Measurement's "observationMoment" vs "date"

[M-casado]

Summary

Measurement's date and observationMoment overlap in the semantic space. Their use seems ambiguous.

Details

The date property, added on top of observationMoment (renamed from Phenopacket's time_observed), seems to overlap in use.

It's unclear to me what additional information date can provide compared to observationMoment. Unless, the definitions were swapped and the idea was for observationMoment to store the individual's age at the time of the measurement, and the date to contain the exact moment in time where the measurement was performed.

Term	Definition	Provided example (value)
observationMoment	Moment in time at which measurement was performed. RECOMMENDED. Renamed compared to GA4GH Phenopackets v2 Measurement.time_observed	"age": { "iso8601duration": "P53Y8M" }
date	Date of measurement. Addition compared to Phenopackets model.	"2017-05-03"
time_observed	The time at which the measurement was made.	timestamp: "2021-01-01T10:54:20.021Z"

I would suggest to either remove the property date, or re-write both properties' descriptions.

[mbaudis]

@M-casado If I remember correctly this was done since one rarely has an absolute date or time but frequently a relative one like an age (however the use of timeElement here is permissive). So mostly the "Renamed compared to GA4GH Phenopackets v2 Measurement.time_observed" is misleading since it is in fact a replacement.

So probably the best solution is to

• better define the observationMoment as something "age equivalent"
• remove the time_observed note
• replace date with timeObserved (type: string, format: date-time)

... as a first recommendation. This can be part of a general "Phenopackets alignment" cleanup¹.

Footnotes

1. Phenopackets v2 got through refinements while we were finishing Beacon v2; and many Beacon model parameters were adopted at this point, sometimes being modified due to different coding style and translation from the protobuf definitions etc. ↩

[M-casado]

@mbaudis I agree especially with the following:

however the use of timeElement here is permissive

I think the leeway offered by timeElement for the observationMoment is proper: either when the measurement was performed, or the age of the individual at the time.

I have mixed feelings with keeping date. Even with timeObserved it's a bit ambiguous for the user of the model, in my opinion. Given that one can encode both the timestamp of the measurement or the age of the patient at the time, I would personally just leave observationMoment alone.