A worldwide pandemic tends to make you re-visit your priorities …
When we first presented our People Like Me proposal for a secure and scalable data infrastructure and a patient-centric environment for collecting that data and for providing Actionable Insights based on that data to encourage PwPs to participate – in June 2019 in Kyoto – the ‘virus that changed everything’ had not yet made an appearance …
A year later, almost, we live in a world that is changing as we speak.
Suddenly remote assistance from medical professionals to patients, who cannot leave home, has become the ‘new normal’ …
In the recently published Ending Parkinson’s Disease … the authors observe …
“Telemedicine could also save Medicare money and provide patients better care—if only Medicare would cover it. The Veterans Health Administration has embraced virtual visits. Consequently, veterans throughout the country receive care remotely from Parkinson’s specialists and even mental health professionals. However, Medicare’s coverage of telemedicine is limited to clinical settings (e.g. hospitals) in “health professional shortage areas,” which are typically rural counties. Care provided in the home via telemedicine is not covered at all. In 2016, Medicare spent little—less than $30 million, or less than 0.01% of its nearly $600 billion budget—on telemedicine.”
Professor Bas Bloem is one of the authors of this book.
We met him at WPC in Kyoto and presented the Kode Health – People Like Me proposal to him.
Some of the work he has done with his Parkinsons’net initiative and the use of insurance claims data to show that achieving better outcomes at a lower cost was possible, was ground breaking …
Even then, in June 2019, there were clear signs that ‘remote care’ was on the agenda, but adoption and hence funding was an issue with medical practitioners – according to some of the people we spoke to – resisting for reasons of personal liability, among others …
I would expect this funding issue to get addressed quickly now …
A pre-Covid-19 review of healthcare trends by Deloitte, concluded a growing willingness to ‘… use technology to track and maintain their health …’
I recently received the following text message showing encouraging signs of early adoption …
P. D. Hinduja Hospital is facilitating a WhatsApp video call consultation with Dr. CHARULATA SANKHLA. To consult, book your appointment via our call centre 022 67668181 / 022 45108181 and provide your WhatsApp number. You will need to pay consultation charges online. The doctor will give you a WhatsApp video call at the scheduled time.
It will be interesting to see how we can support such remote consultations with a combination of structured and unstructured data …
In India, where this social media-based approach was initiated, patients typically physically carry their personal files to consultations and hospitals do not typically keep records … Precisely for this reason, patients should be in a strong position to retain control over their medical data …
As the researchers try to understand Covid-19 it becomes increasingly clear that access to more data is essential…
Having early-on expressed the view that in order to collect data about Covid-19 symptoms – some of them only now being recognised as relevant – we needed a ubiquitous, highly scalable and familiar platform – like Facebook for example or WhatsApp even – to get started quickly.
It soon became clear, that widespread lack of trust in social media platform providers being able to rise to the challenge and provide an environment where privacy settings and encryption could really be relied on, made that approach difficult.
Having said that, Google and Apple have now stepped up to provide an API to a secure, reliable and highly scalable contact tracing infrastructure which is highly likely to become the global standard it needs to be – even as some governments at this stage still seem to think that the virus can be stopped at the borders somehow …
So, we should not give up too soon doing what makes sense in technology terms …
With the state and employers adding their voice to insurers, practitioners and technology companies, the drive towards adoption can only accelerate.
Whilst continuing to focus on PD, we have decided to make ‘symptoms data’ under patient control our first priority …
We have always known that PLM had a wider potential than just PD and as we can now see a shift to remote management of long-term conditions like PD, it makes sense to make the underlying data model configurable for any condition and retain the principle of patient-centricity.
So this what the PLM prototype will focus on … a conversation with Alexa about symptoms and a spreadsheet-based or meta-data upload of basic medical history.
Based on a solid architecture, we will build in the APIs to initiate and capture unstructured data sources such as the video recordings of finger taps or voice samples or posture measurement by specialist cameras, but without initially focusing on the complexities of working with actual devices like an Alexa camera or voice recording, Google Fitbit biometrics or a specialist Microsoft Azure Kinect camera …
So, in terms of the structure of data around one symptom – pain …
A patient has Symptoms
One of the Symptoms is Pain
Pain has a Location, Type, Severity, date/time Period, Frequency and Impact.
‘Overall’ is a pain Location, as is left foot, left elbow tendon etc – patients can create new locations e.g. skin
Type of pain is a word e.g. dull, shooting, sensitive
Severity is a number associated with a chosen pain scale, which maps onto the PLM symptoms scale.
Date / time Period is a word – e.g. today, yesterday, last night, last week or an exact date/time if uploading pain metrics in real time.
Frequency is a word e.g. all the time, daily, regularly, from time to time
We will, of necessity, have to start with an existing widely recognised list of symptoms, using a dictionary of synonyms. Symptoms added under Patient control will need to be reviewed on a regular basis to tie them to a common term, where appropriate.
The team has suggested we adopt the WHO (World Health Organisation) ICD-10 (International Classification of Diseases – revision 10) standard codification. This is expected to help improved interoperability of EHRs (Electronic Health Records) systems.
The WHO clarifies the origin of ICD to be statistical.
The problem with adopting any specific classification standard lies in the fact that you then have no way to make changes without consensus – something which can constrain innovation …
That may be one reason why there are many country variations of ICD-10 already. Some are simply local language versions but some are more significant and depend on whether the original statistical focus remains or – for example – there is a focus on billing – as in the US.
A timely reminder not to forget the need to accommodate language variants of codified symptoms and their synonyms …
We have decided to maintain a PLM list, alongside an ICD one and allow for future – and past – versions of ICD as well as variants.
That should enable us to recognise real world entities such as condition, symptom, medicine and procedure and recognise any ´many to many´ relationships – in traditional E-R database design fashion – such as
- A ‘symptom’ can be associated with multiple ‘conditions’ with attributes ‘severity’, ‘frequency’ etc
- A ‘symptom’ can be caused by ‘medication’ as a side-effect or in combination with another ‘medication’
- A ‘symptom’ can be alleviated or caused by a ‘procedure’
- etc …
Looking at ICD-10 and into PD specifically, it is immediately clear – code G20 Parkinson’s – that there is a fundamental assumption that the most relevant symptoms are motor symptoms, I suppose because they are essential for clinical diagnosis …
You find ‘pain’ … under ‘general’
It seems, the hierarchical nature of ICD/10 can make it complex to represent many to many relationships … Drill down on joint pain – looking for wrist pain caused by Parkinsonian tremor – and you get …
ICD-11, due to be introduced in 2021, looks like a major change from ICD-10, which may have started to recognise these limitations …
With ´related´ conditions like dementia easier´ to find …
I had an opportunity to discuss these issues with Nick, a professional auditor of clinical codification. He confirmed that in most cases, a patient diagnosed with Parkinson´s would be recorded as just that and only in case of very specific issues would the available hierarchy of codes for symptoms and locations of pain be recorded. One of the additions in the UK ICD-10 versions deals with more granular locations and even then it does not provide for left-right granularity, as I understood.
He also confirmed that moving from ICD-10 to ICD-11 will be a big task involving massive retraining.
We clearly have a lot to learn … but I think the decision to go with a PLM list of symptoms – although counter-intuitive – will prove right in the longer term.
This ICD-10 PD coding tutorial concludes …
To paraphrase – sharing the patient’s story is critical …
We couldn’t agree more !
Based on this, the initial dialogue with Alexa will follow this sort of interaction model …
Let’s talk about People like Me
Hi Koen, how are you today.
Shall we talk about symptoms ?
The last time we talked about your symptoms was a week ago.
Have any of your symptoms changed ?
Shall I list your symptoms and you rate them same, better or worse ?
How much worse ?
You can use the PLM score or the King’s College pain score
Pain in your left shoulder
That is the same as before ?
Pain in your right shoulder
That is the same as before ?
Pain in your feet
This is a change from before
You reported this in our last chat as a new symptom.
When do you experience this pain ?
In the morning when I get up
In the morning ?
Does it impact you day ?
How is your day impacted ?
I have problems walking
Difficulty walking ?
Can I record this change and tell your neurologist ?
No problem I will send your neurologist, Dr Smith, a message …
Shall we continue to review other symptoms ?
Are the rest of your symptoms as before ?
Any new symptoms ?
Would you like to talk about something else ?
No, thank you
Can I schedule another chat for next week Wednesday at four pm ?
I hope you feel better soon.
Can I end our PLM chat ?