You appear to be making the assumption that data sharing only benefits the private sector. I’m assuming quite the opposite based on my conversations with front line staff across multiple disciplines. The common thread is one best described as ‘confusion’ as to why there are inconsistent or non-existent data sharing practices.
You appear to be making the assumption that data sharing only benefits the private sector. I’m assuming quite the opposite based on my conversations with front line staff across multiple disciplines. The common thread is one best described as ‘confusion’ as to why there are inconsistent or non-existent data sharing practices.
I recognise there are many potential difficulties but I’m firmly of the belief that (relatively) simple solutions could go a long way towards releasing funds. In my mind, a combination of political will (by Labour) combined with legislation AND technical work that dovetail to reflect required outcomes could provide the right framework for this type of value delivering work.
Firstly, I was head of the information team for social care for a medium sized county in the UK for many years, so I'm talking from experience rather than theory.
We developed an inhouse info system that improved data sharing between teams & practitioners, so that, for example, home care teams had access to the OT assessments for their service users. And that extended to community health teams where there were satisfactory data sharing agreements - that is, where the service user agreed. It worked on an individual basis, giving control to the service user.
That sort of data sharing is a real benefit to everyone, but it's very difficult to implement even on a relatively small scale like a county council. Believe me, it was a nightmare. I really don't think I trust government IT - or private IT companies, given fiascos like the Fujitsu fiasco at the Post Office, to take the intense care that we did, or do the requisite detailed and costly testing. It took years to get it to work, and build the cross-checking and security so that everyone knew that if they misused the data, they would be caught.
However, what the government is proposing is that data will be available across the board, with no patient agreement, on the basis that it is "pseudonymised" - meaning that the data users can't identify individuals. There is NO system like this that actually works: either the data user has enough information to be useful (and can work back to identify the patient if they want to) or they don't have enough information to be any use at all.
That sort of data sharing would be incredibly valuable to a company like Palantir, who would be able to use it to leverage it as Cambridge Analytica did. Given access to social media data, Palantir will be able to work out who to target for adverts on the basis of their health conditions. They'll make a fortune and potentially be able to influence elections as well.
Do those sound like the sort of "potential difficulties" that ought to make you stop & think?
The substance of your comment doesn’t surprise me. What you are saying about development is symptomatic of how it becomes hard to develop tech standards. I’ve heard that Trusts are using a spaghetti soup of whatever the CFO’s fave of the week is. That’s absolutely a recipe for trouble, both operationally and for data sharing. I’ve seen this in a several cases.
Were your difficulties centred around data cleansing? Security? Reaching consensus? All of the above and more?
The comment about Palantir and data ‘pseudonomising’ (a new term on me!) appears aimed at solving a different problem.
Back in the day I saw the value of financial data collected by SaaS accounting firms but the hurdles to sharing anonymised pattern analytics were too high for those firms. That has changed and it is possible to benchmark business types in ways that benefit customers but as I say, that solves a different problem to the one I’m imagining.
If government is mandating a Palantir type solution then why? What’s the rationale?
Did your solution end up proprietary to your circumstances or is/was it possible to use at other counties?
Because I came from a background of complexity science, our solution was very specifically designed to be a cultural fit with the people I was working with. We were able to share some of the methodology - we developed a "soft cluster" technique, for example, taking a sample of case files & using the intuitive skills of social workers to categorise them; we then had to find a way to harmonise these local categories with the "national" ones defined by & for the Dept of Health annual returns (which in itself meant we HAD to have a fully non-normalised data warehouse).
That's how we did the reaching consensus/data cleansing work, and the data warehouse allowed us to develop "management information reports" that reflected the mental models of different management groups & even individuals, and allow managers to see the data "through other people's eyes" - an incredibly powerful technique for solving disagreements - because they could look at their team via someone else's report, or other people's teams via their own.
As to why the Govt want a Palantir type solution, they've been sold the idea of the power of data warehousing, without understanding the cultural sensitivity required to make such a structure work in practice when you're dealing with a complex setup like health & social care. The environment is chaotic (in the mathematical sense) and requires an intensely flexible response. That sounds like something AI could cope with, except that current AI is hugely dependent on the datasets provided for it to learn from, and in this environment you'd be needing a new training dataset every week, potentially...
The problem is that when you change the data structures that an organisation uses to monitor itself, you try to change the culture of the organisation as well as its administrative processes; I suspect someone in the Dept of Health imagines that they could have more control over how the NHS & social care work if they had complete control over their information systems, not realising that (as you say) the consequence would be an ever more variable spaghetti of local systems, get-arounds and work-arounds that would take such a "national" system down the plughole in short order.
My system worked with those local standards & allowed people to see how their ever-changing recording practice affected our "performance indicators", costs and whatever they wanted to use as "outcome measures". It was tightly integrated with the case files system, too, so there were very few additional requirements for paperwork or data inputting.
But achieving that on a national level? In the NHS? Where there are fiercely independent trusts? Snowball's chance in hell, I think.
A great response that pI imagine makes an awesome case study in what you have to do. My sense is that you’ve identified another problem in government - where are the experts (like you) and the people with PR smarts (??) to take this type of initiative to Whitehall in a convincing manner?
The problem is you'd have to get Whitehall to rethink its entire culture, away from public-school-inspired simple linear thinking to appreciating the real world with all its chaotic trans-contextual demands.
I did watch Nora Bateson try to start the process many years ago with some shiny new faces from the coalition government; it was most entertaining but not very successful - she still laughs ruefully when reminded about it.
That’s sad. Perhaps the next best thing then is to ‘have a voice’ by joining Labour and sticking it to the Shadow Health Sec and Starmer? That’s my plan, however Quixotic it may seem.
You appear to be making the assumption that data sharing only benefits the private sector. I’m assuming quite the opposite based on my conversations with front line staff across multiple disciplines. The common thread is one best described as ‘confusion’ as to why there are inconsistent or non-existent data sharing practices.
I recognise there are many potential difficulties but I’m firmly of the belief that (relatively) simple solutions could go a long way towards releasing funds. In my mind, a combination of political will (by Labour) combined with legislation AND technical work that dovetail to reflect required outcomes could provide the right framework for this type of value delivering work.
Firstly, I was head of the information team for social care for a medium sized county in the UK for many years, so I'm talking from experience rather than theory.
We developed an inhouse info system that improved data sharing between teams & practitioners, so that, for example, home care teams had access to the OT assessments for their service users. And that extended to community health teams where there were satisfactory data sharing agreements - that is, where the service user agreed. It worked on an individual basis, giving control to the service user.
That sort of data sharing is a real benefit to everyone, but it's very difficult to implement even on a relatively small scale like a county council. Believe me, it was a nightmare. I really don't think I trust government IT - or private IT companies, given fiascos like the Fujitsu fiasco at the Post Office, to take the intense care that we did, or do the requisite detailed and costly testing. It took years to get it to work, and build the cross-checking and security so that everyone knew that if they misused the data, they would be caught.
However, what the government is proposing is that data will be available across the board, with no patient agreement, on the basis that it is "pseudonymised" - meaning that the data users can't identify individuals. There is NO system like this that actually works: either the data user has enough information to be useful (and can work back to identify the patient if they want to) or they don't have enough information to be any use at all.
That sort of data sharing would be incredibly valuable to a company like Palantir, who would be able to use it to leverage it as Cambridge Analytica did. Given access to social media data, Palantir will be able to work out who to target for adverts on the basis of their health conditions. They'll make a fortune and potentially be able to influence elections as well.
Do those sound like the sort of "potential difficulties" that ought to make you stop & think?
The substance of your comment doesn’t surprise me. What you are saying about development is symptomatic of how it becomes hard to develop tech standards. I’ve heard that Trusts are using a spaghetti soup of whatever the CFO’s fave of the week is. That’s absolutely a recipe for trouble, both operationally and for data sharing. I’ve seen this in a several cases.
Were your difficulties centred around data cleansing? Security? Reaching consensus? All of the above and more?
The comment about Palantir and data ‘pseudonomising’ (a new term on me!) appears aimed at solving a different problem.
Back in the day I saw the value of financial data collected by SaaS accounting firms but the hurdles to sharing anonymised pattern analytics were too high for those firms. That has changed and it is possible to benchmark business types in ways that benefit customers but as I say, that solves a different problem to the one I’m imagining.
If government is mandating a Palantir type solution then why? What’s the rationale?
Did your solution end up proprietary to your circumstances or is/was it possible to use at other counties?
Because I came from a background of complexity science, our solution was very specifically designed to be a cultural fit with the people I was working with. We were able to share some of the methodology - we developed a "soft cluster" technique, for example, taking a sample of case files & using the intuitive skills of social workers to categorise them; we then had to find a way to harmonise these local categories with the "national" ones defined by & for the Dept of Health annual returns (which in itself meant we HAD to have a fully non-normalised data warehouse).
That's how we did the reaching consensus/data cleansing work, and the data warehouse allowed us to develop "management information reports" that reflected the mental models of different management groups & even individuals, and allow managers to see the data "through other people's eyes" - an incredibly powerful technique for solving disagreements - because they could look at their team via someone else's report, or other people's teams via their own.
As to why the Govt want a Palantir type solution, they've been sold the idea of the power of data warehousing, without understanding the cultural sensitivity required to make such a structure work in practice when you're dealing with a complex setup like health & social care. The environment is chaotic (in the mathematical sense) and requires an intensely flexible response. That sounds like something AI could cope with, except that current AI is hugely dependent on the datasets provided for it to learn from, and in this environment you'd be needing a new training dataset every week, potentially...
The problem is that when you change the data structures that an organisation uses to monitor itself, you try to change the culture of the organisation as well as its administrative processes; I suspect someone in the Dept of Health imagines that they could have more control over how the NHS & social care work if they had complete control over their information systems, not realising that (as you say) the consequence would be an ever more variable spaghetti of local systems, get-arounds and work-arounds that would take such a "national" system down the plughole in short order.
My system worked with those local standards & allowed people to see how their ever-changing recording practice affected our "performance indicators", costs and whatever they wanted to use as "outcome measures". It was tightly integrated with the case files system, too, so there were very few additional requirements for paperwork or data inputting.
But achieving that on a national level? In the NHS? Where there are fiercely independent trusts? Snowball's chance in hell, I think.
A great response that pI imagine makes an awesome case study in what you have to do. My sense is that you’ve identified another problem in government - where are the experts (like you) and the people with PR smarts (??) to take this type of initiative to Whitehall in a convincing manner?
The problem is you'd have to get Whitehall to rethink its entire culture, away from public-school-inspired simple linear thinking to appreciating the real world with all its chaotic trans-contextual demands.
I did watch Nora Bateson try to start the process many years ago with some shiny new faces from the coalition government; it was most entertaining but not very successful - she still laughs ruefully when reminded about it.
That’s sad. Perhaps the next best thing then is to ‘have a voice’ by joining Labour and sticking it to the Shadow Health Sec and Starmer? That’s my plan, however Quixotic it may seem.