Abstract
<title>Abstract</title> <p> <bold>Purpose</bold> : In personal injury cases, parties must support medical claims with expert opinions, which can be biased, and the question is whether AI can detect such bias. <bold>Results:</bold> In a sample of malpractice cases, Claude and OpenEvidence rapidly and effectively retrieved published literature relevant to disputed issues, and cross-model agreement was high. <bold/> Evidence-based conclusions, reached by AI tools within seconds or minutes, often aligned with the court's ruling, which had been reached after years of deliberation. A scale to measure deviation from evidence in experts' positions was developed and validated on a separate sample. Defense expert testimony often deviated from published evidence, even in open contradiction, consistently and significantly more often than plaintiff testimony did. <bold/> <bold>Conclusion</bold> : AI tools can provide rapid, accurate summaries of evidence in malpractice cases and detect deviations from that evidence in expert testimony, suggesting a role for such tools in supporting expert preparation, judicial review, focusing the dispute, and settling mediation. </p>