Biomedical Informatics Theorem

Expressions of the Fundamental Theorem of Biomedical Informatics
Ariel Skelley / Getty Images

A theoretically-grounded definition of biomedical informatics (BMI) was lacking for a long time. To bring some focus to this scientific field, Charles Friedman, Ph.D., proposed the fundamental theorem of biomedical informatics. It states that “a person working in partnership with an information resource is ‘better’ than that same person unassisted.” Friedman’s theorem is not actually a formal mathematical theorem (which is based on deduction and is accepted as true), but rather a distillation of the essence of BMI.

The theorem implies that biomedical informaticians are concerned with how information resources can (or cannot) help people. When referring to a ‘person’ in his theorem, Friedman suggests that this could either be an individual (a patient, a clinician, a scientist, an administrator), a group of people or even an organization.

Furthermore, the proposed theorem has three corollaries that help define informatics better:

  1. Informatics is more about people than technology. This implies that resources should be built for the benefit of people.
  2. The information resource must include something the person does not already know. This suggests that the resource needs to be both correct and informative.
  3. The interaction between a person and a resource determines if the theorem holds. This corollary recognizes that what we know about the person alone or the resource alone cannot necessarily predict the result.

Friedman’s contribution has been recognized as defining BMI in a simple and easy-to-understand way. However, other authors have suggested alternative viewpoints and additions to his theorem. For instance, Professor Stuart Hunter of Princeton University emphasized the role of the scientific method when dealing with data. A group of scientists from the University of Texas also advocated that the definition of BMI should include the notion that information in informatics is ‘data plus meaning’. Other academic institutions provided elaborate definitions that recognized the multidisciplinary nature of BMI and focused on data, information, and knowledge in the context of biomedicine.

Expressions of Friedman’s Fundamental Theorem  

It is useful to consider expressions of the theorem in terms of the people or organizations that would use the information resources. Whether the theorem holds true in a given scenario can be empirically tested with randomized controlled trials and other studies.

Below are some examples of how Friedman’s theorem could be applied in the context of current health care from the perspective of different users.

Patient Users

  • A patient using a medication reminder app will be more adherent to her medication regimen than the same patient not using the app.
  • A patient trying to lose weight who tracks diet and exercise on a smartphone app will lose more weight than the same patient without the app.
  • A patient who uses a patient portal to communicate with his physician will feel more engaged in his care than the same patient without the portal.
  • A patient who uses a patient portal to view tests results will express higher satisfaction with her care than the same patient without the portal.
  • A patient who participates in an online forum for rheumatoid arthritis will cope more effectively with her disease than the same patient without the forum.

Clinician Users

  • A pediatrician using an electronic health record (EHR) with vaccination reminders will be more likely to order timely vaccinations than the same physician without the reminders.
  • An emergency medicine provider with access to a local health information exchange (HIE) will order fewer duplicate tests than the same provider without the HIE.
  • A nurse who uses a wireless system to transmit vital signs directly into the EHR will make fewer documentation errors than the same nurse without the wireless system.
  • A case manager using a patient registry will identify more patients with uncontrolled hypertension than the same case manager without the registry.
  • A surgical team using a safety checklist will have fewer surgical site infections than the same surgical team without a checklist. (Note that the checklist is an example of an information resource that does not need to be computerized.)
  • A physician using a clinical decision support (CDS) tool for antibiotic dosing is more likely to prescribe the appropriate antibiotic dose than the same physician without the CDS tool.

Health Care Organization Users

  • A hospital with a computerized deep venous thrombosis (DVT) risk assessment program in the EHR will have fewer DVTs than the same hospital without the program.
  • A hospital with a mobile computerized physician order entry (CPOE) platform will have fewer telephone orders than the same hospital without mobile CPOE.
  • A hospital that uses an HIE to send discharge summaries to primary care providers will have fewer readmissions than the same hospital without the HIE.
  • A nursing home using sensor technologies will have a lower rate of patient falls than the same nursing home without the sensors.
  • A student health clinic that sends text message reminders will achieve higher vaccination rates for human papillomavirus (HPV) than a clinic without the text messaging system.
  • A rural health clinic using telemedicine for virtual consultations with specialists will send fewer patients to the emergency room, compared to the same clinic without telemedicine.
  • A medical practice with a quality improvement dashboard will identify gaps in healthcare provision more rapidly than the same practice without the dashboard.

    The Latest on Biomedical Informatics

    Sometimes biomedical informatics studies complex problems that can be difficult to capture. This field includes a broad spectrum of research, ranging from evaluations of organizations to genomic datasets analyses (e.g. cancer research). It can also be used to develop clinical prediction models, which are being supported by electronic health records (EHR). Two scholars from the University of Pittsburgh, Gregory Cooper and Shyam Visweswaran, are currently working on designing clinical prediction models from data using artificial intelligence (AI), machine learning (ML) and Bayesian modeling. Their work could contribute to the development of patient-specific models. Models which are now becoming crucial in modern medicine.

    Was this page helpful?
    Article Sources