Chapter 2

📖 The problem FHIR wants to solve

To fully utilize computers in healthcare, information must be in a digital format that machines can process automatically, without human intervention. This requires using standardized formats that are structured for machine readability.

Machine Readability

Even with the advancements in artificial intelligence that can read texts, having pre-structured data remains a significant advantage. With data in a standardized, structured format, a computer knows exactly how to process the information it receives. If information needs to be extracted from text first, ambiguities can arise, leading to errors or missing important details. Current large language models may produce different results across multiple attempts to extract data from the same text, which should be avoided at all costs. Structured, machine-readable data is reliable and ensures consistent results, making standardized data formats essential for health data exchange.

Artificial intelligence can indeed be a valuable tool in the process of packaging information into these standardized formats. In general, healthcare practitioners are trained to think and document in a highly structured manner, so even a letter from a physician presented as plain text is likely to be far more organized and formalized than a random piece of text.

Making data machine-readable is crucial for automating processes, analyzing data, and enabling efficient data exchange between different systems.

Interoperability is the ability of different systems, software, or devices to exchange information and use it with a shared understanding. In healthcare, the goal is for authorized people and systems to use information such as patient records or measurements safely and consistently, even when different software is involved. A standard such as FHIR supports that goal, but it cannot eliminate misunderstandings or errors by itself.

Example 1: Fahrenheit vs. Celsius

A lack of interoperability can lead to significant problems, even with something as simple as temperature measurement. For example, in the United States, body temperature is measured in Fahrenheit, where a normal reading is 98.6°F. However, in Europe, body temperature is measured in Celsius, and a normal reading is 37°C. A misunderstanding between these two scales can lead to serious errors, as 98.6°C or 37°F would indicate a situation incompatible with life. Without standardized systems, such basic discrepancies can cause confusion and potentially dangerous outcomes in medical care.

Example 2: Microsoft Office vs. Libre Office Impress

Another situation you have probably experienced is this: imagine you create a presentation in Microsoft PowerPoint and want to share it with a colleague who uses an open-source software like LibreOffice Impress. If Microsoft PowerPoint and LibreOffice Impress don’t fully support each other’s features, issues can arise: the formatting might be off, animations may not work as intended, or some elements might be missing altogether. This lack of compatibility can lead to miscommunication and extra work to fix the presentation. Ensuring systems can talk to each other correctly is essential to avoid these errors and keep the content consistent.

Developers of healthcare standards aim to make different systems more interoperable. The standard provides a shared foundation; real integration also depends on agreed meaning, workflow, security, governance, testing, and reliable software.

Consider a patient in a hospital who requires a chest X-ray. Think about all the steps in which information must be generated and/or exchanged. Think about your own experiences in healthcare: how were these steps performed? Here is a suggestion of what the key steps at which clinical information is created or exchanged may be:

  1. The physician documents the X-ray request, patient details, and clinical notes.
  2. The request is sent to the radiology department.
  3. Radiology schedules the procedure and informs the physician and patient.
  4. The patient undergoes the X-ray.
  5. The radiology team processes the images and prepares a report.
  6. The images and report are sent back to the requesting physician.
  7. The physician reviews the results, records the findings, and decides on the next steps.

Nowadays, dedicated computer systems and software are used to perform this kind of information exchange between departments and to store data for documentation.

An average-size hospital has a complex digital infrastructure to coordinate all the different bits and pieces, from encounters to procedures and the communication between different departments.

From your own experience, you might have seen how things can go wrong when exchanging medical information between practitioners, departments or encounters. Oftentimes, errors arise when different players use different systems that are not fully compatible with each other.