Question about periodic interval events

Technical (archive)
1

Hi all,

As usual I’m reading the specs and have a question about periodic interval events.

I’m not sure how the period is calculated in a series. Let’s say we have interval events on a time line:

—E1.start___________E1.end----------E2.start____E2.end-------…

Note: interval events can have different durations.

Is the period calculated from E1.start to E2.start or from E1.end to E2.start?

This is of course to know when E3 should start.

Thanks!

2

As a doctor I would say it depends on what E is, medically.

Karsten

3

“Ex” is just an interval event, “interval” in terms of the openEHR specs.

Here clinical context is not relevant, the point is how time calculations are done to use the period (HISTORY.period http://www.openehr.org/releases/RM/Release-1.0.2/docs/data_structures/data_structures.html#_history_class).

But if you think this has to do with the medical interpretation, please elaborate. Maybe I’m missing something.

Thanks.

4

When one assumes that the all events are the same kind of events but occurring at various points in time,

Then the period can be expressed as Frequency (n/T) or equally as Period n/F).

Where n=number of events and T= total time from begin of the first event till the last (E.1 - E.n)
Clinical example:
Heart rate = 120 beats per minute.

Or as Period (T=n/F) expressed as time units that is the time between two events.
Clinical example:
Period between heart beats = 60/120 = 0,5 seconde. I.E the time between E(n+1) - E(n)

In your example
Period = 1 divided by Time (E2.start - Time E1.start),
or, Period = 1 divided by Time(E3.start) - Time(E2.start)
or, Period = 2 divided by Time (E3.start - Time E1.start)

Where in the OPENEHR specs can I find Frequency?
When we document we talk about numbers but often in qualitative terms?
What is the complete list of terms to describe these things using a classification?

Gerard Freriks
+31 620347088
gfrer@luna.nl

Kattensingel 20
2801 CA Gouda
the Netherlands

5

Hi Gerard,

6

See below

GF

Gerard Freriks
+31 620347088
gfrer@luna.nl

Kattensingel 20
2801 CA Gouda
the Netherlands

Hi Gerard,

Any Event (process) has an end date and start date an can be small but never zero.
End dates are always different from start dates.
So what do you mean?

Its meaning can be several things.
It can be the rate (frequency) at a real point in time. Time=T e.g. now, yesterday.
Or it can be the average over a period of time.e.g. the last minute,the last 10 minutes, the last month.
Actually it is a little bit more complex.

Frequency at point in a Unit Time.

INTERVAL–EVENTs as interval between events measured over a period of time.

What is NOT defined?

7

Hi Gerard,

8

> Any Event (process) has an end date and start date an can be small but
> never zero.
> End dates are always different from start dates.
> So what do you mean?

I'm talking about openEHR events by the definition of the specs, not trying
to come up with another definition. In openEHR POINT_EVENT is the record of
an event that to the record is just a point in time, doesn't really matters
if in reality the event took 10 seconds to execute, like a BP reading. That
is not clinically relevant in most contexts.

So, if you say that it is not clearly defined in the specs

That is not
clearly defined in the specs (going back to my original question).

it should a) get defind and b) one would, for current
practical purposes, stick to the same moment in time inside
each in-event period, be it either start, end, or middle.

Given that real-world actions of the same type take variable
amounts of time, one would assume the start times to be "more
correct" and closer to the intended period than end times.

Eventually, the "middle" times distance will be closer to
what *actually* happened to the patient.

Karsten

9

Hi,

I perceive the need to change/improve the OpenEHR spec because of this discussion.

What is the interpretation by Thomas?

GF

Gerard Freriks
+31 620347088
gfrer@luna.nl

Kattensingel 20
2801 CA Gouda
the Netherlands

10

always from start to start, assuming there is periodicity.

11

That is the answer I’m looking for :slight_smile:

Task for the SEC: clarify that on the specs (I’ll check current baseline)

On that case, is it possible to have overlapping INTERVAL_EVENTs on the same series? Thinking of one event taking too long and overlapping the start of the next one because the constraint for period is 5 minutes and the INTERVAL_EVENT.width took 6 minutes.

Or it is stated somewhere that we consider INTERVAL_EVENT.width < HISTORY.period? This can be controlled at design time and/or in software.

Best,
Pablo.