Towards Precision Therapy in Hepatocellular Carcinoma: A Clinical-Reasoning LLM for Risk Stratification and Treatment Guidance

# Towards Precision Therapy in Hepatocellular Carcinoma: A Clinical-Reasoning LLM for Risk Stratification and Treatment Guidance
Source: [https://arxiv.org/abs/2607.08602](https://arxiv.org/abs/2607.08602)
## Computer Science \> Artificial Intelligence

**arXiv:2607\.08602**\(cs\)

Authors:[Peng Cui](https://arxiv.org/search/cs?searchtype=author&query=Cui,+P),[Jitao Wang](https://arxiv.org/search/cs?searchtype=author&query=Wang,+J),[Siyan Xue](https://arxiv.org/search/cs?searchtype=author&query=Xue,+S),[Yao Huang](https://arxiv.org/search/cs?searchtype=author&query=Huang,+Y),[Haoming Xia](https://arxiv.org/search/cs?searchtype=author&query=Xia,+H),[Dong Li](https://arxiv.org/search/cs?searchtype=author&query=Li,+D),[Dengxiang Liu](https://arxiv.org/search/cs?searchtype=author&query=Liu,+D),[Weilin Wang](https://arxiv.org/search/cs?searchtype=author&query=Wang,+W),[Liping Liu](https://arxiv.org/search/cs?searchtype=author&query=Liu,+L),[Leida Zhang](https://arxiv.org/search/cs?searchtype=author&query=Zhang,+L),[Yunfu Cui](https://arxiv.org/search/cs?searchtype=author&query=Cui,+Y),[Tao Peng](https://arxiv.org/search/cs?searchtype=author&query=Peng,+T),[Daolin Ji](https://arxiv.org/search/cs?searchtype=author&query=Ji,+D),[Haitao Zhao](https://arxiv.org/search/cs?searchtype=author&query=Zhao,+H),[Wei Zhang](https://arxiv.org/search/cs?searchtype=author&query=Zhang,+W),[Xiaojuan Wang](https://arxiv.org/search/cs?searchtype=author&query=Wang,+X),[Weijie Ma](https://arxiv.org/search/cs?searchtype=author&query=Ma,+W),[Zongren Ding](https://arxiv.org/search/cs?searchtype=author&query=Ding,+Z),[Jinlong Li](https://arxiv.org/search/cs?searchtype=author&query=Li,+J),[Yuan Ding](https://arxiv.org/search/cs?searchtype=author&query=Ding,+Y),[Jiajing Zhao](https://arxiv.org/search/cs?searchtype=author&query=Zhao,+J),[Zhiyu Chen](https://arxiv.org/search/cs?searchtype=author&query=Chen,+Z),[Chengkun Yang](https://arxiv.org/search/cs?searchtype=author&query=Yang,+C),[Ziyue Huang](https://arxiv.org/search/cs?searchtype=author&query=Huang,+Z),[Jiaqi Liu](https://arxiv.org/search/cs?searchtype=author&query=Liu,+J),[Fusheng Liu](https://arxiv.org/search/cs?searchtype=author&query=Liu,+F),[Yang Zhou](https://arxiv.org/search/cs?searchtype=author&query=Zhou,+Y),[Xiaojuan Wang](https://arxiv.org/search/cs?searchtype=author&query=Wang,+X),[Zhongquan Sun](https://arxiv.org/search/cs?searchtype=author&query=Sun,+Z),[Shiyun Bao](https://arxiv.org/search/cs?searchtype=author&query=Bao,+S),[Xiaojun Wang](https://arxiv.org/search/cs?searchtype=author&query=Wang,+X),[Ming Yang](https://arxiv.org/search/cs?searchtype=author&query=Yang,+M),[Guangxin Li](https://arxiv.org/search/cs?searchtype=author&query=Li,+G),[Bin Shu](https://arxiv.org/search/cs?searchtype=author&query=Shu,+B),[Yong Liao](https://arxiv.org/search/cs?searchtype=author&query=Liao,+Y),[Hongxuan Li](https://arxiv.org/search/cs?searchtype=author&query=Li,+H),[Yao Tang](https://arxiv.org/search/cs?searchtype=author&query=Tang,+Y),[Shizhong Yang](https://arxiv.org/search/cs?searchtype=author&query=Yang,+S),[Yongyi Zeng](https://arxiv.org/search/cs?searchtype=author&query=Zeng,+Y),[Yufeng Yuan](https://arxiv.org/search/cs?searchtype=author&query=Yuan,+Y),[Yinpeng Dong](https://arxiv.org/search/cs?searchtype=author&query=Dong,+Y),[Jihui Hao](https://arxiv.org/search/cs?searchtype=author&query=Hao,+J),[Jun Zhu](https://arxiv.org/search/cs?searchtype=author&query=Zhu,+J),[Jiahong Dong](https://arxiv.org/search/cs?searchtype=author&query=Dong,+J)

[View PDF](https://arxiv.org/pdf/2607.08602)

> Abstract:Hepatocellular carcinoma \(HCC\) is a common malignancy and a leading cause of cancer\-related mortality\. Current guidelines and staging systems provide coarse categories, but often miss within\-stage heterogeneity and the clinical context in electronic medical records \(EMRs\)\. We present HCC\-STAR \(Hepatocellular Carcinoma Staging, Treatment And pRognosis\), a clinically aligned large language model that reads routine EMR narratives and jointly outputs risk score\-based staging, ranked guideline\-consistent treatments with evidence\-based rationales, and individualized survival estimates\. We curated about 30,000 HCC cases from SEER and expanded them into EMR\-style narrative training data using a clinician\-validated, prompt\-based augmentation workflow\. On this corpus, we developed a knowledge\-aligned reasoning framework optimized with a step\-verifiable composite reward, moving beyond text\-level memorization of clinical guidelines\. In a multi\-center cohort of 6,668 patients from 12 hospitals in China, HCC\-STAR achieved state\-of\-the\-art performance in treatment recommendation and risk stratification compared with clinical guidelines and competitive models, including GPT\-5 and Gemini\-2\.5 Pro\. Hypothetical overall\-survival analysis showed a median survival of 51 months under adherence to HCC\-STAR recommendations, compared with 29 and 32 months under BCLC and CNLC\. In clinician\-centric evaluations, blinded hepatobiliary specialists rated HCC\-STAR's reasoning and evidence\-based justifications as trustworthy\. The model surpassed resident and attending physicians in treatment accuracy and helped physicians make more accurate decisions faster when used as an assistant\. These findings support HCC\-STAR as a reliable and verifiable decision\-support system for risk stratification and precision therapy in HCC\.

## Submission history

From: Peng Cui \[[view email](https://arxiv.org/show-email/dccd4cb5/2607.08602)\] **\[v1\]**Thu, 9 Jul 2026 15:33:08 UTC \(17,722 KB\)

Bibliographic Tools

## Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Code, Data, Media

## Code, Data and Media Associated with this Article

Demos

## Demos

Related Papers

## Recommenders and Search Tools

About arXivLabs

## arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website\.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy\. arXiv is committed to these values and only works with partners that adhere to them\.

Have an idea for a project that will add value for arXiv's community?[**Learn more about arXivLabs**](https://info.arxiv.org/labs/index.html)\.