Agentic neural architecture search for clinical prediction from electronic health records: Cross-hospital knowledge transfer and computational efficiency

Deyi Li; Qi Xu; Lingyao Li; Tiansheng Wang; Muxuan Liang; Mei Liu

cs.AIAug 22, 2026

Agentic neural architecture search for clinical prediction from electronic health records: Cross-hospital knowledge transfer and computational efficiency

Authors: Deyi Li, Qi Xu, Lingyao Li, Tiansheng Wang, Muxuan Liang, Mei Liu

Organizations: Department of Health Outcomes and Biomedical Informatics, College of Medicine, University of Florida, Gainesville, FL, USA · College of Information Science, University of Arizona, Tucson, AZ, USA · Department of Pharmaceutical Health Outcomes and Policy, College of Pharmacy, University of Houston, Houston, TX, USA · Department of Epidemiology, Gillings School of Global Public Health, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA ·

Abstract

Transformer architectures for clinical prediction from electronic health records (EHRs) often require costly manual tuning and may vary across hospitals and tasks. We introduce ATHENA (Agentic Transfer across Hospitals for EHR Neural Architecture Search), a knowledge-guided agentic neural architecture search framework for Transformer-based EHR modeling. A weight-sharing supernet is pretrained once per hospital, allowing candidate architectures to inherit weights and be evaluated through fine-tuning rather than independent pretraining. A two-layer cross-hospital prior supplies high-performing source architecture examples and task-specific component preferences summarized by SHapley Additive exPlanations (SHAP)-based meta-regression. These priors guide a multi-agent large language model search using target-hospital validation feedback. Across six clinical prediction tasks at one held-out OneFlorida+ site and one external MIMIC-IV site, ATHENA significantly outperforms all four baselines in 9 of 12 site-task evaluations under a strict equal-compute comparison. In a separate comparison using a common pretrained AutoFormer supernet to isolate search strategy performance, ATHENA ranks first in 9 of 12 evaluations at a maximum search budget of 30 candidate evaluations. It also shows more consistent architecture selection across repeated searches. ATHENA supports more efficient architecture selection for Transformer-based clinical prediction models.

Figures & tables

Fig 1: The ATHENA framework. Stage 1 evaluates sampled architectures across source hospitals and tasks to obtain architecture–performance metadata. The metadata yield a retrieval prior of high-performing source architectures and a SHAP-based architecture-effect prior. Stage 2 combines these priors with target-validation feedback to guide agentic NAS while adaptively balancing exploration and exploitation .

Fig 2: Illustration of the four configurable Transformer architecture parameters in the NAS space.

Aspect	Layer 1 : task-driven retrieval	Layer 2 : architecture-effect prior
Input	Source and target training-set task descriptors; evaluated source architecture–performance records.	Architecture–performance records for the same task across source hospitals.
Goal	Retrieve relevant, high-performing architecture examples.	Estimate cross-hospital performance associations for architectural choices.
Output	Top $K=5$ evaluated source configurations and their metrics.	SHAP-based level preferences, confidence intervals, feature importance, and pairwise rules.
Use	Concrete examples for candidate proposal and refinement.	Directional evidence for candidate proposal and refinement.

Table 1: Complementary roles of the two cross-hospital prior layers.

Agent	Implementation	Main function	Output
Proposal Agent $A_{p}$	LLM	Proposes architectures using task information, cross-hospital priors, search history, and the current strategy; revises proposals based on critique feedback.	Candidate architectures and design rationales.
Critic Agent $A_{c}$	LLM	Reviews proposed designs for constraint compatibility and novelty, flags concerns about prior preferences, and recommends revisions.	Structured critiques and revision recommendations.
Experiment Agent	Deterministic evaluator $A_{e}$ and LLM strategy module $A_{s}$	$A_{e}$ fine-tunes accepted subnetworks and computes target-validation metrics; $A_{s}$ uses updated search feedback to select exploration or exploitation for the next round.	Validation records and the next-round strategy with its rationale.

Table 2: Roles of the three agents in ATHENA.

Fig 3: The Proposal Agent prompt template. The implementation populates each block with task-specific values before invoking the LLM.

Prior source pool
	Pretraining		Fine-tuning (Binary)
Cohort	Patients	Share	Train	Val.	Test
Site A	22,186	42.0%	6,158	12,234	12,253
Site B	29,222	52.4%	5,300	10,566	10,647
Site C	37,152	42.0%	10,134	20,594	20,647
Site D	37,259	63.9%	4,204	8,409	8,410

Table 3: Pretraining-pool sizes and patient-level fine-tuning splits.

Prior source pool
Cohort	Task	Train	Val.	Test
Site A	Phenotype (6M)	11,083	8,147	8,405
	Phenotype (12M)	12,235	9,124	9,100
	Drug Recommendation †	78,999	59,249	59,251
Site B	Phenotype (6M)	9,229	6,892	6,921
	Phenotype (12M)	10,285	7,792	7,643

Table 4: Pretraining-pool sizes and patient-level fine-tuning splits (continued).

Task	$\rho$ (AUROC)	$\rho$ (AUPRC)
Mortality	0.544	0.567
Stay $>$ 7d	0.809	0.806
Readmission (3M)	0.570	0.412
Phenotype (6M)	0.951	0.944
Phenotype (12M)	0.940	0.944
Drug Recommendation	0.944	0.958

Table 5: Supernet ranking fidelity on MIMIC-IV.

$N$	Shared	Scratch	Speedup	Shared	Scratch	Speedup
	OneFlorida+ held-out			MIMIC-IV
5	40	160	4.0 $\times$	34	150	4.4 $\times$
10	50	320	6.4 $\times$	39	300	7.7 $\times$
20	70	640	9.1 $\times$	49	600	12.2 $\times$
30	90	960	10.7 $\times$	59	900	15.3 $\times$

Table 6: Search compute under weight sharing, in GPU-minutes on one NVIDIA L4.

Target	Task	Random	EA	GENIUS	CoLLM-NAS	ATHENA
OneFlorida+	Mortality	58.95 $\pm$ 1.01	58.95 $\pm$ 1.01	59.38 $\pm$ 0.43	58.86 $\pm$ 2.07	58.85 $\pm$ 0.84
OneFlorida+	Stay $>$ 7d	71.76 $\pm$ 2.31	71.76 $\pm$ 2.31	71.56 $\pm$ 0.33	71.16 $\pm$ 4.24	73.90 $\pm$ 0.27 ∗
OneFlorida+	Readmission (3M)	53.18 $\pm$ 0.38	53.18 $\pm$ 0.38	53.45 $\pm$ 0.43	53.20 $\pm$ 0.28	52.56 $\pm$ 0.24
OneFlorida+	Phenotype (6M)	29.42 $\pm$ 2.72	29.42 $\pm$ 2.72	29.76 $\pm$ 0.29	30.48 $\pm$ 0.96	31.06 $\pm$ 0.55
OneFlorida+	Phenotype (12M)	30.96 $\pm$ 2.37	30.96 $\pm$ 2.37	31.46 $\pm$ 0.14	31.70 $\pm$ 0.81	32.54 $\pm$ 0.40 ∗
OneFlorida+	Drug Recommendation	18.91 $\pm$ 0.54	18.91 $\pm$ 0.54	19.06 $\pm$ 0.15	19.08 $\pm$ 0.25	19.34 $\pm$ 0.07 ∗

Table 7: Test AUPRC across target hospitals and prediction tasks under an equal-compute comparison between ATHENA and baselines following the conventional pretrain-then-finetune paradigm, with all methods evaluated under a compute budget matched to that of ATHENA’s search with a maximum budget of 30 architecture evaluations (mean

\pm

standard deviation (SD) over five random seeds).

Target	Task	Random Search	EA	GENIUS	CoLLM-NAS	ATHENA
(a) Search budget = 5 evaluations
OneFlorida+	Mortality	57.78 $\pm$ 0.90	57.78 $\pm$ 0.90	58.04 $\pm$ 0.34	57.72 $\pm$ 1.36	58.23 $\pm$ 1.72
OneFlorida+	Stay $>$ 7d	73.25 $\pm$ 0.33	73.25 $\pm$ 0.33	74.00 $\pm$ 0.34	73.59 $\pm$ 0.45	74.03 $\pm$ 0.37
OneFlorida+	Readmission (3M)	52.09 $\pm$ 0.42	52.09 $\pm$ 0.42	52.06 $\pm$ 0.58	52.16 $\pm$ 0.40	52.26 $\pm$ 0.27
OneFlorida+	Phenotype (6M)	29.65 $\pm$ 0.59	29.53 $\pm$ 0.40	29.28 $\pm$ 0.87	29.29 $\pm$ 1.50	30.42 $\pm$ 0.51
OneFlorida+	Phenotype (12M)	30.86 $\pm$ 0.71	30.74 $\pm$ 0.69	30.39 $\pm$ 0.57	30.23 $\pm$ 0.60	32.17 $\pm$ 0.42

Table 8: Test AUPRC across target hospitals and prediction tasks under increasing NAS budgets of 5, 20, and 30 architecture evaluations (mean

\pm

standard deviation over five random seeds).

Fig 4: Search trajectories on the held-out OneFlorida+ target. Panels show ( a ) Mortality, ( b ) Stay $>$ 7d, ( c ) Readmission (3M), ( d ) Phenotype (6M), ( e ) Phenotype (12M), and ( f ) Drug Recommendation. Lines and shaded bands show the mean and SD, respectively, of the best validation AUPRC attained across five random seeds as a function of the number of evaluated architectures. Curves may end before the full budget when proposal saturation yields no new legal architecture.

Fig 5: Validation AUPRC versus parameter count on the held-out OneFlorida+ target. Markers identify the search method and the dashed line denotes the empirical Pareto frontier across evaluated architectures.

	OneFlorida+		MIMIC-IV
Method	Params (M)	Evals	Params (M)	Evals
Random Search	1.28	30.0	2.34	30.0
EA	1.33	30.0	2.11	30.0
GENIUS	1.90	23.6	2.42	20.3
CoLLM-NAS	1.28	10.7	1.93	9.1
ATHENA	1.55	22.5	2.58	19.6

Table 9: Search behavior at a budget of 30 evaluations, averaged over the six tasks and five random seeds.

Target	Task	Random Search	EA	GENIUS	CoLLM-NAS	ATHENA
OneFlorida+	Mortality	—	(64, 2, 2, 2) (2)	(256, 2, 8, 1) (2)	—	—
OneFlorida+	Stay $>$ 7d	—	—	—	—	(128, 4, 4, 1) (2)
OneFlorida+	Readmission (3M)	(128, 1, 8, 1) (2)	—	—	(128, 4, 4, 2) (2)	(128, 2, 8, 2) (2)
OneFlorida+	Phenotype (6M)	—	(256, 1, 8, 2) (2)	(256, 2, 4, 1) (2)	—	(256, 1, 4, 1) (2)
OneFlorida+	Phenotype (12M)	—	(256, 1, 8, 1) (4)	—	(128, 2, 4, 2) (2)	(256, 1, 8, 1) (4)
OneFlorida+	Drug Recommendation	(256, 2, 8, 2) (2)	(256, 2, 4, 2) (2)	(256, 2, 8, 4) (3)	(256, 4, 8, 2) (3)	—

Table 10: Modal architecture selected by each method, reported as the tuple (embed_dim, depth, heads, mlp_ratio), with the number of random seeds selecting that architecture (out of five) in parentheses.

Fig 6: Cross-hospital prior ablation on the external MIMIC-IV target. Bars and error bars show the mean and SD, respectively, of test AUPRC across five random seeds for exact task-matched retrieval, L1-only retrieval without the Layer 2 architecture-effect prior, LOTO Retrieval excluding exact task matches, Cold Start without either prior layer, and CoLLM-NAS as the representative baseline.

Appendix figures & tables8 assets

Supplementary material from the paper’s appendix.

Appendix

Mortality (binary): Whether the patient dies during the target admission ( $1$ if a recorded in-hospital death time is present, $0$ otherwise). The outcome is assessed only during the target admission, from admission to discharge or death, with no post-discharge follow-up.
Stay $>$ 7d (binary): Whether the length of stay of the target admission exceeds seven days ( $1$ if $>7$ days, $0$ otherwise), measured from admission to discharge. The outcome is defined for the single target admission rather than as a cumulative measure across admissions, and all admissions are included regardless of discharge disposition.
Readmission (3M) (binary): Whether the patient has a subsequent hospital admission within 3 months (90 days) of the target admission ( $1$ if a subsequent admission occurs within 90 days, $0$ otherwise). The interval is measured from the discharge date of the target admission to the admission date of the subsequent admission.
Phenotype (6M) (multilabel, 18 classes): The 18 predefined phenotype categories present among the diagnoses recorded at the patient’s next hospital admission, provided that the next admission occurs within 6 months (180 days) of the target admission. Only target admissions followed by a qualifying next admission are included. The 18 categories are: (1) acute and unspecified renal failure; (2) acute cerebrovascular disease; (3) acute myocardial infarction; (4) cardiac dysrhythmias; (5) chronic kidney disease; (6) chronic obstructive pulmonary disease; (7) conduction disorders; (8) congestive heart failure (nonhypertensive); (9) coronary atherosclerosis and related; (10) disorders of lipid metabolism; (11) essential hypertension; (12) fluid and electrolyte disorders; (13) gastrointestinal hemorrhage; (14) hypertension with complications; (15) other liver diseases; (16) other lower respiratory disease; (17) pneumonia; and (18) septicemia (except in labor).
Phenotype (12M) (multilabel, 18 classes): The same 18 phenotype categories are used as in the 6-month task, but the next hospital admission must occur within 12 months (365 days) of the target admission. The label is an 18-dimensional binary vector, and only target admissions followed by a qualifying next admission are included.
Drug Recommendation (multilabel, 55 classes): The medication classes prescribed during the target admission, based on the admission diagnoses and the patient’s prior visit history. Ground truth consists of physician-prescribed drugs mapped to Anatomical Therapeutic Chemical (ATC) level-4 classes. Of the 140 ATC level-4 classes, the 55 classes present in $\geq 1\%$ across admissions (prevalence, 1.0–21.8%) are retained. The resulting label is a 55-dimensional binary vector. For model input, diagnoses from the target admission are retained, whereas medication, laboratory, and procedure tokens from that admission are masked; all tokens from prior visits remain available. The 55 ATC level-4 classes are: (1) antithrombotic agents (B01A); (2) other analgesics and antipyretics (N02B); (3) stomatological preparations (A01A); (4) laxatives (A06A); (5) drugs for peptic ulcer and gastro-oesophageal reflux (A02B); (6) other mineral supplements (A12C); (7) lipid-modifying agents (C10A); (8) irrigating solutions (B05C); (9) calcium (A12A); (10) beta-blocking agents (C07A); (11) opioids (N02A); (12) antidepressants (N06A); (13) antiemetics and antinauseants (A04A); (14) ACE inhibitors, plain (C09A); (15) antacids (A02A); (16) thyroid preparations (H03A); (17) pancreatic hormones (H04A); (18) insulins and analogues (A10A); (19) selective calcium-channel blockers (C08C); (20) other beta-lactam antibacterials (J01D); (21) high-ceiling diuretics (C03C); (22) vitamin B12 and folic acid (B03B); (23) intestinal antiinfectives (A07A); (24) anxiolytics (N05B); (25) adrenergics, inhalants (R03A); (26) antiepileptics (N03A); (27) other cardiac preparations (C01E); (28) thiazide diuretics (C03A); (29) cardiac vasodilators (C01D); (30) topical antihemorrhoidals (C05A); (31) class I and III antiarrhythmics (C01B); (32) angiotensin-II receptor blockers (C09C); (33) quinolone antibacterials (J01M); (34) antipsychotics (N05A); (35) antigout preparations (M04A); (36) potassium (A12B); (37) intestinal anti-inflammatory agents (A07E); (38) penicillins (J01C); (39) general anesthetics (N01A); (40) arteriolar smooth-muscle agents (C02D); (41) anti-inflammatory and antirheumatic NSAIDs (M01A); (42) dermatological corticosteroids (D07A); (43) propulsives (A03F); (44) antipruritics (D04A); (45) hypnotics and sedatives (N05C); (46) other dermatological preparations (D11A); (47) drugs used in addictive disorders (N07B); (48) cardiac stimulants, excluding glycosides (C01C); (49) topical nasal decongestants (R01A); (50) potassium-sparing agents (C03D); (51) direct-acting antivirals (J05A); (52) sulfonamides and trimethoprim (J01E); (53) cough suppressants (R05D); (54) bacterial vaccines (J07A); and (55) blood glucose-lowering drugs, excluding insulins (A10B).

Appendix

Table S1: Definitions of the six clinical prediction tasks. Each task is defined relative to the target admission , i.e., the hospital admission being modeled. ACE denotes angiotensin-converting enzyme; NSAIDs denote nonsteroidal anti-inflammatory drugs.

	OneFlorida+ Source Sites				Internal Target	External Target
Characteristic	Site A	Site B	Site C	Site D	Site E	MIMIC-IV
Demographics
Patients, n	52,831	55,735	88,527	58,282	84,479	60,706
Admissions, n	144,313	119,579	200,911	97,303	135,955	84,492
Female, n (%)	25,428 (48.13)	27,594 (49.51)	39,775 (44.93)	27,952 (47.96)	46,565 (55.12)	30,171 (49.7)
Utilization

Appendix

Table S2: Baseline characteristics of the six study cohorts. Sites A–D are OneFlorida+ source sites used to construct the cross-hospital architecture prior, Site E is the held-out OneFlorida+ internal target, and MIMIC-IV is the external target. Continuous variables are reported as mean (SD) or median [IQR] and categorical variables as n (%). SD denotes standard deviation and IQR denotes interquartile range. Comorbidities are reported as the number and percentage of admissions with the corresponding diagnosis.

Task	Feature pair	Rule supplied to the prompt
Mortality	embed_dim $\times$ heads	Prefer embed_dim $\in\{128,256\}$ , with the preference for 256 applying when heads $\geq 2$ ; discourage embed_dim $\in\{32,64\}$ .
Stay $>$ 7d	depth $\times$ embed_dim	Prefer depth $\geq 2$ ; discourage depth $=1$ across all embed_dim levels.
Readmission (3M)	heads $\times$ embed_dim	Prefer heads $\in\{4,8\}$ ; discourage heads $\in\{1,2\}$ across all embed_dim levels.
Phenotype (6M)	embed_dim $\times$ depth	Prefer embed_dim $\in\{128,256\}$ ; discourage embed_dim $\in\{32,64\}$ across all depth levels.
Phenotype (12M)	embed_dim $\times$ depth	Prefer embed_dim $\in\{128,256\}$ ; discourage embed_dim $\in\{32,64\}$ across all depth levels.
Drug Recommendation	embed_dim $\times$ depth	Prefer embed_dim $\in\{128,256\}$ ; discourage embed_dim $\in\{32,64\}$ across all depth levels.

Appendix

Table S3: Layer 2 pairwise architecture-effect rules supplied to the search prompt (

\mathcal{P}_{\mathrm{meta}}

). For each task, the interaction model uses the two most influential architecture features. The table reports feature-level combinations with estimated effects whose 95% confidence intervals (CIs) exclude zero.

Target	Task	Random Search	EA	GENIUS	CoLLM-NAS	ATHENA
OneFlorida+	Mortality	89.15 $\pm$ 0.72	89.71 $\pm$ 0.51	89.07 $\pm$ 1.15	88.94 $\pm$ 1.49	89.30 $\pm$ 0.95
OneFlorida+	Stay $>$ 7d	90.32 $\pm$ 0.31	90.31 $\pm$ 0.13	90.46 $\pm$ 0.13	90.39 $\pm$ 0.30	90.30 $\pm$ 0.25
OneFlorida+	Readmission (3M)	70.22 $\pm$ 0.28	70.52 $\pm$ 0.48	70.38 $\pm$ 0.19	70.29 $\pm$ 0.37	70.51 $\pm$ 0.35
OneFlorida+	Phenotype (6M)	78.74 $\pm$ 0.55	78.82 $\pm$ 0.31	78.39 $\pm$ 0.32	78.18 $\pm$ 0.71	78.82 $\pm$ 0.25
OneFlorida+	Phenotype (12M)	79.86 $\pm$ 0.40	79.84 $\pm$ 0.44	79.64 $\pm$ 0.26	78.85 $\pm$ 0.25	80.05 $\pm$ 0.10
OneFlorida+	Drug Recommendation	71.21 $\pm$ 0.08	71.26 $\pm$ 0.08	71.31 $\pm$ 0.07	71.20 $\pm$ 0.18	71.28 $\pm$ 0.09

Appendix

Table S4: Test area under the receiver operating characteristic curve (AUROC) at a search budget of 30 architecture evaluations (mean

\pm

standard deviation over five random seeds). Within each hospital–task comparison, the best mean is shown in bold and the second-best mean is underlined ; ties share the same mark. Avg. Rank denotes the mean rank across the 12 hospital–task comparisons (1 = best; ties receive the average rank), with the number of best-in-row comparisons shown in parentheses. EA denotes evolutionary algorithm.

Fig S1: Cross-hospital architecture prior learned by the Layer 2 meta-regression across four OneFlorida+ source sites. Rows denote the six prediction tasks: ( a ) Mortality, ( b ) Stay $>$ 7d, ( c ) Readmission (3M), ( d ) Phenotype (6M), ( e ) Phenotype (12M), and ( f ) Drug Recommendation; columns denote architecture dimensions. Points show mixed-effects SHapley Additive exPlanations (SHAP) estimates with 95% CIs for each hyperparameter level. Green indicates preferred levels (CI $>$ 0), red indicates discouraged levels (CI $<$ 0), and gray indicates inconclusive levels whose CIs span 0. Gray points at zero without CIs denote reference levels.

Fig S2: Search trajectories on the held-out MIMIC-IV target for ( a ) Mortality, ( b ) Stay $>$ 7d, ( c ) Readmission (3M), ( d ) Phenotype (6M), ( e ) Phenotype (12M), and ( f ) Drug Recommendation. Lines and shaded bands show the mean and SD, respectively, of the best validation AUPRC attained across five random seeds as a function of the number of evaluated architectures. Curves may end before the full budget when proposal saturation yields no new legal architecture.

Fig S3: Validation AUPRC versus parameter count on the held-out MIMIC-IV target for ( a ) Mortality, ( b ) Stay $>$ 7d, ( c ) Readmission (3M), ( d ) Phenotype (6M), ( e ) Phenotype (12M), and ( f ) Drug Recommendation. Markers denote search methods, and the dashed line indicates the empirical Pareto frontier across evaluated architectures. Validation performance guided architecture search and selection; the selected architecture was evaluated only once on the held-out test set.

Fig S4: Cross-hospital prior ablation on the held-out OneFlorida+ target. Bars and error bars show the mean and SD, respectively, of test AUPRC across five random seeds for exact task-matched retrieval, L1-only retrieval without the Layer 2 architecture-effect prior, Leave-One-Task-Out (LOTO) Retrieval excluding exact task matches, Cold Start without either prior layer, and CoLLM-NAS as the representative baseline.

Neural Architecture Search

Neural Architecture Search automates neural network design for target tasks, resource budgets, and hardware constraints.

See all 132 papers

Neural at ArchEHR-QA 2026: One Method Fits All: Unified Prompt Optimization for Clinical QA over EHRs
May 11, 2026Abrar Majeedi, Viswanatha Reddy Gajjala, Sai Prasanna Teja Reddy Bogireddy +1Healthcare Clinical QA
University of Wisconsin–Madison · University of Chicago · Independent Researcher
EveryQuery: A Promptable Foundation Model for Clinical Prediction Tasks over Electronic Health Records
Mar 9, 2026Payal Chandak, Gregory Kondas, Liat Antwarg Friedman +2EHR Foundation Models Clinical Outcome Prediction
Harvard-MIT HST · Columbia University · Harvard Medical School
HypEHR: Hyperbolic Modeling of Electronic Health Records for Efficient Question Answering
Apr 22, 2026Yuyu Liu, Sarang Rajendra Patil, Mengjia Xu +1Healthcare Hierarchical Representation Learning
Department of Computer Science, Stony Brook University · Department of Data Science, New Jersey Institute of Technology · Department of Biomedical Informatics, Stony Brook University

May 11, 2026J/K move · Enter open · S save

Neural at ArchEHR-QA 2026: One Method Fits All: Unified Prompt Optimization for Clinical QA over EHRs

Abrar Majeedi, Viswanatha Reddy Gajjala, Sai Prasanna Teja Reddy Bogireddy, Siddhant Rai

University of Wisconsin–Madison · University of Chicago · Independent Researcher

Healthcare Clinical QA Evidence-Grounded Generation Electronic Health Records EHR QA

Open paper AskPDF arXiv

Automated question answering (QA) over electronic health records (EHRs) demands precise evidence retrieval, faithful answer generation, and explicit grounding of answers in clinical notes. In this work, we present Neural1.5, our method for the ArchEHR-QA 2026 shared task at CL4Health@LREC 2026, which comprises four subtasks: question interpretation, evidence identification, answer generation, and evidence alignment. Our approach decouples the task into independent, modular stages and employs DSPy"s MIPROv2 optimizer to automatically discover high-performing prompts, jointly tuning instructions and few-shot demonstrations for each stage. Within every stage, self-consistency voting over multiple stochastic inference runs suppresses spurious errors and improves reliability, while stage-specific verification mechanisms (e.g., self-reflection and chain-of-verification for alignment) further refine output quality. Among all teams that participated in all four subtasks, our method ranks second overall (mean rank 4.00), placing 4th, 1st, 4th, and 7th on Subtasks 1-4, respectively. These results demonstrate that systematic, per-stage prompt optimization combined with self-consistency mechanisms is a cost-effective alternative to model fine-tuning for multifaceted clinical QA.

Abstract

Neural Architecture Search

Explore similar work

Neural at ArchEHR-QA 2026: One Method Fits All: Unified Prompt Optimization for Clinical QA over EHRs

EveryQuery: A Promptable Foundation Model for Clinical Prediction Tasks over Electronic Health Records

HypEHR: Hyperbolic Modeling of Electronic Health Records for Efficient Question Answering