In 2026, AI is being used to make clinical trial eligibility criteria understandable for patients. ASCO 2026 Breakthrough found that "eligible patients and caregivers reading clinical trial information prefer to read a version of the information that had been optimized by a LLM rather than the original, standard description provided on ClinicalTrials.gov." Clinical Leader notes that when patients ask "Is this for someone like me?" they "are not reciting inclusion criteria. They are assessing personal relevance" — and need "structured plain- language framing of fit."

Nature Communications published TrialGPT, "an end-to-end framework for zero-shot patient-to-trial matching with large language models" that "first performs large-scale filtering to retrieve candidate trials, then predicts criterion-level patient eligibility, and finally generates trial-level scores." Nature Communications also published research on "human-AI teaming to improve accuracy and efficiency of eligibility criteria prescreening for oncology trials." JMIR published a framework for "structuring and visualizing clinical trial eligibility criteria at scale." PMC notes that "registries like ClinicalTrials.gov often present complex medical jargon that is difficult for the general public to understand."

But the ASCO finding is the key signal: patients prefer AI- simplified trial information over the standard ClinicalTrials.gov descriptions. The preference exists because the standard descriptions are unreadable. And unreadable eligibility criteria mean patients cannot determine if they qualify for a trial that could save their life.

The clinical trial eligibility readability problem

1. Medical jargon in patient-facing trial descriptions

Clinical trial eligibility criteria use medical terminology: "Inclusion: histologically confirmed non-small cell lung cancer (NSCLC), stage IIIB or IV, ECOG performance status 0-1, measurable disease per RECIST 1.1, adequate organ function (ANC >= 1.5 x 10^9/L, platelets >= 100 x 10^9/L, creatinine <= 1.5x ULN). Exclusion: prior systemic therapy for advanced disease, active CNS metastases, uncontrolled autoimmune disease, pregnancy." The patient needs to know: "You may qualify if: you have lung cancer (specific type, stage 3B or 4), you're well enough to care for yourself (ECOG 0-1), your tumors can be measured on scans, your blood counts and kidney function are adequate. You don't qualify if: you've already had chemo for advanced cancer, you have cancer spread to the brain (unless treated), you have an autoimmune disease that's not controlled, or you're pregnant." The medical version is for the investigator. The plain version is for the patient who is searching for a trial that could save their life.

2. The ClinicalTrials.gov readability gap

PMC identifies the problem directly: ClinicalTrials.gov "often presents complex medical jargon that is difficult for the general public to understand." ClinicalTrials.gov is the primary public registry of clinical trials — the database patients search when looking for experimental treatments. When the eligibility criteria on this database are at a college reading level with medical jargon, the patient searching for hope cannot understand whether they qualify.

3. The personal relevance gap

Clinical Leader's insight is key: patients ask "Is this for someone like me?" They are not checking inclusion criteria — they are assessing whether the trial is relevant to their situation. When the trial description is in medical jargon, the patient cannot answer the personal relevance question. They need a plain-language summary that says: "This trial is for people with stage 4 lung cancer who have not had chemotherapy yet. If that's you, ask your doctor about this trial."

4. The enrollment gap

The ultimate consequence of unreadable eligibility criteria is underenrollment. Nature Communications notes "patient recruitment is challenging for clinical trials" — and TrialGPT was developed specifically to address this. But the recruitment challenge is partly a readability challenge: patients cannot enroll in trials they cannot understand. If the eligibility criteria are in jargon, the patient does not know they qualify, does not ask their doctor, and does not enroll — missing access to potentially life-saving experimental treatment.

Why AI-generated trial descriptions are hard to read

1. Medical protocol content as training data

AI trial tools train on clinical trial protocols, medical databases, and ClinicalTrials.gov entries — all written in medical language at a high reading level. The generated descriptions reproduce this register.

2. Scientific accuracy over patient comprehension

AI trial tools optimize for scientific accuracy: correct criteria, correct staging, correct biomarkers. This produces descriptions that satisfy the investigator but not the patient who is searching for a trial on their phone at 2 AM.

3. The ASCO preference signal

ASCO found patients prefer AI-simplified information over standard descriptions. The preference is the market signal: standard descriptions are not meeting patient needs. AI simplification works — but it needs to go all the way to the patient's reading level.

4. The ClinicalTrials.gov gap

ClinicalTrials.gov is the primary public database, but its entries are written for researchers, not patients. The "plain language" option on ClinicalTrials.gov is limited and not universally available. The readability gap is structural — the database is designed for the research community, not the patient community.

The fix: readable clinical trial eligibility criteria

The fix is to make clinical trial eligibility criteria readable for the patients who are searching for trials:

1. AI generates the trial description with all eligibility

criteria

1. Simplify the patient-facing version with

ELI5 AI — rewrite at a 5th-grade reading level with four levels side-by-side

1. Use the appropriate level:

• Adult level for the full protocol (for the investigator)
• Teenager level for the patient-facing trial summary
• 5-year-old level for the key question: who qualifies, what

the trial tests, how to ask about enrollment

1. **Publish the simplified version alongside the ClinicalTrials.gov

entry** — so patients searching for trials find readable descriptions

How ELI5 AI helps clinical trial teams

ELI5 AI is a free, no-login text simplifier that takes any clinical trial eligibility criteria and rewrites it at a 5th-grade reading level with four levels side-by-side: 5-year-old, 10-year-old, teenager, and adult.

For clinical trial teams, the workflow is:

1. AI generates the trial description (using TrialGPT, NLP tools,

or manual drafting from the protocol)

1. Paste the eligibility criteria into ELI5 AI — get a plain-

language version at four reading levels

1. Use the appropriate level:

• Adult level for the full protocol (for the research team)
• Teenager level for the patient-facing trial summary (for

ClinicalTrials.gov and patient recruitment sites)

• 5-year-old level for the key question: "This trial is for

people with stage 4 lung cancer who haven't had chemo yet. It tests a new drug that helps your immune system fight cancer. If that's you, ask your doctor about this trial or call [number]."

1. Publish the simplified version on the trial's patient

recruitment page and alongside the ClinicalTrials.gov entry

What clinical trial teams should do now

• Audit your trial eligibility criteria for readability. Sample

10 recent trial descriptions. Can a non-medical reader understand who qualifies? If not, patients are missing trials they could enroll in

• Simplify the patient-facing summary first. Use

ELI5 AI to produce a plain-language version of every trial description. The patient needs: who qualifies, what the trial tests, how to ask

• Follow the ASCO evidence. ASCO found patients prefer AI-

simplified information. Use ELI5 AI to deliver the simplified information patients prefer

• Follow the Clinical Leader principle. Patients ask "Is this

for someone like me?" Answer the question in plain language — not in inclusion criteria

• Publish simplified criteria on recruitment pages. The

ClinicalTrials.gov entry can stay in medical language for researchers. The patient recruitment page should be in plain language

• Track enrollment against readability. If simplified

descriptions lead to more patient inquiries and higher enrollment, readability is a trial access metric

The bottom line

Clinical trial eligibility criteria determine whether patients can access experimental treatments that could save their lives. When criteria are in medical jargon on ClinicalTrials.gov, patients cannot determine if they qualify — and miss access to trials that could help them. The ASCO evidence shows patients prefer AI-simplified information. The preference is clear. The need is clear. The fix is readability.

The clinical trial teams that add a readability layer to their AI trial descriptions will give patients the information they need to find and enroll in trials. The teams that do not will continue publishing eligibility criteria in medical jargon — and watching patients miss trials they could have enrolled in because they could not read the description.

Clinical trial eligibility criteria the patient cannot read do not enable enrollment. They prevent it. Plain language turns them into criteria that open the door to potentially life-saving treatment.

Try it: paste any clinical trial eligibility criteria, trial description, or patient recruitment communication into ELI5 AI and get a plain-English version patients can understand and act on — four reading levels, no account required.