Data and the NHS
Posted: 20 Apr 2022, 11:10
There is a man called Marc Warner who owns an Artificial Intelligence company called Faculty.
He is trying to use AI to help the NHS. To help with the cancer backlog, to help with the throughput of NHS patients, to help with waiting lists.
To ensure a good result in one area without creating a bad result somewhere else.
So to do this you have to analyse data. When Warner started off in Kettering Hospital he had a shock, because the data available was virtually useless.
Patient information was split by speciality within the hospital. By the NHS's own admission the left hand can't see what the right hand is doing.
This is interesting to me because I once worked on a similar project in a police force. Criminal information was split across a crime system, an intelligence system, a command and control system, etc. We wanted to centralise the information, to create a central nominal across all systems.
So I feel for Mr Warner. Is the Ronald Smith in oncology the same R Smith who has a record in Rheumatology? Do they live at the same address? Oh they do. Same guy, right? So merge the record. Then it turns out that R Smith is Raymond Smith, Ronald Smith's son. Ooops. So now you have to untangle the merge. Oh, you didn't save the data? So now you don't know which one is in Oncology and which in Rheumatology?
Data quality is going to be a HUGE issue. Warner was working with the NHS at the start of the pandemic and was having his staff copying and pasting thousands of Excel spreadsheets into the format that he needed to get data ready to analyze. The prospect of having to do this across the NHS is enough to make anyone shudder.
The data will be held on different platforms, in different formats. The data Warner needs may not even be held by some departments. Some may be held in free text format, badly spelled, wrongly spelled. Then you have the issue that GP records are seperate from hospital records (or used to be) and may be held in different formats again.
Good luck Mr Warner, you'll need it. But if it comes off it will be a massive step forward.
He is trying to use AI to help the NHS. To help with the cancer backlog, to help with the throughput of NHS patients, to help with waiting lists.
To ensure a good result in one area without creating a bad result somewhere else.
So to do this you have to analyse data. When Warner started off in Kettering Hospital he had a shock, because the data available was virtually useless.
Patient information was split by speciality within the hospital. By the NHS's own admission the left hand can't see what the right hand is doing.
This is interesting to me because I once worked on a similar project in a police force. Criminal information was split across a crime system, an intelligence system, a command and control system, etc. We wanted to centralise the information, to create a central nominal across all systems.
So I feel for Mr Warner. Is the Ronald Smith in oncology the same R Smith who has a record in Rheumatology? Do they live at the same address? Oh they do. Same guy, right? So merge the record. Then it turns out that R Smith is Raymond Smith, Ronald Smith's son. Ooops. So now you have to untangle the merge. Oh, you didn't save the data? So now you don't know which one is in Oncology and which in Rheumatology?
Data quality is going to be a HUGE issue. Warner was working with the NHS at the start of the pandemic and was having his staff copying and pasting thousands of Excel spreadsheets into the format that he needed to get data ready to analyze. The prospect of having to do this across the NHS is enough to make anyone shudder.
The data will be held on different platforms, in different formats. The data Warner needs may not even be held by some departments. Some may be held in free text format, badly spelled, wrongly spelled. Then you have the issue that GP records are seperate from hospital records (or used to be) and may be held in different formats again.
Good luck Mr Warner, you'll need it. But if it comes off it will be a massive step forward.