A medical provider's summary of a patient visit serves several critical
purposes, including clinical decision-making, facilitating hand-offs between
providers, and as a reference for the patient. An effective summary is required
to be coherent and accurately capture all the medically relevant information in
the dialogue, despite the complexity of patient-generated language. Even minor
inaccuracies in visit summaries (for example, summarizing "patient does not
have a fever" when a fever is present) can be detrimental to the outcome of
care for the patient.
This paper tackles the problem of medical conversation summarization by
discretizing the task into several smaller dialogue-understanding tasks that
are sequentially built upon. First, we identify medical entities and their
affirmations within the conversation to serve as building blocks. We study
dynamically constructing few-shot prompts for tasks by conditioning on relevant
patient information and use GPT-3 as the backbone for our experiments. We also
develop GPT-derived summarization metrics to measure performance against
reference summaries quantitatively. Both our human evaluation study and metrics
for medical correctness show that summaries generated using this approach are
clinically accurate and outperform the baseline approach of summarizing the
dialog in a zero-shot, single-prompt setting.

本研究使用 GPT-3 模型，将医学对话摘要问题离散成多个对话理解任务，并动态构建少样本提示进行实验，开发了基于 GPT 的度量标准，评估任务性能，经过研究评价证实，此方法生成的摘要临床准确性优于基准方法。