@inproceedings{feng-etal-2026-mcplanner,
title = "{MCP}lanner: Hierarchical Adaptive Planning-Based Medical Consultation with Large Language Models",
author = "Feng, Qing and
Jia, Zhihao and
Duan, Junwen",
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.1799/",
pages = "38455--38471",
ISBN = "979-8-89176-462-0",
abstract = "Medical consultation (MC) is an interactive decision-making process that requires dynamically adapting inquiry directions according to evolving patient feedback. Although large language models (LLMs) have demonstrated strong planning capabilities in medical applications, existing LLM-based MC methods typically make turn-level independent inquiry decisions without adaptive control over global inquiry directions, making them prone to inefficient local exploration. To address this limitation, we propose MCPlanner, a hierarchical adaptive planning-based MC framework in which a high-level planner adaptively controls the inquiry direction, while a low-level planner executes inquiry actions under the selected direction. Additionally, MCPlanner introduces expert-guided distillation and a smoothed confidence calibration strategy to improve high-level planning efficiency and diagnostic accuracy. Experimental results on DXY, GMD, and CMD show that MCPlanner achieves best diagnostic accuracies of 95.2{\%}, 90.8{\%}, and 76.1{\%}, respectively. Notably, on CMD, MCPlanner outperforms the strongest baseline by 5.0 percentage points. Code is publicly available at \url{https://github.com/zh-jia/MCPlanner}."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="feng-etal-2026-mcplanner">
<titleInfo>
<title>MCPlanner: Hierarchical Adaptive Planning-Based Medical Consultation with Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Qing</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhihao</namePart>
<namePart type="family">Jia</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junwen</namePart>
<namePart type="family">Duan</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>Medical consultation (MC) is an interactive decision-making process that requires dynamically adapting inquiry directions according to evolving patient feedback. Although large language models (LLMs) have demonstrated strong planning capabilities in medical applications, existing LLM-based MC methods typically make turn-level independent inquiry decisions without adaptive control over global inquiry directions, making them prone to inefficient local exploration. To address this limitation, we propose MCPlanner, a hierarchical adaptive planning-based MC framework in which a high-level planner adaptively controls the inquiry direction, while a low-level planner executes inquiry actions under the selected direction. Additionally, MCPlanner introduces expert-guided distillation and a smoothed confidence calibration strategy to improve high-level planning efficiency and diagnostic accuracy. Experimental results on DXY, GMD, and CMD show that MCPlanner achieves best diagnostic accuracies of 95.2%, 90.8%, and 76.1%, respectively. Notably, on CMD, MCPlanner outperforms the strongest baseline by 5.0 percentage points. Code is publicly available at https://github.com/zh-jia/MCPlanner.</abstract>
<identifier type="citekey">feng-etal-2026-mcplanner</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1799/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>38455</start>
<end>38471</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T MCPlanner: Hierarchical Adaptive Planning-Based Medical Consultation with Large Language Models
%A Feng, Qing
%A Jia, Zhihao
%A Duan, Junwen
%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 feng-etal-2026-mcplanner
%X Medical consultation (MC) is an interactive decision-making process that requires dynamically adapting inquiry directions according to evolving patient feedback. Although large language models (LLMs) have demonstrated strong planning capabilities in medical applications, existing LLM-based MC methods typically make turn-level independent inquiry decisions without adaptive control over global inquiry directions, making them prone to inefficient local exploration. To address this limitation, we propose MCPlanner, a hierarchical adaptive planning-based MC framework in which a high-level planner adaptively controls the inquiry direction, while a low-level planner executes inquiry actions under the selected direction. Additionally, MCPlanner introduces expert-guided distillation and a smoothed confidence calibration strategy to improve high-level planning efficiency and diagnostic accuracy. Experimental results on DXY, GMD, and CMD show that MCPlanner achieves best diagnostic accuracies of 95.2%, 90.8%, and 76.1%, respectively. Notably, on CMD, MCPlanner outperforms the strongest baseline by 5.0 percentage points. Code is publicly available at https://github.com/zh-jia/MCPlanner.
%U https://aclanthology.org/2026.findings-emnlp.1799/
%P 38455-38471
Markdown (Informal)
[MCPlanner: Hierarchical Adaptive Planning-Based Medical Consultation with Large Language Models](https://aclanthology.org/2026.findings-emnlp.1799/) (Feng et al., Findings 2026)
ACL