Clinical Informatics

Clinical informatics is the branch of health informatics that concerns the use of information and communication technology in the direct care of patients.

11 frameworks, from 1968 to today · Health Informatics · Medicine & Health

still a working view ended · each leads to its entry below

Overview

Clinical informatics is the branch of health informatics that concerns the use of information and communication technology in the direct care of patients. Its practitioners—clinicians who also understand computing, and computer scientists who work inside health care—design, build, implement, and evaluate the systems that hold patient data, the standards that let those systems exchange data, and the decision aids that help clinicians reason with it1. What is at stake is concrete: whether a physician can see a patient's allergy history from another hospital, whether a drug interaction is flagged before the order is signed, whether a sepsis risk score reaches the bedside in time to matter. The field's history is the story of how medicine moved from paper charts and isolated computers to networked records and algorithmic advice, and of the hard lessons learned about why so many of those systems failed to deliver what their builders promised.

Paper Records and the First Clinical Computing

The substrate of clinical informatics is the medical record, and for most of medical history that meant paper. In the early twentieth century, hospitals and professional bodies worked to standardize the record itself—uniform forms, consistent terminology, and the professionalization of medical record librarians—so that patient data could be compared and audited rather than merely stored. The record was a legal and administrative document as much as a clinical one, and its fragmentation across departments was already a recognized problem.

The first electronic clinical systems emerged in the 1950s and 1960s, mostly in academic medical centers with access to mainframe computers. At the University of Utah, Homer Warner and colleagues built computer-based aids for diagnosing congenital heart disease from catheterization data in the early 1960s. At Massachusetts General Hospital, Octo Barnett's laboratory developed the MUMPS programming language and database system in the late 1960s, which became the technical foundation for hospital information systems for decades. These systems were departmental and local: a laboratory system here, a pharmacy system there, each solving one unit's problem.

The Problem-Oriented Medical Record

A conceptual turning point came from Lawrence Weed, a physician at Case Western Reserve University, who in the late 1960s proposed the problem-oriented medical record (POMR). Weed's insight was that the traditional narrative record buried clinical reasoning in prose and made it impossible to follow. His alternative organized the chart around an explicit list of the patient's problems, with each problem carrying its own subjective findings, objective data, assessment, and plan—the SOAP structure that became ubiquitous in medical education. Weed argued that this structure made the record a tool for thought and audit rather than a passive log, and that it could be computerized to link problems, data, and orders. The POMR spread widely through medical schools and shaped how generations of clinicians were taught to document, even where the paper form was never fully computerized.

The Computer-Based Patient Record and Early Decision Support

Through the 1970s and 1980s, a handful of institutions built comprehensive electronic records. The Regenstrief Institute in Indianapolis, under Clement McDonald, developed a medical record system that by the 1970s was generating automated reminders to physicians—for example, flagging when a patient's laboratory results suggested a needed change in care—and later demonstrated that such reminders could change clinician behavior. At the LDS Hospital in Salt Lake City, Homer Warner's group and later Reed Gardner and colleagues built the HELP system, which combined a clinical database with a decision-support engine and was used for applications such as antibiotic monitoring and, in the 1990s, computerized physician order entry. At the Beth Israel Hospital in Boston, the CareGroup system developed by a team including Charles Safran and others supported order entry and results review.

These systems shared a pattern: they were built in-house by clinician-computer scientists at institutions willing to invest heavily, and they demonstrated that electronic records with embedded logic could improve care. They also shared a limitation: they were expensive, idiosyncratic, and hard to replicate elsewhere. The Institute of Medicine's 1991 report The Computer-Based Patient Record: An Essential Technology for Health Care gave the field a name and a goal—a record designed to support clinical decisions, not just to store data—and argued that paper records were fundamentally inadequate for modern care.

Expert Systems and Their Limits

In parallel, the 1970s and 1980s saw a wave of medical expert systems built on artificial intelligence methods. MYCIN, developed at Stanford by Edward Shortliffe and colleagues in the 1970s, diagnosed bacterial infections and recommended antibiotics using a few hundred rules and a certainty-factor scheme for reasoning under uncertainty; in evaluations it performed comparably to human experts. INTERNIST-1, built by Jack Myers and Harry Pople at the University of Pittsburgh, tackled the broader problem of internal medicine diagnosis. The field of medical decision-making grew around these efforts, with researchers like Lee Lusted and others arguing that clinical judgment could be formalized and studied quantitatively1.

The expert systems were intellectually influential but rarely deployed in routine care2. They demanded laborious knowledge acquisition, could not easily access patient data from hospital systems, and offered advice that clinicians often found hard to trust or fit into workflow. The lesson—that a correct recommendation is useless if it arrives at the wrong moment or cannot be explained—became a founding principle of clinical decision support.

Standards and Interoperability

As systems multiplied, the inability of one to read another's data became the field's central practical problem. Health Level Seven (HL7) was founded in 1987 and produced its first messaging standard in the late 1980s; HL7 version 2 became the dominant way hospitals exchanged orders, results, and admissions data, though its flexibility meant implementations varied. For terminology, the Systematized Nomenclature of Medicine (SNOMED) developed from the 1960s work of Roger Côté in pathology into a comprehensive clinical vocabulary, and SNOMED CT emerged in 2002 from the merger of SNOMED and the UK's Clinical Terms project. Laboratory data found a standard in LOINC, created by Clement McDonald and Stanley Huff in the 1990s to give every lab test a universal code. Later, HL7 version 3 and its Reference Information Model attempted a more rigorous, model-driven approach, and the Clinical Document Architecture (CDA) specified how to exchange clinical documents. These standards were technical achievements, but adoption was uneven, and the gap between a standard's existence and its use in practice remained wide.

The Sociotechnical Turn

By the 1990s and 2000s, a generation of researchers had learned that the hardest problems in clinical informatics were not technical. Systems that worked in one hospital failed in another; order entry that improved safety in one setting disrupted work in the next. The field developed a sociotechnical perspective: health information technology succeeds or fails as a joint product of technology, people, workflows, organizational structures, and policy. Researchers such as Joan Ash, Marc Berg, and others documented how clinicians adapted, worked around, or abandoned systems, and how implementation was itself a design activity. The sociotechnical model of health IT, articulated in work by the informatics community in the 2000s and 2010s, formalized this into dimensions—hardware and software, clinical content, human-computer interface, people, workflow and communication, organizational policies, and external rules—that any evaluation must consider. Evaluation research matured alongside it, moving from simple before-and-after studies toward rigorous designs that asked not just whether a system worked but for whom, under what conditions, and at what cost.

Meaningful Use and the National Push

The decisive policy turning point came in the United States with the Health Information Technology for Economic and Clinical Health (HITECH) Act of 2009, which created the Medicare and Medicaid EHR Incentive Programs—the "meaningful use" program. It paid hospitals and eligible professionals to adopt certified electronic health records and to demonstrate that they were using them in specified ways, such as recording structured problem lists, e-prescribing, and reporting quality measures. Adoption, which had been slow and concentrated in large academic centers, rose sharply within a few years. The program had real effects and real critics: it accelerated digitization but also drove a wave of implementations optimized for attestation rather than clinical value, and it exposed how much of the field's knowledge about workflow and usability had not been translated into vendor products. The policy debate that followed—about documentation burden, interoperability, and whether the incentives rewarded the right things—shaped the next decade of the field.

Predictive Analytics and the Data-Rich Era

With records now digital and networked, the field turned toward using the accumulated data. Clinical predictive analytics—statistical and machine-learning models that estimate a patient's risk of deterioration, readmission, or other outcomes—moved from research prototypes toward deployed tools. Early warning scores for sepsis and deterioration, built on routinely collected vital signs and laboratory values, were implemented in hospital systems and studied for their effect on outcomes. The methods drew on the broader rise of machine learning, but clinical informatics contributed its own hard-won cautions: models trained on one institution's data often failed elsewhere, alert fatigue could blunt the effect of even accurate predictions, and the question of how a prediction changes clinician action—not just whether it is accurate—became central. The field's evaluation tradition and its sociotechnical awareness were applied to algorithms as they had been to order sets and reminders.

A Medical Subspecialty

As the work grew more specialized, its practitioners sought formal recognition. In the United States, clinical informatics was recognized as a medical subspecialty, with the American Board of Preventive Medicine administering the first certifying examination in 2013, following approval of the subspecialty by the American Board of Medical Specialties in 20111. Certification pathways were created for physicians from a range of primary specialties, and fellowship training programs emerged1. This professionalization reflected a shift: clinical informatics was no longer a scattering of interested physicians and programmers but a recognized discipline with its own body of knowledge, training expectations, and standards of practice.

The Present

Today clinical informatics works across the layers its history built. Electronic health records are near-universal in high-income hospital settings, and the field's attention has moved to making them usable, interoperable, and safe—through standards such as FHIR (Fast Healthcare Interoperability Resources), which since the 2010s has offered a more web-oriented approach to exchanging data, and through regulatory pressure for information blocking to end. Decision support has moved from rule-based reminders toward predictive models embedded in workflow, with the same questions of alert fatigue, explainability, and equity that earlier systems raised. Evaluation remains a core method, now applied to algorithms and to the organizational arrangements that surround them. The open questions are recognizably descended from the field's founding ones: how to get the right information to the right clinician at the right moment, how to make systems that fit the work rather than fight it, and how to know whether any of it actually helps patients.

References

  1. Health informatics (Tools to support practitioners, United States, Competencies, education and certification), Wikipedia
  2. Artificial intelligence in healthcare (History), Wikipedia

The frameworks of Clinical Informatics

Each framework is a way of working the field has used: when it held, and the question it answers.

  1. 1968-2005

    Computer-Based Patient Record

    How can a patient's full record be held and used electronically?

    Fly to its tower in Nousfarer

  2. 1968-1990

    Problem-Oriented Medical Record

    How should a medical record be organized to support clinical reasoning?

    Fly to its tower in Nousfarer

  3. 1972-1995

    Medical Expert Systems

    Can computers diagnose disease as well as experts?

    Fly to its tower in Nousfarer

  4. 1975-Present

    Clinical decision support system

    How can computers help clinicians make better decisions at the point of care?

    Fly to its tower in Nousfarer

  5. 1985-Present

    Informatics Evaluation Research

    Does health information technology actually improve care?

    Fly to its tower in Nousfarer

  6. 1987-Present

    Health Information Exchange Standards

    How can separate health systems exchange patient data?

    Fly to its tower in Nousfarer

  7. 1991-Present

    Electronic health record

    What should a comprehensive electronic patient record contain and do?

    About Electronic health record
    At its core, the Clinical Information Systems and Electronic Health Records framework addresses a deceptively simple but historically transformative problem: how to capture, organize, and retrieve clinical data in a structured, computer-based form that supports both patient care and medical inquiry. Before this framework took hold, patient data existed primarily on paper charts, scattered across departments, often illegible, and nearly impossible to aggregate for systematic analysis. This framework established the foundational layer of clinical informatics by creating the infrastructure—the electronic record itself—that all later analytical and decision-support frameworks would depend on.
    To understand this framework, students need to orient themselves around a few essential conceptual clusters. The core organizing concept is the Computer-Based Patient Record, the digital successor to the paper chart. This record is built around a Clinical Data Repository, a centralized database that stores patient information from multiple sources. The data in this repository is structured through Structured Clinical Documentation, which replaces free-text notes with coded fields, checkboxes, and standardized forms. This structure enables the creation of a Longitudinal Health Record—a patient's medical history that persists across encounters, not just a visit-by-visit snapshot. Achieving this requires careful Clinical Workflow Integration, meaning the system must fit naturally into how clinicians already work, rather than forcing them to adapt to the machine. The primary functions of these early systems were Data Capture and Organization, accomplished through thoughtful Clinical Database Design that balances normalization (for integrity) with performance (for real-time clinical use). A critical usability component is the User Interface for Clinical Data Entry, which must minimize cognitive load while ensuring completeness and accuracy. The contents of the record typically include Patient Demographics and Encounter Data (who the patient is, when and why they visited) and Laboratory and Pharmacy Systems Integration (connecting results and medication orders directly into the record). Finally, the system must support Clinical Information Retrieval—the ability for clinicians to efficiently find the relevant data they need at the point of care.
    The history of this framework begins in the mid-1960s, when a handful of pioneering institutions recognized that computers could do more than process billing data. At the Regenstrief Institute in Indianapolis, researchers led by Clement J. McDonald sought to create a computer-based patient record that could streamline clinical workflow and simultaneously provide data for research. Their work produced the Regenstrief Medical Record System (RMRS), one of the earliest and most enduring clinical information systems. Around the same time, the LDS Hospital in Salt Lake City, under the leadership of Homer R. Warner, developed the Health Evaluation through Logical Processing (HELP) system. The HELP system was notable for integrating a clinical data repository with early Rule-Based Decision Support—for example, the MYCIN system at Stanford demonstrated how If-Then Rules derived from Medical Knowledge Encoding could produce Algorithmic Treatment Recommendations. While rule-based decision support eventually grew into its own distinct framework (Rule-Based Clinical Decision Support), it is important to note that it emerged alongside, and often within, the same early EHR systems. The key early contribution of the Clinical Information Systems framework was not the rules themselves but the underlying infrastructure of Observational Medical Data and the ability to structure, store, and retrieve that data reliably. By the 1970s and 1980s, frameworks such as Clinical Knowledge Representation began to formalize the ways clinical data and concepts could be modeled, while Clinical Data Standards and Interoperability (from the late 1980s onward) tackled the challenge of sharing data across systems. The early EHR pioneers wrestled with fundamental issues that remain relevant: System Interoperability Basics (how to get different systems to talk to each other), Data Quality and Completeness (ensuring the data in the record is accurate and usable), and the need for Clinical Data Standards (Local) before national or international standards existed. Over time, this framework's insights were absorbed into the mainstream of health IT, becoming the basis for enterprise EHR systems used worldwide. It coexists with its successor frameworks—providing the essential raw material for EHR-Based Predictive Analytics, Evidence-Based Clinical Decision Support, Learning Health Systems, and Clinical Artificial Intelligence and Machine Learning—but it never became obsolete; every subsequent framework still depends on the foundation that this one built.
    The 11 concepts its tower teaches
    • Clinical Data Repository
    • Clinical Database Design
    • Clinical Information Retrieval
    • Clinical Workflow Integration
    • Computer-Based Patient Record
    • Data Capture and Organization
    • Laboratory and Pharmacy Systems Integration
    • Longitudinal Health Record
    • Patient Demographics and Encounter Data
    • Structured Clinical Documentation
    • User Interface for Clinical Data Entry

    Fly to its tower in Nousfarer

  8. 1995-Present

    Sociotechnical Model of Health IT

    Why do health IT systems succeed or fail in real clinical settings?

    Fly to its tower in Nousfarer

  9. 2009-2018
  10. 2010-Present

    Clinical Predictive Analytics

    Can clinical data predict which patients will deteriorate?

    Fly to its tower in Nousfarer

  11. 2011-Present

    Clinical Informatics Subspecialty

    Should clinical informatics be a recognized medical specialty?

    Fly to its tower in Nousfarer

How the ideas connect

Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.

Computer-Based Patient RecordProblem-Oriented Medical RecordMedical Expert SystemsClinical decision support systemInformatics Evaluation ResearchHealth Information Exchange StandardsElectronic health recordSociotechnical Model of Health ITMeaningful UseClinical Predictive AnalyticsClinical Informatics Subspecialty

Beyond the field

Where Clinical Informatics's ideas came from in other fields, and what they led to there: the strongest 12 of 89 links.

A city in Nousfarer: a red-carpeted stair climbing terraces to a crowned tower on an acropolis, more towers round it, forest and snowy mountains beyond.
A city in Nousfarer, the game played in the browser.

Clinical Informatics is a planet in Nousfarer

Its city has a tower for each of these 11 frameworks. Land, walk in, and climb one concept at a time.

Land on Clinical Informatics

Related fields

Fields its ideas are linked to

More in Health Informatics

See where Clinical Informatics sits on the Map of Human Knowledge