Personalized nutrition intervention for patients with multimorbidity is critical for improving health outcomes, yet remains challenging because it requires the simultaneous integration of heterogeneous clinical conditions, medications, and dietary guidelines. Single-agent large language models (LLMs) often suffer from context overload and attention dilution when processing such high-dimensional patient profiles. We introduce NutriOrion, a hierarchical multi-agent framework with a parallel-then-sequential reasoning topology. NutriOrion decomposes nutrition planning into specialized domain agents with isolated contexts to mitigate anchoring bias, followed by a conditional refinement stage. The framework includes a multi-objective prioritization algorithm to resolve conflicting dietary requirements and a safety constraint mechanism that injects pharmacological contraindications as hard negative constraints during synthesis, ensuring clinical validity by construction rather than post-hoc filtering. For clinical interoperability, NutriOrion maps synthesized insights into the ADIME standard and FHIR R4 resources. Evaluated on 330 stroke patients with multimorbidity, NutriOrion outperforms multiple baselines, including GPT-4.1 and alternative multi-agent architectures. It achieves a 12.1 percent drug-food interaction violation rate, demonstrates strong personalization with negative correlations (-0.26 to -0.35) between patient biomarkers and recommended risk nutrients, and yields clinically meaningful dietary improvements, including a 167 percent increase in fiber and a 27 percent increase in potassium, alongside reductions in sodium (9 percent) and sugars (12 percent).
Figures & tables
Figure 1. Conceptual comparison of reasoning paradigms for high-stakes decision making. (a) Monolithic LLM Approach: Standard single-agent models process heterogeneous data in a unified context window. This often leads to information entanglement, where conflicting constraints are hallucinated or ignored due to attention dilution. (b) NutriOrion (Ours): A hierarchical modular framework that decouples reasoning into specialized streams. By isolating domain contexts and enforcing deterministic safety gates, NutriOrion synthesizes complex inputs into precise, verifiable, and structured interventions without reasoning collapse.
Figure 2. The NutriOrion System Pipeline. The framework integrates three modular components: (Left) Retrieval-Augmented Grounding ( R ): Ingestion of patient profiles and vectorization of clinical guidelines (e.g., DASH, ADA); (Center) Hierarchical Agent Reasoning ( H ): A three-stage orchestration consisting of parallel diagnostic analysis (Stage 1), sequential dietary refinement (Stage 2), and safety-aware synthesis (Stage 3); (Right) Output Standardization ( O ): Projection of synthesized ADIME-structured insights into interoperable FHIR R4 clinical resources.
Method
Type
Logic
# Agents
Core Test
NutriOrion (Ours)
Multi-Agent
Structured ADIME
10
SOTA Performance
+ Specialized Roles
ReAct ( yao2022react )
Single
Dynamic Tool Loop
1
Dynamic Reasoning
CoT ( wei2022chain )
Single
Linear Planning
1
Linear Reasoning
Self-Refine ( madaan2023self )
Single
Iterative Correction
1
Self-Correction
API ReAct
Single
Dynamic Tool Loop
1
Model Capability
Table 1. Summary of baseline methods and ablations. All methods use identical tools and temperature = 0.0.
Model
Diagnosis
Interv.
Logic
Detail
Total
Act.%
Claude-Sonnet-4
2.0
2.0
1.5
1.5
7.0
91.2
GPT-4.1
2.0
1.5
1.5
1.5
6.5
91.2
NutriOrion
2.0
2.0
2.0
1.5
7.5
97.8
React-Qwen-2.5-72B
2.0
1.5
1.5
1.0
6.0
20.8
React-LLaMA-3-70B
1.5
1.0
1.5
0.5
4.5
18.1
Models below produce non-structured outputs:
Table 2. NCP-QUEST Scores & Actionability Rates. NCP-QUEST scores (0–8) averaged from two independent RDs ( κ=0.82 ). Act.% (Actionability) denotes the success rate of generating concrete, identifiable food items (Calculated as 1−Non-specific Rate ).
Model
Eval. ( N )
Violation
Rate
High Actionability Models (Act. >85% in Table 2 )
Claude-Sonnet-4
157
32
11.5%
GPT-4.1
157
35
16.6%
Ablation-NoSpecialist
157
34
16.6%
NutriOrion (Ours)
132
23
12.1%
Moderate Actionability Baselines
Table 3. Drug-Food Interaction Violations. Eval. ( N ) denotes the number of outputs containing actionable food items. Rate is the percentage of outputs triggering FDA contraindications. Rows in gray indicate models with insufficient Actionability ( <40% in Table 2 ). Their near-zero violation rates are artifacts of silence (failing to prescribe food) rather than safety.
Figure 3. Personalization Heatmap. Pearson correlations between patient biomarkers and nutrient intake. Blue (Negative) indicates effective personalization (e.g., High BP → Low Sodium), while Red (Positive) implies unsafe recommendations. NutriOrion consistently achieves the strongest protective correlations across all five metrics, significantly outperforming baselines.
Figure 4. Multi-dimensional Performance (Bubble Chart). X-axis : Actionability; Y-axis : Food Compass Score 2.0 (Nutritional Quality); Bubble Size : Dietary Diversity; Color : NutriScore Health Gradient (Red → Green). NutriOrion (top-right) achieves the optimal balance: high quality, actionable output, and high diversity. Gray labels indicate unstructured baselines, as detailed in Table 2 .
Appendix figures & tables4 assets
Supplementary material from the paper’s appendix.
Appendix
NutriOrion Internal Field
ADIME Stage
FHIR R4 Target Element
hypertension_diet (e.g., DASH)
Assessment
oralDiet.type.coding (e.g., SNOMED: 182922004)
diabetes_diet (e.g., CarbCount)
Assessment
oralDiet.type.coding
nutrient_flags.sodium
Diagnosis
oralDiet.nutrient.modifier (e.g., LOINC)
nutrient_flags.limit_val
Diagnosis
oralDiet.nutrient.amount
intervention.replace
Intervention
oralDiet.excludeFoodModifier
intervention.add
Intervention
oralDiet.foodPreferenceModifier
Appendix
Table 4. Full Mapping Schema: From NutriOrion ADIME Output to FHIR R4 NutritionOrder
Figure 5. RD Clinical Appropriateness Evaluation. Mean scores (1–5 scale) across six dimensions for NutriOrion outputs
Warfarin (n=28)
K-Sparing Diuretics (n=15)
Statins + Stroke (n=114)
Model
Eval
Viol
Rate
Eval
Viol
Rate
Eval
Viol
Rate
Claude-Sonnet-4
28
6
21.4%
15
11
73.3%
114
1
0.9%
NutriOrion
26
9
34.6%
12
7
58.3%
94
0
0.0%
GPT-4.1
28
13
46.4%
15
11
73.3%
114
2
1.8%
Ablation-NoSpec.
28
16
57.1%
15
10
66.7%
114
0
0.0%
Single-Qwen-72B
23
5
21.7%
10
1
10.0%
93
0
0.0%
Appendix
Table 5. Drug-Food Interaction Violations by Clinical Scenario (DailyMed-based). The three scenarios represent high-risk drug-food interactions prevalent in stroke patients. Warfarin users face the highest violation rates across all high-quality models due to the “healthy food trap” where vitamin K-rich vegetables (spinach, broccoli, kale) are commonly recommended.
Figure 6. Average dietary nutrients from original foods and NutriOrion recommendations. Bars show absolute amounts, and markers indicate percent change relative to original intake.
College of Information and Electrical Engineering, China Agricultural University · Key Laboratory of Precision Nutrition and Food Quality, China Agricultural University
Department of Electrical and Computer Engineering, Johns Hopkins University, Baltimore, MD 21218 USA · Geriatric Engineering at Johns Hopkins University, Baltimore, MD 21218 USA · Departments of Neurology and Neurosurgery, Johns Hopkins Medicine, Baltimore, MD 21287 USA