@inproceedings{wenbo-etal-2026-ask,
title = "Ask Better Questions: Learning Efficient Information-Seeking Policy for Clinical Cold-Start Dialogue",
author = "Wenbo, Li and
Cai, Le and
Liu, Jiawei and
Yu, Jinhong and
Chen, Zhuo and
Luo, Zhuoran and
Qingqing, Yang and
Cheng, Gao and
Cheng, Qikai and
Lu, Wei and
Liu, Xiaozhong",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.2387/",
pages = "51298--51328",
ISBN = "979-8-89176-462-0",
abstract = "Despite strong performance in medical question answering, large language models remain limited in clinical cold-start settings, where little patient information is initially available and key evidence must be acquired through dialogue. This requires models to decide what to ask under a limited turn budget, balancing broad history taking with targeted clarification of high-risk cues. To address the aforementioned challenges, we propose **MedCIA**, a two-stage framework for improving clinical information acquisition. First, it reconstructs patient records into structured inquiry trajectories, enabling supervised learning of expert-guided information-seeking intents and the breadth-depth trade-off in clinical questioning. Second, it refines the policy with Adaptive Tree Policy Optimization (ATPO), an uncertainty-aware reinforcement learning procedure that allocates rollout computation to ambiguous dialogue states in a profile-anchored patient simulator. Furthermore, we curate a multi-turn dataset (MedIntake) for active clinical intake with progressive patient information disclosure and diverse inquiry strategies. Experiments on an EHR-derived benchmark show that **MedCIA** improves importance-weighted coverage, early acquisition, and cost control under controlled simulation."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wenbo-etal-2026-ask">
<titleInfo>
<title>Ask Better Questions: Learning Efficient Information-Seeking Policy for Clinical Cold-Start Dialogue</title>
</titleInfo>
<name type="personal">
<namePart type="given">Li</namePart>
<namePart type="family">Wenbo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Le</namePart>
<namePart type="family">Cai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiawei</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jinhong</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhuo</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhuoran</namePart>
<namePart type="family">Luo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yang</namePart>
<namePart type="family">Qingqing</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gao</namePart>
<namePart type="family">Cheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qikai</namePart>
<namePart type="family">Cheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaozhong</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Despite strong performance in medical question answering, large language models remain limited in clinical cold-start settings, where little patient information is initially available and key evidence must be acquired through dialogue. This requires models to decide what to ask under a limited turn budget, balancing broad history taking with targeted clarification of high-risk cues. To address the aforementioned challenges, we propose **MedCIA**, a two-stage framework for improving clinical information acquisition. First, it reconstructs patient records into structured inquiry trajectories, enabling supervised learning of expert-guided information-seeking intents and the breadth-depth trade-off in clinical questioning. Second, it refines the policy with Adaptive Tree Policy Optimization (ATPO), an uncertainty-aware reinforcement learning procedure that allocates rollout computation to ambiguous dialogue states in a profile-anchored patient simulator. Furthermore, we curate a multi-turn dataset (MedIntake) for active clinical intake with progressive patient information disclosure and diverse inquiry strategies. Experiments on an EHR-derived benchmark show that **MedCIA** improves importance-weighted coverage, early acquisition, and cost control under controlled simulation.</abstract>
<identifier type="citekey">wenbo-etal-2026-ask</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2387/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>51298</start>
<end>51328</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Ask Better Questions: Learning Efficient Information-Seeking Policy for Clinical Cold-Start Dialogue
%A Wenbo, Li
%A Cai, Le
%A Liu, Jiawei
%A Yu, Jinhong
%A Chen, Zhuo
%A Luo, Zhuoran
%A Qingqing, Yang
%A Cheng, Gao
%A Cheng, Qikai
%A Lu, Wei
%A Liu, Xiaozhong
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F wenbo-etal-2026-ask
%X Despite strong performance in medical question answering, large language models remain limited in clinical cold-start settings, where little patient information is initially available and key evidence must be acquired through dialogue. This requires models to decide what to ask under a limited turn budget, balancing broad history taking with targeted clarification of high-risk cues. To address the aforementioned challenges, we propose **MedCIA**, a two-stage framework for improving clinical information acquisition. First, it reconstructs patient records into structured inquiry trajectories, enabling supervised learning of expert-guided information-seeking intents and the breadth-depth trade-off in clinical questioning. Second, it refines the policy with Adaptive Tree Policy Optimization (ATPO), an uncertainty-aware reinforcement learning procedure that allocates rollout computation to ambiguous dialogue states in a profile-anchored patient simulator. Furthermore, we curate a multi-turn dataset (MedIntake) for active clinical intake with progressive patient information disclosure and diverse inquiry strategies. Experiments on an EHR-derived benchmark show that **MedCIA** improves importance-weighted coverage, early acquisition, and cost control under controlled simulation.
%U https://aclanthology.org/2026.findings-emnlp.2387/
%P 51298-51328
Markdown (Informal)
[Ask Better Questions: Learning Efficient Information-Seeking Policy for Clinical Cold-Start Dialogue](https://aclanthology.org/2026.findings-emnlp.2387/) (Wenbo et al., Findings 2026)
ACL
- Li Wenbo, Le Cai, Jiawei Liu, Jinhong Yu, Zhuo Chen, Zhuoran Luo, Yang Qingqing, Gao Cheng, Qikai Cheng, Wei Lu, and Xiaozhong Liu. 2026. Ask Better Questions: Learning Efficient Information-Seeking Policy for Clinical Cold-Start Dialogue. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 51298–51328, Budapest, Hungary. Association for Computational Linguistics.