Authority faces challenge of messily heterogeneous data sets
The recently established NHS Counter Fraud Authority is seeking a supplier of commercial fraud analytics software as it looks to tackle an estimated £1.25 billion annual fraud problem across the National Health Service.
An exploratory contract notice puts meat on the bones of a 2017-2020 strategy in which interim chairman Simon Hughes said: “We will use intelligence to paint a comprehensive and detailed picture of the evolving fraud risks faced by the NHS.”
The authority was established in November last year as a “single expert intelligence-led organisation providing a centralised investigation capacity for complex economic crime matters in the NHS”.
Lots of Data, Lots of Different Places…
The authority requires an off-the-shelf software data analytics tool that can “integrate with various data sources including (but not limited to) Oracle databases, Excel spreadsheets, SQL databases”, it said, without specifying a budget.
The software should be able to “filter and query large datasets, analyse, model and profile data to identify trends, patterns and anomalies, manipulate, manage, merge and extract parts of free text, text mining to model structured and unstructured data”.
The contract provides a snapshot of the issues so many public authorities find themselves facing: the need to harmonise heterogeneous data sets, spot patterns in them and then visual findings in a way that doesn’t alienate a non-technical audience.
The authority is open about the scale of the task ahead of them and note that: “More comprehensive intelligence is needed on specific fraud risks.”
“Without thematic intelligence exploring a particular problem in more detail, there is insufficient intelligence to properly inform the planning, resourcing and focus of specific crime reduction initiatives.”
No budget has been listed for the contract as at this stage it is purely “a request for information from market operators who supply a tried and tested business intelligence system,” the authority said.
Hard to Run Machine Learning on Faxes…
Cloud Strategist at UKCloud Bill Mew told Computer Business Review: “The real challenge here is the fragmented and outdated nature of so much of the NHS technology estate. It is all well and good looking at an analytics solution that will integrate with a broad spectrum of data sources from Oracle databases and Excel spreadsheets, to SQL databases, but this presumes that all Trusts know where all their data held and that it is structured and accessible.”
He added: “However, so much of the NHS is reliant on out-dated and inefficient technologies such as fax machines – which are still in widespread use across the NHS. And many specialised legacy equipment and software cannot be migrated to run on more modern systems. This represents a massive challenge. Mind you, if these challenges can be overcome and it can be made to work, securely, then it will be transformational.”