The enormous growth of informatics in the past decade has resulted in a high demand for trained professionals. Different types of training programs have been developed to serve the full spectrum of our field, from theoretical foundations to practical applications. For example, clinical research data management requires certain professional competencies (p. 737), while clinical data management involves an overlapping, but not identical, set of skills (p. 832). The goal of informatics education is to train a cadre of professionals who can design, develop, implement, and evaluate information-based interventions that impact health sciences and clinical care. These interventions target a variety of users and utilize diverse approaches and technologies. Among them, electronic health records (EHRs) continue to be a critical source of research in informatics, with articles that describe how to identify gravely ill Social Security disability applicants from EHRs (p. 709), compare vital signs recorded on paper and electronically (p. 717), review digital interventions to improve cardiovascular health (p. 867), and analyze policies that promote safe and usable EHRs (p. 769), among many others.
Identifying adverse drug events (ADEs) from electronic health records is an important area of investigation in informatics. Authors report on the decline in in-hospital ADEs resulting from meaningful use of information technology (p. 729), show how Twitter posts can serve as input for deep-learning pharmacovigilance models to label ADEs (p. 813), and describe a system to detect ADEs from nursing notes and laboratory test results (p. 697). Related to medication safety, different studies compare knowledge bases used for detection of drug-drug interactions (p. 806), analyze medication order voiding in computerized provider order entry systems (p. 762), and describe a pharmacogenomics information resource for pharmacists (p. 822).
Any analysis based on electronic health data must rely on harmonized data and reproducible processes. Examples of such processes are a phenotyping algorithm for the Elixhauser Comorbidity Index (p. 845), a method for embedding nursing interventions into the World Health Organization classification (p. 722), and a system to automatically classify eligibility criteria in clinical trials (p. 781). Related to data organization are also a new implementation of MetaMap (p. 841) and a method to detect missing relations and concepts in SNOMED CT (p. 788).
Genomic information is increasingly being utilized in clinical care. We report on the decision-support needs of primary care pediatricians (p. 851) and a strategy for mitigating privacy risks in genomic data queries (p. 799). Decision-support tools are important to fill information gaps and to continuously educate health care providers and patients on various topics, eg, the information needs of generalists and specialists (p. 754), communication of patient health information to guide health information technology design (p. 680), shared decision-making using personal health record technology (p. 857), and systems to decrease unnecessary vitamin D testing (p. 776) and to monitor mood (p. 746).
As exemplified in this issue of the journal and documented in JAMIA for over 20 years, informatics is a diverse and fascinating field that not only supports and augments the work of health scientists and clinicians, but also changes the way they approach problems. It is thus our responsibility to ensure that our information interventions result in maximal societal benefit and are improved or replaced by better products generated by a new generation of informatics professionals.