How to Daubert Challenge an Expert Witness in 2026 (AI-Enhanced Strategy)

The Daubert standard sits at the center of every significant medical malpractice trial. Under Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), and the subsequent Federal Rule of Evidence 702 amendments, federal courts (and most state courts) require expert testimony to rest on reliable methodology applied to sufficient facts and data. In practice, most Daubert challenges fail — not because the expert's methodology is sound, but because the attorney didn't surface the methodology gaps in the deposition.

This guide walks through how plaintiff (and defense) attorneys structure Daubert challenges in 2026, where AI tools add real leverage during the deposition itself, and where human judgment still wins.

The four Daubert factors (and one Rule 702 addition)

Federal Rule of Evidence 702 (as amended 2023) requires the expert's testimony to:

1. Be based on sufficient facts or data
2. Use reliable principles and methods
3. Reflect a reliable application of those methods to the facts
4. Help the trier of fact understand the evidence or determine a fact in issue

The classic Daubert factors for assessing reliability:

• Has the theory or technique been tested?
• Has it been subjected to peer review and publication?
• What is the known or potential rate of error?
• Are there standards controlling its operation?
• Is it generally accepted in the relevant scientific community?

The 2023 amendment also emphasized that courts must find, by a preponderance of the evidence, that each Rule 702 requirement is satisfied before admitting expert testimony. That tightened the gatekeeping standard meaningfully.

The three most productive Daubert attacks in medical malpractice

1. Methodology not disclosed or not generally accepted

When an expert says "in my experience" or "based on my clinical judgment" to justify an opinion about, say, standard of care — that's not a methodology. Push for:

• What specific peer-reviewed literature forms the basis of the opinion?
• What diagnostic criteria (NIH guidelines, specialty society position statements, Bradford Hill factors for causation) are being applied?
• Has the methodology been validated in the expert's own publications?

If the expert can't articulate a methodology beyond "I just know," a Daubert motion has legs.

2. Insufficient facts or data

This is the most common Daubert win — and the most overlooked attack angle. It requires forcing the expert to concede that they DID NOT review:

• The complete medical record (often they review only what opposing counsel sent)
• Post-incident records showing alternative causation
• The deposition transcripts of co-treating providers
• Relevant imaging studies in their original form (not just radiologist reports)

When an expert testifies about causation or standard of care while admitting they didn't review X, Y, and Z — that's Rule 702(a) "sufficient facts or data" failure. Judges care about this a lot more now post-2023 amendment.

3. Unreliable application (the Rule 702(d) play)

Even when the underlying methodology is sound, the expert might misapply it to these facts. Examples:

• Differential diagnosis done without ruling out plausible alternatives
• Causation opinion based on a study population different from this patient
• Standard-of-care opinion from a different geographic / institutional context than the defendant

Where AI deposition analysis changes the Daubert game

Here's what we've learned running live AI analysis across dozens of medical malpractice depositions: the human attorney's conscious mind can't hold a two-hour testimony stream in working memory while simultaneously preparing the next question. That's exactly the problem AI solves.

What AI catches in real time

• Methodology-disclaimed-while-opining patterns. When an expert says "I don't have the screen" or "I didn't review that record" in minute 30, then offers a prevalence claim in minute 90, AI cross-links them. A human attorney tracking the next question usually doesn't.
• Internal contradictions within the deposition. The expert defines "standard of care" narrowly early, then applies it broadly later.
• Prior testimony inconsistencies. If you've uploaded 50+ pages of the expert's prior depositions, the AI surfaces direct contradictions with exact quotes from both.
• Non-responsive answers that signal evasion. "Your question was different" is a tell. AI flags these patterns and scores the expert's evasiveness score — useful for the jury presentation.

What AI still doesn't do

AI won't:

• Decide STRATEGY (when to press, when to pivot, when to let a bad answer hang)
• Read body language or tone
• Assess jury dynamics (that's what closing argument is for)
• Replace your medical expert's review of the same record

The 10-question Daubert deposition outline

Regardless of specialty, these 10 questions surface Daubert vulnerabilities reliably. Use them verbatim or adapt.

1. Qualifications scoping: "Are you testifying as a general expert in [field], or specifically as an expert in [narrower sub-specialty that matters in this case]?"
2. Literature foundation: "What specific peer-reviewed articles or treatises form the basis of your opinion on [key issue]?"
3. Methodology articulation: "Can you walk me through the step-by-step methodology you used to arrive at your opinion on [causation / standard of care]?"
4. Data sufficiency: "What medical records, deposition transcripts, and other materials did you review before forming your opinion? What did you NOT review?"
5. Differential diagnosis completeness: "What alternative causes did you consider and rule out? How did you rule each out?"
6. Rate of error / uncertainty: "In your clinical experience, what is the approximate rate of misdiagnosis for [condition]? What factors increase that rate?"
7. Validation: "Have you ever published, peer-reviewed, or validated the specific methodology you're applying here?"
8. Prior inconsistent positions: "Have you testified differently in any prior case about [related issue]?"
9. Compensation transparency: "What is your total compensation for this case, and what portion of your annual income comes from expert witness work?"
10. Opinion reliability: "Is there any aspect of your opinion you hold with less than reasonable degree of medical certainty?"

Each of these maps to a specific Rule 702 prong. The AI tools running during the deposition flag when any answer opens a Daubert vulnerability — giving you the follow-up question before you'd otherwise have thought of it.

Common mistakes that kill Daubert motions

• Waiting until the Daubert motion to raise methodology issues. The deposition record must already contain the damaging admissions. If it doesn't, the motion fails.
• Not getting the expert to commit. Ambiguous answers ("generally I would consider...") are worthless. Press for yes/no or specific citations.
• Focusing on the conclusion, not the methodology. Judges care whether the methodology is reliable, not whether you disagree with the conclusion.
• Missing the 702(b) "sufficient facts" attack. This is the most winnable prong and the most commonly overlooked.

After the deposition: structuring the Daubert motion

Open with the strongest single quote from the deposition — ideally one where the expert disclaims methodology or data. Organize the brief around which Rule 702 prongs fail, with citations to the deposition transcript at each point.

For each argued defect:

1. State the Rule 702 prong (a/b/c/d)
2. Quote the expert's admission verbatim with transcript cite
3. Explain why the admission fails the prong (with legal citation)
4. Argue why the testimony should be excluded or limited

Related reading

• How AI Analyzes a Two-Hour Deposition in Real Time — walkthrough with actual output samples
• How AI Catches Expert Witness Inconsistencies During Live Depositions
• Medical Malpractice Discovery Checklist