Re-imagining the Digital Observation/ TPR Chart

[Wai Keong]

Everyday, tens of thousands of patients get their blood pressure, heart rate, respiratory rate, temperature and Oxygen Saturations recorded. This is now captured digitally in a few select hospitals. Current systems however, are highly unimaginative in the way this data is displayed (they try to recreate the paper experience, badly!).

My challenge for this Hackday is to prototype and create a data visualisation method to enable

+ The ability to view 1 weeks worth of observation data on a a smartphone sized screen

+ Reuse this design to integrate into electronic patient list and patient dashboards

+ Ability to simultaneously view 10 patients x 1 weeks worth observation data in one desktop sized screen to get a 'birds-eye' view of the condition of my patients.

This idea was inspired by an article written by the father of Data Viz, Edward Tufte, in a Lancet Article published in 1994

http://www.edwardtufte.com/bboard/q-and-a-fetch-msg?msg_id=0003mm

Rob Dyke is going to provide the 'raw' anonymised observation data from real patients for us to *play* with. These were created using the Wardware platform that Kings Hospital is using.

So I guess I'm looking for

+ Data Viz experts

+ Designers

+ Coders 

+ Clinicians

Who wants to join me in maximising the data-to-ink(pixels) ratio? ;)

Wai Keong

[Ian McNicoll]

Hi Wai-Keong,

I would be keen to help with some clinical data modelling. @RobD how
are you providing the data - some kind of feed from a Wardware test
env. or as an export file?

Ian

> --
> You received this message because you are subscribed to the Google Groups
> "nhshackday" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to nhshackday+...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>



--
Dr Ian McNicoll
office +44 (0)1536 414 994
fax +44 (0)1536 516317
mobile +44 (0)775 209 7859
skype ianmcnicoll
ian.mc...@oceaninformatics.com

Clinical Modelling Consultant, Ocean Informatics, UK
Director openEHR Foundation www.openehr.org/knowledge
Honorary Senior Research Associate, CHIME, UCL
SCIMP Working Group, NHS Scotland
BCS Primary Health Care www.phcsg.org

[VJ]

Count me in.

Sounds like fun.

 

VJ

--

[Rob Dyke]

Mysql file / CSV.

[Rob Dyke]

Hi WK, good hack project. Shame I can't be there that weekend :(

A few thoughts on this.

Something similar was hacked at a recent RWS event ... a hack called MEDIC. After the ribbing I got from some of the team on Twitter about 'openness' I was expecting to find some code online to point you at, but instead I found a wordpress theme... (any other links people?) Anywho... The team took a vast feed of data from bedside monitoring equipment and plotted results in a simple dashboard. Looks great. Good data-ink ratio there.

The thing with obs is that the frequency of them is not uniform. One patient may have 3 observation sets taken in a 24hr period, others may have 48 sets. This variance is driven by some sort of Track/Trigger algorithm, for example the RCN/RCP National Early Warning Score.  Score 0 on that scale and you'll be on 6 or 8 hourly obs. If a patient begins to deteriorate the obs frequency will increase. This creates a challenge for data recording and presentation that the paper-world doesn't easily adapt to, yet computers can cope with much easier.

Here are two chats, one patient on 6hrly obs, another on 30min obs.

Varying sample intervals create display problems .... Clinicians need consistency of interface so that 'at a glance' the data can be understood. With fixed presentation style and axis the data presented can be easily understood. These charts show T, P/HR, BP, SpO2%, RR, and EWS in fixed positions. The X and Y are also fixed. Everything is shown with the same 12hr time interval marked on the chart.

Happy to kick this around more on the list.

BW, Rob

[@bluepill979]

We are planning to work on charting functionality for our iNEWS app, perhaps we can work together on this project

[Wai Keong]

Sounds good. What's iNEWs? Is it an early warning score type app?

On 15 May 2013 21:53, @bluepill979 <tony...@hotmail.com> wrote:

We are planning to work on charting functionality for our iNEWS app, perhaps we can work together on this project

[Martin Green]

The graphs in the article look good, although data to ink could be increased with more use of colour.  I have a little experience of plotting data in javascript and could produce graphs like them, if that helps.

See you,

Martin

On 14 May 2013 23:13, Wai Keong <wongwa...@gmail.com> wrote:

[Neville Dastur]

A couple of observations

On 16 May 2013, at 00:23, Martin Green <martin...@gmail.com> wrote:

The graphs in the article look good, although data to ink could be increased with more use of colour.  I have a little experience of plotting data in javascript and could produce graphs like them, if that helps.

See you,

Martin

On 14 May 2013 23:13, Wai Keong <wongwa...@gmail.com> wrote:

Everyday, tens of thousands of patients get their blood pressure, heart rate, respiratory rate, temperature and Oxygen Saturations recorded. This is now captured digitally in a few select hospitals. Current systems however, are highly unimaginative in the way this data is displayed (they try to recreate the paper experience, badly!).

I suspect they do this as we (doctors and other healthcare staff) are used to looking at results like this. I can walk on to ITU and the 24 hour chart is easy to inspect. The wards however, have a multitude of different representations and time lines and we are forever complaining.

My challenge for this Hackday is to prototype and create a data visualisation method to enable

+ The ability to view 1 weeks worth of observation data on a a smartphone sized screen

Is this not just a smoothed average?? Over a week the individual data points matter little

+ Reuse this design to integrate into electronic patient list and patient dashboards

Patient dashboard displays are often different from smartphones. I.e they are bigger, so I don't think the solutions are that transferable.

+ Ability to simultaneously view 10 patients x 1 weeks worth observation data in one desktop sized screen to get a 'birds-eye' view of the condition of my patients.

Are you able to elaborate on the use case here. Why 10? Why 1 week. Why do the data points need representing? By the last statement what I mean is should this not be replaced by analysis. So a colour coded table with say an average systolic coded with traffic lights. So the representation is not limited to a fixed number. I just scroll through looking for red and then drill down to my poorly controlled BP patients.

Neville

--

Surgeons Net Education: http://www.surgeons.org.uk

Clinical Software Solutions: http://www.clinsoftsolutions.com

Find our free and paid apps on the iTunes Apple store and Android Google Play store

LinkedIn: http://www.linkedin.com/profile/view?id=49617062

[Rob Dyke]

"...a colour coded table with say an average systolic coded with traffic lights. So the representation is not limited to a fixed number. I just scroll through looking for red and then drill down to my poorly controlled BP patients."

See all the good work by the RCN / RCP on the NEWS score.

http://www.rcplondon.ac.uk/sites/default/files/documents/the-national-early-warning-score--a3-size-_0.pdf

Simples.

--
Rob Dyke

Director, Tactix4
Office: +44 (0)845 003 9159
Mobile: +44 (0)783 345 0022
Web: tactix4.com @tactix4 LinkedIn

---

From: "Neville Dastur" <neville...@gmail.com>
To: nhsha...@googlegroups.com
Sent: Thursday, 16 May, 2013 2:44:53 PM
Subject: Re: [nhshackday] Re-imagining the Digital Observation/ TPR Chart

[Neville Dastur]

On 16 May 2013, at 14:46, Rob Dyke <r...@tactix4.com> wrote:

"...a colour coded table with say an average systolic coded with traffic lights. So the representation is not limited to a fixed number. I just scroll through looking for red and then drill down to my poorly controlled BP patients."

See all the good work by the RCN / RCP on the NEWS score.

http://www.rcplondon.ac.uk/sites/default/files/documents/the-national-early-warning-score--a3-size-_0.pdf

Exactly a very nice example of colour coding a score. But the table I was suggesting was more like 

for each patient in Wai example.

Neville

[Damian Roland]

This thread is very important and timely. I was at the Rapid Response Systems Conference (http://www.rapidresponsesystems.org) earlier this week and it is clear that EWS and their interface with technology (of all kinds) is going to be increasingly be flavour of the month. However the cost of systems are huge and anything that can be done to allow the recording and sharing of data in a cost efficient manner will be welcomed by trusts I suspect. 

An interesting point made during the conference was that although there was a great deal of attention paid to the science of numbers (i.e what score predicts what), virtually no attention has been made to the science of design - apparently red isn't a great colour to use on observations charts...

Some tweets from the conference

http://storify.com/Damian_Roland/rapid-response-systems-rrs2013-day-one

http://storify.com/Damian_Roland/rapid-response-systems-rrs2013-day-two

Cheers

Damian

Dr. Damian Roland

NIHR Doctoral Research Fellow

Inaugural NICE Scholar

@damian_roland

E: damian...@me.com

M: 07727158213

On 16 May 2013, at 15:05, Neville Dastur wrote:

On 16 May 2013, at 14:46, Rob Dyke <r...@tactix4.com> wrote:

"...a colour coded table with say an average systolic coded with traffic lights. So the representation is not limited to a fixed number. I just scroll through looking for red and then drill down to my poorly controlled BP patients."

See all the good work by the RCN / RCP on the NEWS score.

http://www.rcplondon.ac.uk/sites/default/files/documents/the-national-early-warning-score--a3-size-_0.pdf

Exactly a very nice example of colour coding a score. But the table I was suggesting was more like 

<PastedGraphic-1.png>

[Wai Keong]

Great discussion and plenty of food for thought. 

I guess when I was imagining this design hack I wanted to play with idea on how to present a large amount of observation data in a limited space (hence the smartphone-sized screen example), Like the attached Tufte Lancet article.

The design should also allow the appreciation of trends and patterns (ie. spikes of temperatures, decreasing or increasing HRs). I think the human brain appreciate patterns more than specific numbers.

Early warning scores should be included but it only emphasises observations at a moment time. EWS is very important but this is not the point of this proposed design challenge.

I hope that we can come up with all sorts of wild and wacky ideas and nothing is off limits! So let's forget about what we are used to and imagine what's possible! 

Looking forward to it.

[Michelle Teo]

It would be useful to be able to capture (sort of a Cmd+shift+4) specific portions of the observations - and link them to a particular entry in the patient's ward round notes! To give a quick visual representation of what particular trend in the obs is worrying, or is reassuring? Just a quick thought!

Looking forward to the hackday! Wai Keong tells me its a crazy time!

Best wishes,

Michelle

[Wai Keong]

Some more reading material and food for though. 

http://s3.amazonaws.com/edwardtufte.com/MUIchapter7.pdf

[Wai Keong]

Tufte's Sparklines: http://www.edwardtufte.com/bboard/q-and-a-fetch-msg?msg_id=0001OR

I think the application of this data viz approach to this project might work. I've also found some Javascript Implementations of this here:

http://omnipotent.net/jquery.sparkline/#s-about

[Martin Green]

Wai,

I intend to have a go at implementing interactive javascript graphs as
described by Tufte. I am impressed by the way that he has manipulted
the scales on the axes such that vastly different data can be
displayed in the same way. With the first 10% of the time axes
showing ~two
years of data and the last ~90% showing ~one week. The y axes are
perhaps more impresive,
being split into five linear regions, of normal, raised/lowered and
very raised/lowered, ie. how scared the
clinican should be.

I want to try any make these graphs and have them interactive. ie,
hover over a reading to see detail. Perhaps incorporate the side bar
of events/ free text observations, and have that bring up a line on
all the other graphs to show when the event happened. Be able to do
some sort of rescaling of the time axis, and have all the graphs
change.

Although there is a risk that I will quickly become limited in my
ability to write javascript.

Regards,
Martin

>>>> Office: +44 (0)845 003 9159 <callto:+44%20%280%29845%20003%209159>
>>>> Mobile: +44 (0)783 345 0022 <callto:+44%20%280%29783%20345%200022>
>>>> Web: tactix4.com <http://www.tactix4.com/>
>>>> @tactix4<https://twitter.com/#%21/tactix4>
>>>> LinkedIn <http://www.linkedin.com/company/tactix4>
>>>>
>>>> ------------------------------
>>>> *From: *"Neville Dastur" <neville...@gmail.com>
>>>> *To: *nhsha...@googlegroups.com
>>>> *Sent: *Thursday, 16 May, 2013 2:44:53 PM
>>>> *Subject: *Re: [nhshackday] Re-imagining the Digital Observation/ TPR

>>>> Chart
>>>>
>>>> A couple of observations
>>>>
>>>> On 16 May 2013, at 00:23, Martin Green <martin...@gmail.com> wrote:
>>>>
>>>> The graphs in the article look good, although data to ink could be
>>>> increased with more use of colour. I have a little experience of
>>>> plotting
>>>> data in javascript and could produce graphs like them, if that helps.
>>>> See you,
>>>> Martin
>>>>
>>>>
>>>> On 14 May 2013 23:13, Wai Keong<wongwa...@gmail.com> wrote:
>>>>
>>>>> Everyday, tens of thousands of patients get their blood pressure,
>>>>> heart
>>>>> rate, respiratory rate, temperature and Oxygen Saturations recorded.
>>>>> This
>>>>> is now captured digitally in a few select hospitals. Current systems
>>>>> however, are highly unimaginative in the way this data is displayed
>>>>> (they
>>>>> try to recreate the paper

>>>>> experience<http://www.rcplondon.ac.uk/sites/default/files/documents/news-observation-chart-with-explanatory-text.pdf>,

>>>> Apple<http://itunes.apple.com/gb/artist/clinical-software-solutions/id522940810>
>>>> store
>>>> and Android Google
>>>> Play<https://play.google.com/store/apps/developer?id=Clinical+Software+Solutions>

>>>> store
>>>> LinkedIn: http://www.linkedin.com/profile/view?id=49617062
>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google
>>>> Groups "nhshackday" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send
>>>> an email to nhshackday+...@googlegroups.com.
>>>> For more options, visit https://groups.google.com/groups/opt_out.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google
>>>> Groups "nhshackday" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send

>>>> an email tonhshackday...@googlegroups.com.

>>>> For more options, visit https://groups.google.com/groups/opt_out.
>>>>
>>>>
>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google
>>>> Groups "nhshackday" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send
>>>> an email to nhshackday+...@googlegroups.com.
>>>> For more options, visit https://groups.google.com/groups/opt_out.
>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google
>>>> Groups "nhshackday" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send
>>>> an email to nhshackday+...@googlegroups.com.
>>>> For more options, visit https://groups.google.com/groups/opt_out.
>>>>
>>>>
>>>>
>>>
>>>
>>
>

[Wai Keong]

Thanks Martin,

Can't wait to see them. Also can't wait to see what other ideas others have. I'm sure that we can create something by Sunday. 

Now, just waiting for Rob Dyke to feed us some 'real' data ;)

WK

[Rob Dyke]

Sent --> @wai2k