Exploring electronic phenotyping for clinical practice in Norway
The data in EHR represent a limited view of a patient’s condition, and are by definition incomplete and often biased. Electronic phenotyping provides a more complete view of the condition. It uses a combination of clinical notes, International Classification of Diseases codes (ICD-10), medication lists, and laboratory tests. It enables direct identification of cohorts based on population characteristics, risk factors, and complications that have been demonstrated in similar populations, which can be used to conduct clinical trials and comparative effectiveness research. The cohort identification process can also be integrated with the EHR for real-time clinical decision support. This corresponds to secondary use of EHR data. Machine-learning approaches along with NLP methods are keys for electronic phenotyping.
The project aim is to gain knowledge on electronic phenotyping, including technologies and methods for its development, as well as identify its clinical relevance in Norwegian settings. This knowledge will be valuable for Helseplattformen and Directorate for E-health, providing prerequisites and demands for developing EHR systems.
The project activities will include among others knowledge gaining on electronic phenotyping through conferences and research visits to institutions working on this topic, identification of clinical needs for electronic phenotyping on the national level as well as data types required for phenotyping of the identified clinical needs and check the accessibility of the data, and phenotype modelling for the defined clinical use case(s).