A roadmap for a smart EHR. 2) Data Model

After explaining why does HL7 FHIR represent a suitable architecture for intelligent FSE, let’s see how to set up the data model.

If a new EHR shall provide doctors with a rapid and complete view of a patient’s health state and to enable the implementation of intelligent services, which we will describe in the next article, then it is necessary to abandon the current approach, which sees the ESF as a mere container of documents.

At present, EHR contains the digital version of paper documents relevant to the clinical activity (hospitalization summary, emergency report), visit and examination prescriptions, guidance to the patient (discharge letter), examination results (diagnostic reports).

Instead of starting from the status quo, let’s try to overturn the reasoning starting from the objectives outlined above, and how to set up a data model for the smart EHR.

The following figure describes a data model based on some FHIR classes. Each patient may have one or more encounters, the contacts he has with different actors of the health system (family doctors, territorial structures, hospitals, etc.) in different ways (outpatient, inpatient, emergency accesses – first aid, home care, and so on).

An encounter can link clinical conditions, observations, prescriptions, interventions procedures, medications, diagnostic reports and images.

Adverse effects, intolerances or allergies can be detected in a patient, and health goals can be set. More encounters may constitute an episode of care or be part of a care plan.

But how can we feed this data model? There are two possibilities: direct or indirect mode. In the first case, information systems publish data in the EHR when certain processes are managed, such as the discharge of a patient, the issue of a prescription, access to the emergency room. This is the same logic used to create data files for information exchange towards regional systems; the only difference is the timing: instead of the cumulative way at the end of a period (month, quarter), it will be a real-time operation.

The indirect mode, on the contrary, foresees that data are taken from available digital documents. In addition to storing them as HL7 CDA files, they should be processed to extract detailed data and publish them in the above model.

However, the indirect mode could not grant the same detail than the direct one. The two modes can be used together, with a hybrid approach.

To be continued.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s