Towards Efficient Medical Reasoning with Minimal Fine-Tuning Data

Supervised Fine-Tuning (SFT) plays a pivotal role in adapting Large Language Models (LLMs) to specialized domains such as medical reasoning. However, existing SFT practices often rely on unfiltered datasets that contain redundant and low-quality samples, leading to substantial computational costs and suboptimal performance. Although existing methods attempt to alleviate this problem by selecting data based on sample difficulty, defined by \textit{knowledge} and \textit{reasoning} complexity, they overlook each sample's optimization utility reflected in its gradient. Interestingly, we find that gradient-based influence alone favors easy-to-optimize samples that cause large parameter shifts but lack deep reasoning chains, while difficulty alone selects noisy or overly complex cases that fail to guide stable optimization. Based on this observation, we propose a data selection strategy, \textit{\textbf{D}ifficulty-\textbf{I}nfluence \textbf{Q}uadrant} \textbf{(DIQ)}, which prioritizes samples in the “high-difficulty–high-influence” quadrant to balance complex clinical reasoning with substantial gradient influence, enabling efficient medical reasoning with minimal fine-tuning data. Furthermore, Human and LLM-as-a-judge evaluations show that DIQ-selected subsets demonstrate higher data quality and generate clinical reasoning that is more aligned with expert practices in \textit{differential diagnosis}, \textit{safety check}, and \textit{evidence citation}, as DIQ emphasizes samples that foster expert-like reasoning patterns. Extensive experiments on medical reasoning benchmarks demonstrate that DIQ enables models fine-tuned on only \textbf{1\%} of selected data to match full-dataset performance, while using \textbf{10\%} consistently outperforms baseline methods, highlighting the superiority of principled data selection over brute-force scaling.