A Multi-Agent Framework for Interpreting Multivariate Physiological Time Series

Davide Gabrielli; Paola Velardi; Stefano Faralli; Bardh Prenkaj

cs.LGMar 4, 2026

A Multi-Agent Framework for Interpreting Multivariate Physiological Time Series

Authors: Davide Gabrielli, Paola Velardi, Stefano Faralli, Bardh Prenkaj

Organizations: Sapienza University of Rome Rome, Italy · ISTC-CNR & Sapienza University of Rome Rome, Italy · Technical University of Munich Munich, Germany

Abstract

Continuous physiological monitoring is central to emergency care, yet deploying trustworthy AI is challenging. While LLMs can translate complex physiological signals into clinical narratives, it is unclear how agentic systems perform relative to zero-shot inference. To address these questions, we present Vivaldi, a role-structured multi-agent system that explains multivariate physiological time series. Due to regulatory constraints that preclude live deployment, we instantiate Vivaldi in a controlled, clinical pilot to a small, highly qualified cohort of emergency medicine experts, whose evaluations reveal a context-dependent picture that contrasts with prevailing assumptions that agentic reasoning uniformly improves performance. Our experiments show that agentic pipelines substantially benefit non-thinking and medically fine-tuned models, improving expert-rated explanation justification and relevance by +6.9 and +9.7 points, respectively. Contrarily, for thinking models, agentic orchestration often degrades explanation quality, including a 14-point drop in relevance, while improving diagnostic precision (ESI F1 +3.6). We also find that explicit tool-based computation is decisive for codifiable clinical metrics, whereas subjective targets, such as pain scores and length of stay, show limited or inconsistent changes. Expert evaluation further indicates that gains in clinical utility depend on visualization conventions, with medically specialized models achieving the most favorable trade-offs between utility and clarity. Together, these findings show that the value of agentic AI lies in the selective externalization of computation and structure rather than in maximal reasoning complexity, and highlight concrete design trade-offs and learned lessons, broadly applicable to explainable AI in safety-critical healthcare settings.

Figures & tables

Figure 1. System architecture as a five-scene Emergency Department narrative. The area on the right represents the Shared Memory Buffer (SMB) that Vivaldi manages, orchestrating communication with it and the agents, or between agents, as an intermediary: e.g., think of this as API calls where read (dashed lines) and write (dotted lines) operations happen in the SMB. The circles at the edge of the shaded area on the right illustrate the agents’ entry point for communicating with Vivaldi. We remove the details of read/write interactions between the SMB and Vivaldi for illustration purposes, and defer them to the single scenes. Same-colored lines represent a single logical flow whose completion might depend on multiple interactions between Vivaldi and the agents therein (e.g., 5.b, 8, 9, and 10).

Metric	Thinking		Non-Thinking
	Zero-shot	Agentic	Zero-shot	Agentic
Factuality	88.6 $\pm$ 21.4	83.3 $\pm$ 28.9	87.0 $\pm$ 22.3	91.0 $\pm$ 19.4
Justification	93.2 $\pm$ 17.6	83.3 $\pm$ 28.9	88.9 $\pm$ 21.2	96.2 $\pm$ 13.5
Relevance	95.5 $\pm$ 14.7	81.0 $\pm$ 29.5	87.0 $\pm$ 22.3	97.4 $\pm$ 11.2
Trust	90.9 $\pm$ 25.1	83.3 $\pm$ 28.9	90.7 $\pm$ 19.8	93.6 $\pm$ 16.9

Table 1. Aggregated expert explanation quality (0–100 scale; Yes=100, Partial=50, No=0). Green shading marks the best strategy per category; bold denotes the global best. Thinking models: Gemini 3 Pro, Claude 4.5 Opus. Non-Thinking: GPT 5.2, Llama 4 Maverick, MedGemma 27B.

	Factuality			Justification			Relevance			Trust
	Zero-shot	Agentic	$\mathbf{\Delta}$	Zero-shot	Agentic	$\mathbf{\Delta}$	Zero-shot	Agentic	$\mathbf{\Delta}$	Zero-shot	Agentic	$\mathbf{\Delta}$
Gemini 3 Pro	88.9	81.8	-7.1%	94.4	81.8	-12.6%	94.4	77.3	-17.2%	88.9	81.8	-7.1%
Claude 4.5 Opus	88.5	85.0	-3.5%	92.3	85.0	-7.3%	96.2	85.0	-11.2%	92.3	85.0	-7.3%
GPT 5.2	85.7	84.6	-1.1%	92.9	96.2	+3.3%	92.9	100.0	+7.1%	85.7	88.5	+2.7%
Llama 4 Maverick	90.0	100.0	+10.0%	90.0	96.2	+6.2%	85.0	96.2	+11.2%	90.0	92.3	+2.3%
MedGemma	85.0	88.5	+3.5%	85.0	96.2	+11.2%	85.0	96.2	+11.2%	95.0	100.0	+5.0%

Table 2. Explanation quality comparison for Thinking models (top) versus Non-Thinking models (bottom) on a normalized 0–100 scale.

\mathbf{\Delta}

represents the difference in percentage points between Agentic and Zero-shot performance.

Clinical Metric	Thinking		Non-Thinking
	Zero-shot	Agentic	Zero-shot	Agentic
ESI (F1 $\uparrow$ )	61.0	64.6	40.7	65.4
Pain Score (MAE $\downarrow$ )	1.9	1.8	2.4	2.8
LOS (MAE $\downarrow$ )	1.5	1.9	1.6	2.0
qSOFA (F1 $\uparrow$ )	63.6	100.0	48.1	100.0
SI (MAE $\downarrow$ )	0.1	0.0	0.1	0.0

Table 3. Average model performance on triage-related clinical metric estimation tasks under zero-shot and agentic settings. Green-shaded cells indicate improved performance within the model type; bold indicates the best across the board per metric. (ESI=Emergency Severity Index; LOS=Length of Stay; MAP=Mean Arterial Pressure; SI=Shock Index).

Figure 2. ESI Level Confusion Matrices. Zero-shot models (left) show a critical failure to detect Level 1 (highest acuity) emergencies, misclassifying all such cases.

Figure 3. Performance shifts in Chart Comprehensibility (left) and Clinical Utility (right) across models. Each arrow represents the transition from zero-shot (open circle) to the agentic workflow (solid circle). Green arrows indicate performance gains; red, degradations.

Figure 4. Time efficiency analysis of Zero-Shot versus Agentic pipelines. Stacked bars illustrate the temporal distribution of agents, each annotated with its share of total execution time. Gray bars represent baseline Zero-Shot latency.

Figure 5. Prompt and Completion token usage by model.

Explainable Artificial Intelligence

Explainable artificial intelligence develops ways to make AI decisions understandable, testable, and useful to people.

See all 326 papers

Agentic AI in medicine: architectures, applications, evaluation, and challenges for clinical translation
Jul 28, 2026Zheng Tong, Yang Liu, Wanshu Fan +6Medical AI Agents AI Agent Evaluation
School of Software Engineering, Dalian University, Dalian, China · Department of Surgery, Affiliated Zhongshan Hospital of Dalian University, Dalian, China · The Chinese University of Hong Kong, Shenzhen, China +3
CARE: Towards Clinical Accountability in Multi-Modal Medical Reasoning with an Evidence-Grounded Agentic Framework
Mar 2, 2026Yuexi Du, Jinglu Wang, Shujie Liu +2Medical VLMs Medical Reasoning
Microsoft Research Asia · Department of Biomedical Engineering, Yale University · Department of Radiology & Biomedical Imaging, Yale University
HeartAgent: An Autonomous Agent System for Explainable Differential Diagnosis in Cardiology
Mar 11, 2026Shuang Zhou, Kai Yu, Song Wang +11Clinical Decision Support Differential Diagnosis
Division of Computational Health Sciences, Department of Surgery, University of Minnesota, Minneapolis, MN, USA · Department of Computer Science, University of Central Florida, Orlando, FL, USA · College of Science and Engineering, University of Minnesota, Minneapolis, MN, USA +6

Jul 28, 2026J/K move · Enter open · S save

Agentic AI in medicine: architectures, applications, evaluation, and challenges for clinical translation

Zheng Tong, Yang Liu, Wanshu Fan, Jing Qin+5

School of Software Engineering, Dalian University, Dalian, China · Department of Surgery, Affiliated Zhongshan Hospital of Dalian University, Dalian, China · The Chinese University of Hong Kong, Shenzhen, China+3

Medical AI Agents AI Agent Evaluation Agentic Workflows Multimodal Agents Multi-Agent LLMs Agentic AI

Open paper AskPDF arXiv

Large language models and multimodal foundation models are enabling medical artificial intelligence (AI) systems to move beyond isolated prediction and undertake multistep clinical tasks that require planning, tool use, memory, iterative correction, and coordination among specialized agents. However, the scope of agentic AI in medicine remains unsettled, and current evaluation practices are not yet aligned with the requirements of clinical use. We conducted a scoping review with systematic evidence mapping across five electronic sources, screened 1,649 exportable records, and provisionally included 557 unique studies that met predefined criteria for goal-directed task execution, tool use, interaction with external resources, feedback-based refinement, or multi-agent collaboration. The included studies describe single agents that use external tools, workflows supported by retrieval and external knowledge, multimodal agents, and multi-agent systems applied to medical question answering, image interpretation, electronic health record analysis, drug safety, and clinical trial prediction. The evidence base remains dominated by public benchmarks, simulated settings, retrospective datasets, and small-scale expert evaluation. Process reliability, evidence traceability, uncertainty, safety, workflow impact, and external validity are evaluated less consistently. Clinical translation will depend on clearer definitions, reproducible evaluation, auditable oversight, interoperable system design, and prospective validation in real-world clinical workflows.

Abstract

Explainable Artificial Intelligence

Explore similar work

Agentic AI in medicine: architectures, applications, evaluation, and challenges for clinical translation

CARE: Towards Clinical Accountability in Multi-Modal Medical Reasoning with an Evidence-Grounded Agentic Framework

HeartAgent: An Autonomous Agent System for Explainable Differential Diagnosis in Cardiology