Using algorithms to detect fraud and find criminal behaviour

In California the owner of a firm supplying power wheelchairs paid for by the Medicare and Medi-Cal funds, was convicted on 20 March 2015 of fraud of $3.5million (1). She had been paid almost $2million, and sentencing is set for June 2015.

Fraud within public funds is not new. In the UK the programme of Individual Learning Accounts had to be abruptly closed and abandoned because of fraudulent claims by criminals using powerful computers. They searched for weaknesses in the system’s data and then made hundreds of automatic false claims on each ‘find’, each claim for over £150.

In the USA the authorities are making it known that all financial claims are now monitored by anti-fraud algorithms which look for known patterns of criminal behaviour. Clearly these algorithms are not themselves in the public domain, but common sense suggests that any statistically significant ‘spikes’ in the claims will raise a red flag.

But more important than money is life itself. The same algorithms can also be used to interrogate health data as well as health funding.

Harold (Fred) Shipman, the mass murderer, killed patients at an estimated one a month for 25 years, mostly elderly women living alone. “By 1998 … a local taxi driver, who often ferried old ladies around, noticed that they seemed to die shortly after seeing Shipman. Linda Reynolds, a nearby GP, noticed that his patients were dying three times more frequently than hers.” (2) The subsequent Inquiry determined that he murdered 210 people, plus a possible 45 others.

The Mid-Staffordshire hospital scandal of appalling local health care was also first detected by patients’ relatives who knew that something was wrong. In retrospect, the data told the story in terms of falsely coding people’s deaths. The senior managers in charge at the time tried to “explain away” these sudden lurches in unlikely deaths, but the relatives’ stories and the data supported each other and the truth emerged.

Back to funding in the UK, and at a more routine level thankfully, we have the recent convictions for fraud by some A4E staff making false claims to the Department for Work and Pensions and to the European Social Fund.

Auditors are usually well aware that – if it looks too good to be true, then it probably is suspect. Such as projects which achieve all their outputs and outcomes, with everything spot-on the original profile, along with wonderfully neat files all signed in the same shade of ink. Hmmm.

It might not be as high-tech as super-computer algorithms, but auditors as well as machines can know which trends or characteristics in the data will need further investigation.


(2) Forensics: an anatomy of crime, Val McDermid, 2014, p107.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s