Research

Clinical Trials

Clinical Trial Intelligence and Optimization

Clinical trials are a full lifecycle problem. Sunlab builds data, models, and agents that support review, design, recruitment, execution, and outcome reasoning.

01Trial records and literature
02Structured trial knowledge
03AI agents and predictors
04Better design and recruitment

Recent papers

What this program is building

Selected recent and foundational papers, summarized around the task, why it matters, and the main technical result.

2026Nature Health

A Large-Scale Database for Clinical Trial Outcomes and Features

1Trial records
2Outcome features
3Trial intelligence database
Task
Build a large database connecting clinical trial features with trial outcomes.
Why it matters
Trial design and portfolio decisions need better empirical evidence about what predicts success or failure.
Main result
The database creates a foundation for benchmarking outcome prediction and trial intelligence models.
Paper details
2026Nature Health

Matching Clinicians with Clinical Trials Using AI

1Clinical expertise
2Trial criteria
3Clinician-trial match
Task
Recommend relevant trials to clinicians who can identify and refer eligible patients.
Why it matters
Clinician-trial matching can improve recruitment by connecting studies to the right clinical networks.
Main result
The work reframes recruitment as a matching problem across clinicians, specialties, patients, and trial needs.
Paper details
2025npj Digital Medicine

Accelerating Clinical Evidence Synthesis with Large Language Models

1Study corpus
2LLM extraction
3Evidence table
Task
Use LLMs to assist evidence extraction and synthesis for clinical research.
Why it matters
Trial teams and guideline groups need faster ways to turn studies into structured evidence.
Main result
The paper shows LLM-supported synthesis workflows can reduce manual burden while preserving review structure.
Paper details
2025ACL

SECRET: Semi-supervised Clinical Trial Document Similarity Search

1Trial documents
2Semi-supervised search
3Similar protocol set
Task
Find similar clinical trial documents with limited labeled supervision.
Why it matters
Trial designers can reuse precedent and compare protocols more efficiently when trial documents are searchable by meaning.
Main result
Semi-supervised similarity search improves retrieval for clinical trial documents.
Paper details
2024Patterns

FRAMM: Fair Ranking with Missing Modalities for Clinical Trial Site Selection

1Site data gaps
2Fair ranking model
3Site selection list
Task
Rank clinical trial sites fairly when some data modalities are missing.
Why it matters
Site selection affects recruitment speed, representativeness, and operational cost.
Main result
FRAMM addresses fairness and missing data together for more reliable site ranking.
Paper details
2024Nature Communications

Matching Patients to Clinical Trials with Large Language Models

1Patient evidence
2Eligibility language
3Trial match
Task
Match patients to trial eligibility criteria from structured and unstructured clinical information.
Why it matters
Better matching can reduce screening burden and help eligible patients find appropriate studies.
Main result
LLMs support eligibility reasoning by aligning patient evidence with protocol criteria.
Paper details

Representative publication links

Interested in this program?

Send a concise note with the program name, your role, the problem you want to work on, and any relevant data, code, clinical setting, or research experience.

Contact Sunlab
Clinical AI Medical LLMs Drug Discovery