Skip to content
DopeSwagYolo

AI in Biotech

What are clinical trial phases?

Clinical trial phases are the numbered stages of testing a new drug or vaccine in people. Each phase generally enrolls more participants than the last and answers a different question, from basic safety to whether the treatment works.

Also known as: Phase 1, 2 and 3 trials, phases of clinical research

Researched and fact-checked by AI, with no human review. 6 sources listed below. How we verify

Last updated

How it works

In the United States, a developer must submit an Investigational New Drug (IND) application to the Food and Drug Administration before testing a drug in people. That is according to the FDA, which says it has 30 days to review the application. Federal regulation 21 CFR 312.21 divides the study of a previously untested drug into three phases. These generally run in sequence but may overlap.

The FDA's patient guide, dated January 2018, describes them this way:

  • Phase 1: 20 to 100 healthy volunteers or patients, over several months, to check safety and dosage. About 70% of drugs move on.
  • Phase 2: up to several hundred patients, for several months to two years, to assess efficacy (whether the drug works) and side effects. About 33% move on.
  • Phase 3: 300 to 3,000 volunteers with the condition, for one to four years, to assess efficacy and monitor adverse reactions. About 25% to 30% move on.
  • Phase 4: several thousand volunteers with the condition. The National Institutes of Health says these studies track safety after a drug is approved.

These are typical ranges, and sources differ. The regulation puts Phase 1 at 20 to 80 subjects. NIH puts Phase 3 at 1,000 to 3,000.

Why it matters for AI in biotech

Drugs found or designed with AI go through the same kind of phased trials. One example is rentosertib. The trial's authors describe it as an AI-generated molecule aimed at a target also found with generative AI. Results of a 12-week Phase 2a trial in 71 patients with idiopathic pulmonary fibrosis, a progressive lung disease, appeared in Nature Medicine in June 2025. Insilico Medicine sponsored the trial and employs several authors. Adverse event rates were similar across drug and placebo groups. A measure of lung capacity rose at the highest dose. The authors called for larger, longer trials.

Where things stand in 2026

As of a May 2026 update, the FDA says its drug center saw more than 500 submissions with AI components from 2016 to 2023. It issued draft guidance in January 2025 on using AI to support regulatory decisions. In January 2026 it published 10 guiding principles developed with the European Medicines Agency.

A Perspective in Nature Reviews Drug Discovery, published in August 2026, said evidence of AI methods' clinically relevant impact on drug discovery was, so far, limited.

ShareXLinkedInFacebookRedditBlueskyEmail

Sources

  1. Step 3: Clinical Research, U.S. Food and Drug Administration (FDA)
  2. 21 CFR 312.21 -- Phases of an investigation., Electronic Code of Federal Regulations (eCFR)
  3. The Basics, National Institutes of Health (NIH)
  4. A generative AI-discovered TNIK inhibitor for idiopathic pulmonary fibrosis: a randomized phase 2a trial, Nature Medicine
  5. Artificial Intelligence for Drug Development, U.S. Food and Drug Administration (FDA)
  6. Artificial intelligence in drug discovery — what it is, where we stand and the path forward, Nature Reviews Drug Discovery

Articles on AI in Biotech