the individual modelling approach in most cases.