Clinical Requirements for Transparent Machine Learning Model Information: A Mixed Methods Study Protocol.

Journal: Studies in health technology and informatics
Published Date:

Abstract

Limited transparency of machine learning models poses risks their effective use. Through semi-structured interviews with physicians, this mixed methods study will qualitatively identify requirements for transparent machine learning model information for a diagnostic decision support system in the emergency department. Then, a prototype will be developed and tested, aiming to align clinical needs with regulatory requirements and improve the responsible use of artificial intelligence in healthcare.

Authors

  • Louis Agha-Mir-Salim
    Institute of Medical Informatics, Charité - Universitätsmedizin Berlin, Germany.
  • Nicolas Frey
    Institute of Medical Informatics, Charité - Universitätsmedizin Berlin, Germany.
  • Lina Mosch
    Institute of Medical Informatics, Charité - Universitätsmedizin Berlin, Germany.
  • Felix Balzer
    Charité - Universitätsmedizin Berlin, corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Department of Anesthesiology and Operative Intensive Care Medicine (CCM, CVK), Charitéplatz 1, 10117, Berlin, Germany.