Hello!
The Q2 project is on Healthcare. I’ve created wiki page here. Please add to it as you can.
We’ve not had a healthcare specific quarterly project since 2020, although we have had Defibrillators more recently. What do you think would be good metrics (if any) for this?
If you’re wanting to get involved in this Quarterly Project, a good place to start might be to have a look at your local area in my comparison tool at https://osm.mathmos.net/healthcare/progress/.
Click through to your postcode area and zoom in the map. The overlay features are as follows:
Dark Green Blobs/Lines - matches between OSM and an official register entry based on a ref value added in OSM.
Light Green Blobs/Lines - matches between OSM and official register entry based on location. You might like to check if the match is correct. (If not, maybe mapping and adding the ref to the right object might fix things.)
Blue Circles - Entries in an official register with no matching OSM object. You might like to see if the facility is already mapped but needs its tags improving so the tool picks it up. If not, perhaps it could be mapped after a survey?
Red Circles - OSM objects that aren’t matched to an official register entry. Perhaps the object has been mis-tagged, or perhaps its since closed. A ground survey might help. (It might also be that it’s correctly mapped and tagged, but this type of facility doesn’t have an entry in the registers that I’ve used.)
Yellow Circles - Unmatched OSM objects where we wouldn’t necessarily expect a match. These can be ignored.
Here’s something else people might like to try on this project. This Overpass Turbo query finds objects tagged with amenity=social_facility that don’t have a social_facility=* sub-tag. There’s currently 819 objects found. Quite a few of these will probably be care homes of one sort of another.
Can you help by reviewing some of these objects and adding appropriate social_facility=* subtags? See Key:social_facility - OpenStreetMap Wiki for the common values. You might be able to find out what the object is from its name or other existing tagging, by finding an official website, or from one of the Healthcare sources in my tool*.
(* All the sources in my tool except the ‘Regulation and Quality Improvement Authority’ in Northern Ireland have their data available under the OGL, and so can be used in OSM. But I’d stay clear of using the addresses, as there’s an exemption for 3rd Party Rights, and they might be contaminated by OS / Royal Mail IP.)
Are there any tags for private/commercial clinics that are relevant to use in the UK?
I’ve particularly found places in shopping centres like “Thérapie Clinic” which offers a variety of cosmetic treatments, and more health-related places like “Medical Foot Clinic Dundee”.
I’ve been enjoying using this tool in the E postcode for the last couple of weeks. Great stuff.
I have a question about the relationship between the two data sources - the CQC and the NHS, which get tagged as e.g. ref:GB:cqc_location=1-541836459 and ref:GB:nhs_ods = F84047 respectively. Some institutions appear on both lists and have both tags.
In the case above, way 868479127 is tagged with ref:GB:cqc_location=1-541836459 already, but the system is showing a light green blob because it doesn’t have the other tag.
That means that to get all the blobs dark green, one has to go around twice, adding the CQC tag, then the NHS tag (a week later following the data getting updated). Is there a way the tool could identify both up front?
Some of the big providers - NHS, Bupa Dental, Damira Dental, Thérapie Clinic, are in the NSI so iD will suggest the key tags as you go along. (Try adding operator=NHS or Bupa… it’ll fill in the wikidata ref and set operator:type=)
The CQC dataset that Robert’s tool links to will tell you who the operator is (a company, a public org, an individual doctor, etc.). You could set operator:type=private if it seems to be private.
As an aside Thérapie Clinic is identified in NSI as a beauty shop. If some of them are actually healthcare facilities (e.g. in the CQC dataset), then you might need to change the suggested tags to e.g. dermatology or whatever it is.
(NB, you can hit F3 in JOSM to get to the NSI suggestions - but I think it’s a lot less fliud than iD for this particular task)
Unfortunately, the way I’ve set up the database tables, each entity can only have a single ref value associated with it. For most of the CQC listings that’s ref:GB:cqc_location=* . But for GPs I’ve used ref:GB:nhs_ods=* as there are additional ODS codes for branch surgeries, and these don’t have separate CQC codes. In the tool, the ref value and key it recognises for that object is shown on the popups on the map and in the data tables below - so you should be able to do it in one refresh cycle.
Hopefully this won’t have much impact, but I’ve just discovered that NSI contains several NHS trusts as operators for amenity=hospital. I’ve no idea why, as they’re unlikely to operate enough sites to justify inclusion.
Unfortunately, someone helpfully added operator:type=private to these and I’ve seen a couple of instances where people blindly accepted it as a tag “upgrade” from iD despite it being obviously wrong.
After filming that, several of the notes (left for myself) have been resolved by local mappers, to my delight. So I’ve added more notes of missing dentists/doctors. They’re still in areas I will get to, but it will be wonderful if other locals get to them in the next 2 months.
Apart from adding previously unmapped healthcare facilities, another thing to do is to improve the tagging of those that are already mapped. The table of postcode areas below the map at https://osm.mathmos.net/healthcare/progress/ includes the percentage of OSM healthcare objects that have names, websites, and postcodes tagged. If you click through to an area, and look at the two tables with OSM objects (e.g. Matched healthcare facilities in the IP postcode area) you can see which are missing any of these tags. The names and postcodes are show if tagged, and the name is a link if there’s a website tagged.
If there’s a missing website, there may be a website defined in the registration data, though sometimes it’s for the operator rather than the specific facility (the latter is preferable) so you might need to navigate through the site to find the facility-specific page. If not then a web search might find a website you can add.
For missing names, if the facility has been matched, then this might be the name. But the matching isn’t perfect, so do do a santity check or try to find a corroborating source.
For missing postcodes, my tool provides a link that takes you to a view of Code-Point open, which highlights the nearest postcode centroid. This may or may not be the actual postcode of the facility, so use this with care.
I’ve been taking part in this. Many thanks to Robert_Whittaker for his excellent website. I have a quick question. In my city, it’s common to have one medical centre but with more than one practice e.g. the Red Practice and the Yellow Practice. The best way I can think to map this is, the building itself with the name, street etc, as a way. The practices as nodes within the way, the nodes would include amenity=doctors etc etc.
@Jack_Regan For cases like that, I’d tag the building as healthcare=centre and then add nodes inside it for any specific mappable healthcare facilities located there. I’d add the address of the centre to all the objects, with an additional addr:housename=“XXX Health Centre” to the nodes inside if the centre has recognisable name. I wouldn’t put the nodes on the outline way of the centre, since the surgeries aren’t centred in the wall!
In the simpler case of a single practice with a pharmacy/dispensary, then I’d usually map the building and tag it as amenity=doctors and add a node inside for the amenity=pharmacy. Again, I’d add the address to both objects.
The “DD” area (Dundee) has been rising in the ranks of matched healthcare objects. I don’t venture out from the city, but other people have been helping.
Here’s the latest video of me hunting-down things to fix, and some chat about how the project has been going in this area.
There’s under a couple of weeks left for this quarter’s project now, so lets see what we can get done in that time. I’ve just updated the source data in my healthcare comparison tool to the latest available, so there may now be some more unmatched objects to check out near you.
We’ve got over 15k maxweight=* tags in the UK according to taginfo, almost all of which are incorrect. These indicate a restriction on the maximum actual weight of a vehicle, something for which signage in TSRGD 2016 does not exist. These restrictions may exist on private service roads and bridges, but are unlikely on the public road network.
What we actually have signs for in the UK are:
Diagram 622.1A “Goods vehicles exceeding the maximum gross weight indicated prohibited”, often with a plate with exceptions for access or loading (destination or delivery in OSM) maxweightrating:hgv=* + maxweightrating:hgv:condtional=*
Diagram 626.2A “Vehicles exceeding the maximum gross weight indicated prohibited from crossing the bridge or other structure”, sometimes with an exemption plate for (e.g.) empty vehicles maxweightrating=*
We also have a lot of maxweight tags added by a prolific vandal in 2015, who decided that adding fictitious weight and height limits all over the place was “Improving [the] road network”. I’m making good, but slow, progress at expunging his legacy in Greater London.
We’re also leaving an open goal for Overture and others by having so much incorrect data provided to routing software.