TheraAgent: Multi-Agent Framework with Self-Evolving Memory and Evidence-Calibrated Reasoning for PET Theranostics

Zhihao Chen; Jiahui Wang; Yizhou Chen; Xiaozhong Ji; Xiaobin Hu; Jimin Hong; Wolfram Andreas Bosbach; Axel Rominger

cs.AIMar 14, 2026

TheraAgent: Multi-Agent Framework with Self-Evolving Memory and Evidence-Calibrated Reasoning for PET Theranostics

Authors: Zhihao Chen, Jiahui Wang, Yizhou Chen, Xiaozhong Ji, Xiaobin Hu, Jimin Hong, Wolfram Andreas Bosbach, Axel Rominger, +3 more

Organizations: Fudan University · University of Bern · Nanjing University · National University of Singapore

Abstract

PET theranostics is transforming precision oncology, yet treatment response varies substantially; many patients receiving 177Lu-PSMA radioligand therapy (RLT) for metastatic castration-resistant prostate cancer (mCRPC) fail to respond, demanding reliable pre-therapy prediction. While LLM-based agents have shown remarkable potential in complex medical diagnosis, their application to PET theranostic outcome prediction remains unexplored, which faces three key challenges: (1) data and knowledge scarcity: RLT was only FDA-approved in 2022, yielding few training cases and insufficient domain knowledge in general LLMs; (2) heterogeneous information integration: robust prediction hinges on structured knowledge extraction from PET/CT, laboratory tests, and free-text clinical documentation; (3) evidence-grounded reasoning: clinical decisions must be anchored in trial evidence rather than LLM hallucinations. In this paper, we present TheraAgent, to our knowledge, the first agentic framework for PET theranostics, with three core innovations: (1) Multi-Expert Feature Extraction with Confidence-Weighted Consensus, where three specialized experts process heterogeneous inputs with uncertainty quantification; (2) Self-Evolving Agentic Memory (SEA-Mem), which learns prognostic patterns from accumulated cases, enabling case-based reasoning from limited data; (3) Evidence-Calibrated Reasoning, integrating a curated theranostics knowledge base to ground predictions in VISION/TheraP trial evidence. Evaluated on 35 real patients and 400 synthetic cases, TheraAgent achieves 75.7% overall accuracy on real patients and 87.0% on synthetic cases, outperforming MDAgents and MedAgent-Pro by over 20%. These results highlight a promising blueprint for trustworthy AI agents in PET theranostics, enabling trial-calibrated, multi-source decision support. Code will be released upon acceptance.

Figures & tables

Figure 1: Overview architecture of TheraAgent

Single LLM
Method	PSA Response		OS $>$ 12m		Overall
Method	Acc (%)	F1 (%)	Acc (%)	F1 (%)	Acc (%)	F1 (%)
GPT-4o	48.6	30.8	48.6	43.8	48.6	37.3
+ CoT	57.1	68.1	48.6	40.0	52.9	54.0
+ Self-Consistency	40.0	36.4	54.3	42.9	47.1	39.6
+ RAG	65.7	66.7	65.7	70.0	65.7	68.3

Table 1: Performance comparison on real clinical data (n=35 patients). TheraAgent significantly outperforms all baselines.

Method	Clear	Ambiguous	Misleading	Overall
	(n=200)	(n=120)	(n=80)
GPT-4o	79.5	63.3	40.0	66.8
GPT-4o w. RAG	89.5	70.0	55.0	76.8
MDAgents	86.0	66.7	51.3	73.3
MedAgent-Pro	83.0	68.3	47.5	71.5
TheraAgent	95.0	79.2	78.8	87.0

Table 2: Accuracy (%) on synthetic data by case complexity.

Configuration	PSA	OS	Overall
TheraAgent (full)	77.1	74.3	75.7
w/o Multi-Expert	62.9	68.6	65.7
w/o SEA-Mem	68.6	71.4	70.0
w/o Evidence-Calibrated	71.4	74.3	72.9
GPT-4o (Single LLM)	48.6	48.6	48.6

Table 3: Ablation study on real data. All values are accuracy (%).

Explore similar work

Jun 23, 2026

RADIANT-PET: Reasoning-Augmented PET/CT Lesion Segmentation with Large Language Models and Reinforcement Learning

Accurate lesion segmentation in PET/CT is critical for oncology, yet remains challenging because physiologic tracer uptake and artifacts can mimic malignant signal. We present RADIANT-PET, a reasoning-augmented framework that couples a high-sensitivity voxel-level segmentation model with lesion-level large language model (LLM) adjudication. Candidate uptake regions are generated with a deliberately permissive segmentation stage, then converted into structured textual descriptions that summarize uptake intensity, morphology, and regional and global anatomical context. An LLM classifies each candidate as true lesion vs. false positive, optionally leveraging the radiology report as additional clinical context. To strengthen lesion-level reasoning, we further optimize a local LLM via reinforcement learning using Group Relative Policy Optimization, rewarding correct lesion classification and anatomically concordant site assignment. Across AutoPET and an OSU test cohort, RADIANT-PET consistently outperforms strong image-only baselines, with the largest improvements observed when radiology reports are provided. Overall, these results demonstrate that LLM-based lesion-level reasoning adds a novel reasoning layer beyond conventional segmentation, suppressing physiologic false positives and aligning voxel-level predictions with clinical interpretation. The project repository is available at: https://github.com/jwang-580/RADIANT-PET.

Jiasheng Wang, Tanun Jitwatcharakomol, Piyawadee Jongpradubgiat +1

Division of Hematology, The Ohio State University Comprehensive Cancer Center, Columbus, OH, USA · Department of Radiology, Mahidol University, Bangkok, Thailand · Department of Radiology, Mettapracharak Hospital, Thailand +1

PET/CT Imaging Medical Image Segmentation Medical LLMs PET/CT Tumor Segmentation

Ask PDF

Jul 13, 2026

Agentic systems for breast cancer treatment recommendations

Large language models (LLMs) are increasingly being explored for clinical decision support, but their reliability in complex oncology treatment planning remains unclear. We evaluated agentic LLM systems for breast cancer treatment recommendation generation using 72 real clinical cases across stages I to IV and 1,147 case-specific rubrics generated through Asymmetric Information Rubric Generation (AIRG), in which the rubric generator had access to real clinical decisions unavailable to the evaluated models. Seven pipelines were compared, including single-LLM baselines, tool-augmented systems, and multi-agent architectures with fact checking and autonomous subagent spawning. The best-performing configuration, Claude Opus 4.8 with the D&C+SA pipeline, achieved a global score of 0.594

\pm

0.025. Tool use and increased agent autonomy had mixed effects, improving performance in some settings but degrading it in others. Performance varied by clinical domain and disease stage, and oncologist-led error analysis revealed persistent clinically relevant failures, including incorrect or missing recommendations, flawed justifications, citation errors, outdated claims, and overconfidence. These findings suggest that agentic LLM systems can generate clinically relevant breast cancer recommendations, but remain insufficient for unsupervised clinical use.

Vinicius Anjos de Almeida, Nícolas Henrique Borges, Leonardo Vicenzi +5

Faculdade de Medicina, Universidade de S˜ao Paulo, Av. Dr. Arnaldo, 455, 01246-903, S˜ao Paulo, S˜ao Paulo, Brazil · Spesia · Laboratory of Artificial Intelligence Applied to Bioinformatics, SEPT, Universidade Federal do Paran´a (UFPR), Rua Alcides Vieira Arcoverde, 1225, 81520-260, Curitiba, Paran´a, Brazil +2

Multi-Agent LLM Systems Clinical Decision Support Oncology Clinical Language Model Evaluation LLM Agent Reliability

Ask PDF

Aug 22, 2026

BioMed-Agent-RL: A Meta Learning, All You Need for Biomedical Applications

The current progress of Clinical Vision Large Language Models (C-VLLMs) has substantially improved digital diagnostics, still these frameworks often endure lesion noises, modality misalignment, hallucination, and missed contextual grounding in complex clinical cases. Moreover, prevailing agent systems usually depend on static and non-adaptable pipelines and lack the versatility necessary for complex medical reasoning. To resolve these difficulties, we present BioMed-Agent-RL, a unified medical agent that incorporates adaptive orchestration, policy, and reward-based reinforcement learning (RL) models for biomedical applications. To ensure reliability, it invokes clinical context-aware preference optimization (CPO), direct preference optimization (DPO), and group relative policy optimization (GRPO) with dynamic entropy regulation. This pipeline utilizes a multimodal meta-learning approach that operates as a field-specific expert and human judgment synthesizer. The agent adaptively utilizes a set of model-level expertise, such as clinical grounding and reasoner, lesion segmenter, and field-specific synthesizer, across various clinical modalities (e.g., X-ray) by utilizing an iterative and adaptive RL approach. The agent learns to seriously synthesize misleading, conflicting vision cues and trust in inherent reasoning, while specialist advice is faulty. An intensive ablation study is conducted across multiple benchmarks, and the agent significantly outperforms existing state of the art models, such as GPT-5, attaining up to ~73% accuracy (gain of ~5%) over contemporary baselines. As a result, the framework suggests a new standard for building factual, reliable, robust, and expert-like intelligent agent systems for independent clinical reasoning.

Md Asaduzzaman Jabin, Zihao Wu, Tianming Liu

The University of Georgia, Athens, Georgia, GA 30605 USA

LLM Agents Multimodal Agents Medical VLMs RL for LLM Agents

Ask PDF