Mini Section:
TRIPOD Statement
Key Idea: What is the TRIPOD Statement?
The Transparent Reporting of Multivariable Prediction Models or Individual Prognosis or Diagnosis (TRIPOD) Statement is a disclosure method for medical prediction models (diagnostic and prognostic) comprised of 22-items which disclose details of a model's development, validation, and improvement.
While not directly developed for ML models, portions of the disclosures suggested for multivariable statistical models may be applicable to some supervised ML models for medical research.
Optional Reading: The Concept Paper
Review the TRIPOD Statement concept paper written by Collins et al.
Origins:
In 2011, 24 expert statisticians, epidemiologists, methodologists, clinicians, and medical journal editors met at the University of Oxford to conceptualize how to improve the reporting of multivariable prediction model development, validation, and improvement. This convening led to the development of the TRIPOD Statement, which was published simultaneously in 2015 by 11 endorsing journals. One additional journal published another paper further elaborating on the TRIPOD statement.
Cite as: Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): The TRIPOD Statement. Ann Intern Med. 2015;162(1):55-63. doi:10.7326/M14-0697
Explore: TRIPOD Statement Website
The TRIPOD Statement website provides an overview of the TRIPOD Statement's scope, history, and usage.
Review the website and take notice of the TRIPOD downloads and external resources made available.
Explore: TRIPOD Checklist
See the TRIPOD Checklist for Prediction Model Development and Validation embedded here.
Individual checklists for model development and validation separately are available on the TRIPOD Statement website.
Tripod-Checklist-Prediction-Model-Development-and-Validation-PDF.pdf