Public-interest biomedical researchClinical trials in a dish · AI drug discovery bottleneckcontact@basedresearch.org
Based Research Foundation

Clinical trials in a dish

Based Research Foundation studies how human-relevant in-vitro systems can help evaluate AI-discovered drug candidates before scarce clinical trial capacity is committed.

Illustration of organoid culture dishes connected to an AI candidate selection network.
The laboratory image illustrates where early biological models can support evidence decisions before a candidate enters human trials.

Story reel

Work in motion

Preclinical signal quality check

Preclinical signal quality check

How candidate ranking changes when evidence quality is treated as a decision gate instead of a decorative metric.

Read research
Infrastructure for candidate triage

Infrastructure for candidate triage

A practical framework for moving from AI-generated hypotheses to a shortlist that can support translational planning.

See programs
Open evidence records

Open evidence records

Why methods, assumptions, and benchmark rationale must be visible before candidates enter expensive downstream work.

About the nonprofit
Clinical signal before patients

Clinical signal before patients

The nonclinical layer should reduce uncertainty, not create new ambiguity in candidate progression decisions.

Read note
Preclinical signal quality check

Preclinical signal quality check

How candidate ranking changes when evidence quality is treated as a decision gate instead of a decorative metric.

Read research
Infrastructure for candidate triage

Infrastructure for candidate triage

A practical framework for moving from AI-generated hypotheses to a shortlist that can support translational planning.

See programs
Open evidence records

Open evidence records

Why methods, assumptions, and benchmark rationale must be visible before candidates enter expensive downstream work.

About the nonprofit
Clinical signal before patients

Clinical signal before patients

The nonclinical layer should reduce uncertainty, not create new ambiguity in candidate progression decisions.

Read note

The bottleneck is evidence.

AI systems can nominate far more candidates than clinical programs can responsibly test. The hard problem is choosing which candidates deserve costly biological and patient-facing studies.

Read the research approach

Dish-scale models can be decision infrastructure.

Organoids and microphysiological systems can help expose weak biological claims early and make candidate prioritization more legible.

Programs

Why start with evidence?

AI discovery expands the candidate list. Clinical translation narrows it. The work in between needs better public methods, not another slogan.

About the nonprofit

News and updates

What a “trial in a dish” should prove

A note on the difference between useful preclinical signal and decorative biological complexity.

Research update · Apr 15Read news

Why candidate triage matters after AI generation

The field needs better filters between model output and the clinical pipeline.

Program note · Apr 10Read blog

About the foundation

Nonprofit status, legal details, and contact information are in the About section.

About page · Apr 3View about