Benchmark Integrity and Reasoning-Trace Errors in Medical Question Answering With Large Language Models: Mixed Methods Study With Sparse Autoencoders - Top-Institutions - MDSpire

Benchmark Integrity and Reasoning-Trace Errors in Medical Question Answering With Large Language Models: Mixed Methods Study With Sparse Autoencoders

  • By

  • Jialin Liu

  • Siru Liu

  • Adam Wright

  • June 12, 2026

  • 0 min

Share

Top Institutions in Internal Medicine

Brief introduction explaining scope and methodology.

  • #1

    Stanford University School of Medicine
    Stanford University School of Medicine

    Stanford, California

    Key Differentiators

    • Medical Informatics

  • #2

    Harvard Medical School
    Harvard Medical School

    Boston, Massachusetts

    Key Differentiators

    • Medical Informatics

  • #3

    Mount Sinai
    Mount Sinai

    New York, New York

    Key Differentiators

    • Medical Informatics

  • #4

    Mass General Brigham
    Mass General Brigham

    Boston, Massachusetts

    Key Differentiators

    • Medical Informatics

  • #5

    NYU Langone Health
    NYU Langone Health

    New York, New York

    Key Differentiators

    • Medical Informatics

Original Source(s)

Benchmark Integrity and Reasoning-Trace Errors in Medical Question Answering With Large Language Models: Mixed Methods Study With Sparse Autoencoders

  • by Jialin Liu, Siru Liu, Adam Wright

  • June 12, 2026

Related Content

Application of large language models in medical diagnosis: A bibliometric review

Peptides Are Outrunning the Rulebook

A new JAMA Viewpoint argues US drug policy is built for a market that no longer behaves like one.

Radiation-associated bioprosthetic valve dysfunction: an initial case-control analysis