About CINEX

Overview

CINEX (Clinical Information Extraction) is a consensus-based reporting guideline for studies that extract information from clinical text. It gives researchers, clinicians and data scientists a structured framework for documenting and reporting how a clinical information extraction study was designed, built and evaluated.

The primary goal is to improve the reproducibility, transparency and comparability of clinical information extraction research. By following the guideline, authors can be confident that the methods they report are complete enough for others to understand, appraise and replicate.

The five dimensions

The 29 reporting items are grouped into five dimensions, each covering a different aspect of an extraction study.

1. Information modelIM1–IM5

   What is extracted, and the clinical and technical models that give it meaning.

2. ArchitectureA1–A5

   The task, the implementation, and the compute behind it.

3. DataD1–D9

   Where the texts come from, how they were split, and who they represent.

4. AnnotationAN1–AN5

   How the reference standard was produced, and how reliable it is.

5. OutcomeO1–O5

   Performance, validation, and what is available to others.

Browse all 29 items with examples →

How to use it

• As a checklist when designing a clinical information extraction study
• As a reporting standard when writing a manuscript
• As a training resource for researchers new to clinical information extraction
• As a quality assurance framework for reviewing existing extraction pipelines

Authors

CINEX was developed by an international team of experts in clinical information extraction.

• 
  Prof. Daniel ReichenpfaderInstitute of Patient-Centered Digital Health, Bern University of Applied SciencesPhD School of Life Sciences, Faculty of Medicine, University of Geneva
• 
  Fabio Dennstädt, MDDepartment of Radiation Oncology, Inselspital, Bern, University Hospital and University of Bern, Bern, Switzerland
• 
  Jamil Zaghir, PhDDivision of Medical Information Sciences, Geneva University Hospitals, Geneva, SwitzerlandDepartment of Radiology and Medical Informatics, University of Geneva, Geneva, Switzerland
• 
  Prof. Stéphane Meystre, MD, PhD, FACMI, FIAHSI, FAMIAInstitute of Digital Technologies for Personalised Healthcare (MeDiTech), University of Applied Sciences and Arts of Southern Switzerland (SUPSI), Lugano, Switzerland
• 
  Prof. Dr. Kerstin DeneckeInstitute of Patient-Centered Digital Health, Bern University of Applied Sciences
• HM
  Prof. Dr. Henning MüllerDepartment of Radiology and Medical Informatics, University of Geneva, Geneva, SwitzerlandInformatics Institute, HES-SO Valais-Wallis, Sierre, Switzerland
• 
  Boya Zhang, MEng, MScFaculty of Medicine, University of Geneva, Geneva, Switzerland
• 
  Boris Delange, MDCHU Rennes, INSERM, LTSI-UMR 1099, Univ Rennes, 35000 Rennes, France
• 
  Prof. Douglas Teodoro, PhDDepartment of Radiology and Medical Informatics, University of Geneva, Geneva, Switzerland
• 
  Anastassia Shaitarova, PhDDepartment of Computational Linguistics, University of Zurich, Switzerland
• 
  Emmanuelle Sylvestre, MD, PhDUniversité de Rennes, CHU Rennes, INSERM, LTSI - UMR 1099, F-35000, Rennes, FranceCHU Martinique, Centre de Données Cliniques, F-97200, Martinique, France
• 
  Elsa Cécilia-Joseph, PhDUniversité de Rennes, CHU Rennes, INSERM, LTSI - UMR 1099, F-35000, Rennes, FranceCHU Martinique, Centre de Données Cliniques, F-97200, Martinique, France

Publication & development

CINEX was developed following an established guideline methodology: a three-round electronic Delphi (eDelphi) study with domain experts, followed by an in-person consensus meeting.

Because CINEX was produced by a consensus exercise, its own reporting follows the ACCORD reporting guideline. The completed checklist is on the ACCORD checklist page.

Get started