Presented at AMIA 2022 Informatics Summit in Session 26 on March 23, 2022. Abstract: Computational phenotyping uses algorithms to identify which patients have a particular clinical condition. The phenotyping community is diverse, which has led to a proliferation of algorithms and methodologies. To date there are no commonly accepted best practices for algorithm development, evaluation, and reporting. Here we present a comprehensive review that considers all EHR-derived cohorts regardless of phenotype, methodology, or publication venue to inform new standards for reporting and evaluation of phenotyping algorithms. A total of 5,942 studies were assess for inclusion, ultimately identifying 342 studies that developed 661 algorithms to identify patient populations from the EHR. Some phentoypes have been the focus of many studies, with 43 different manuscripts describing the development of algorithms for type 2 diabetes alone. Not only does this make algorithm reuse difficult (which algorithm should you choose?), it represents significant duplication of effort and cost to the research enterprise.