Cancer Phenotype Development: A Literature Review

Area of Science:

• Biomedical Informatics
• Health Services Research
• Oncology

Background:

• Electronic Health Records (EHRs) offer valuable data for identifying patient cohorts.
• Computable phenotypes derived from EHRs can streamline cohort identification for research and clinical applications.
• Targeting specific patient populations is crucial for drug development and disease interventions.

Purpose of the Study:

• To review and summarize the literature on the development and application of EHR-based phenotypes for cancer patient cohort identification.
• To identify common approaches and variations in phenotype development, validation, and implementation for cancer studies.

Main Methods:

• A PubMed literature survey was conducted using specific search criteria based on NIH guidelines.
• Identified studies were analyzed to determine the cancer types and the methodologies employed for phenotype development and validation.
• Included studies focused on breast, colorectal, ovarian, and lung cancers.

Main Results:

• Phenotype development and validation approaches varied across the reviewed studies.
• Four studies utilized chart review for phenotype development.
• Machine learning, ontological approaches, and natural language processing (NLP) were also employed.

Conclusions:

• EHR-based phenotypes are effective tools for identifying cancer patient cohorts.
• Diverse methodologies exist for developing and validating these phenotypes, including chart review, machine learning, and NLP.
• Further research can refine these methods for improved cohort identification in oncology.