. Author manuscript; available in PMC: 2021 Feb 10.

Published in final edited form as: Ophthalmology. 2020 Sep 10;127(12):1596–1599. doi: 10.1016/j.ophtha.2020.09.009

Table 1.

Summary of Guidelines for Artificial Intelligence Studies

Name of Artificial Intelligence Extension	Purpose of Artificial Intelligence System^*	Study Design	Phase of Development or Testing of the Artificial Intelligence System	Status
Development and validation phase
STARD AI¹⁴	Diagnosis	Diagnostic accuracy	Testing the diagnostic accuracy of an AI system	In development
TRIPOD ML¹⁵	Diagnosis or prognosis	Studies developing, validating, or updating a prediction model	Development, validation, or updating of an AI system, or a combination thereof	In development
Testing and regulatory phase
CONSORT AI⁶	Any health intervention	Randomized trial (report)	Randomized trial report, results for the effectiveness of an AI system	Published online September 9, 2020, in the British Medical Journal, Lancet Digital Health, and Nature Medicine
SPIRIT AI⁷	Any health intervention	Randomized trial (protocol)	Randomized trial protocol for testing the effectiveness of an AI system	Published online September 9, 2020, in the British Medical Journal, Lancet Digital Health, and Nature Medicine

AI = artificial intelligence; CONSORT = Consolidated Standards of Reporting Trials; ML = machine learning; SPIRIT = Standard Protocol Items: Recommendations for Interventional Trials; STARD = Standards for Reporting of Diagnostic Accuracy; TRIPOD = transparent reporting of a multivariable prediction model for individual prognosis or diagnosis.

Table courtesy of Alastair Denniston and Xiaoxuan Liu.

For example, diagnosis, prognosis, therapeutic decision making, or risk stratification.