LHS Journal Special Issue:  Computational Phenotyping

Guest Editor:  Dr. Vasa Curcin, King’s College London, UK  vasa.curcin@kcl@ac.uk

Editor in Chief:  Dr. Charles Friedman  cpfried@umich.edu

Submit Paper:  https://mc.manuscriptcentral.com/lhsjournal


Learning Health Systems require high-quality routinely collected Electronic Health Record (EHR) data to serve as input to novel machine learning and AI techniques and translate the outputs of these technologies into patient care and service improvement. To achieve this, the data in research databases need not only be of high-quality, but methods associated with its use need to be transparent and reproducible to ensure that any findings can be validated by the research community and generalised to the target population. At the core of this challenge is the ability to reliably identify clinically equivalent research-grade patient phenotypescapturing a particular disease variety, sets of comorbidities, medical histories, demographic profile or any other relevant patient-specific information – a process known as computational phenotyping.

The popularity of EHR data for research created an increased drive to document and share clinical phenotypes derived from research datasets in order to stimulate reuse, reduce variation in phenotype definitions across data sources and ultimately simplify and support the identification of clinically equivalent populations for research and healthcare applications. Reuse of existing phenotype definitions requires access to large sets of validated phenotypes, together with metadata needed to efficiently evaluate and implement them in a new research use case. Such phenotype collections have been used to identify traits and diseases for biomedical research and clinical care, recruitment for clinical trials, quality improvement studies, population-based health outcomes research, disease or drug safety surveillance, and genetic research.

Scope, Description, and More Information

This Special Issue of the Learning Health Systems journal invites submissions that feature original research on methods for developing reusable phenotype definitions, including computable representations, supervised and non-supervised algorithms, high-throughput phenotyping techniques, techniques for phenotype validation, and examples of successful phenotype repositories. A suitable submission must exhibit or discuss how the resulting approaches, techniques, or systems advance the concept of a Learning Health System. Interdisciplinary and applied research is especially encouraged.

Specific areas of interest include, but are not limited to:

  • Use of natural language processing for translating human-readable phenotype definitions (from patient notes or scientific papers) into machine-readable formats for high-throughput phenotyping
  • High-throughput phenotyping techniques for generating thousands of phenotypes with minimal human supervision
  • Workflow models for formalising phenotyping algorithm representation
  • Publishing phenotypes in a reusable form, including metadata, execution environments, software packages
  • Deep learning for feature selection in phenotype definitions
  • Addressing overfitting and local bias in supervised learning techniques
  • Integrating genomic and phenomic data in research studies
  • Models for representing longitudinal phenotypes with complex data relationships
  • Heterogeneous phenotyping including sensor, IoT and social network data
  • Statistical measures of confidence in the assignment of a computable phenotype to a person, e.g. calculations of a fit to an individual compared to the determinants of a computable phenotype.


  • Full paper submission deadline: January 15, 2020
  • Final author notification: Summer 2020
  • Expected publication: Autumn 2020 (see below)

The LHS Journal offers Early View wherein finished articles are made available before their actual inclusion in the issue. In Early View, articles are published online (including all figures and tables) and are fully citable and freely downloadable.

NB: In addition to the standard research paper format, authors are also encouraged to consider submitting Computable Knowledge papers which focus on a model, algorithm or another form of digital knowledge object. The LHS Journal welcomes two types of Computable Knowledge publications: CK-Enhanced and CK-Implementation, details of which can be found here.