Fraud: Can healthcare learn from telecoms?

By Ana Castelo Branco, ‎product marketing specialist, WeDo Technologies

A recent report led by former NHS anti-fraud boss Jim Gee revealed that fraud could be costing the NHS in England as much as £5.7 billion. Among the scams highlighted were dentists claiming money for NHS care they did not carry out and GPs falsifying records for extra payments. The biggest area of fraud was estimated to be payroll, accounting for losses of between £555 million and £1.49 billion.

While these figures are based on estimates as well as detected fraud, the large sums at stake highlight the huge scale of the problem and the burden fraud places on healthcare providers, which comes at a time when the NHS is under increasing financial pressure. In light of these findings, it is clear that the current measures used to prevent fraud are falling short.

Unfortunately, much of today’s healthcare fraud is committed by providers who feel that they are not getting paid what their insurance companies owe them, with fraud often being thought of as a way to recover what is due. As a result, it is vital that healthcare providers, including the NHS, learn lessons from the telco sector and create a multi-layered approach to fraud detection and management that identifies, contains and responds to new types of controls (of fraud), to protect critical assets and data.

When it comes to fraud prevention, it can be assumed that the unscrupulous will find any hole in an organisation’s system and take full advantage of it. It would therefore be in healthcare providers’ best interest to prepare better documentation to support diagnoses or claims that are questioned, and have an automated system in place to analyse records and find any issues that can be corrected. No one can respond to fraud attacks they don’t know about.

Historically, providers have relied on manual or automated rules-based systems to detect fraud. The advantage of using a rules-based system is that businesses can encode knowledge and use automated tools to undertake the complex task of detecting fraud. It adopts a common sense approach, in which you don’t need to process a million variables to find a fraud type, and can instead use pre-determined rules to examine specific complex variables to quickly identify fraud with a high rate of accuracy.

However, this approach will only be partially successful, as it depends entirely on the strength of the rules being employed. Data-driven analytics for instance, can be used to catch emerging behaviour patterns and problems that aren’t yet obvious enough to capture in a policy rule. Rules-based systems in contrast can only detect specific types of fraud for which rules have been established. In other words, rules-based systems can protect your organisation against previous threats, but may often prove ineffective against preventing emerging ones.

In order to stop fraud before it happens and take action before it is too late, it is essential that healthcare providers learn lessons from the telco sector to implement more effective modern fraud management techniques, such as using anti-fraud analytical systems, in order to prepare against potential threats or behaviours.

As an example, by using the combination of a rules-based system together with unsupervised models that use big data technologies to process analytical models, healthcare organisations will be able to look for unknown types of controls (of fraud). This approach collects information from external sources that go well beyond the normal inputs used to create profiles.

By using cases that are 100% matches for fraudulent activity as a baseline, these types of predictive fraud systems rely on analytics to look for providers that have profiles with lower percentage matches, and then create correlations between users fitting that profile. These providers are not automatically identified as ‘fraudsters’, but since parts of their profile match typical fraudster behaviours, they can be highlighted for fraud management teams to investigate further, to determine if it is a new type of fraud for which new rules must be created.

As a result of having tools available that can integrate with existing information for fraud investigation, healthcare managers will be able to make more informed decisions that create a centralised, aggregated view of information from disparate sources. This can also accelerate situational awareness and help support faster and more informed decision making across multiple departments including revenue, cost, productions and provisional monitoring.
Benefits of hybrid fraud management

The benefits of an automated fraud prevention system that combines business rules with predictive analytics and machine learning cannot be overstated. Business rules will help organisations to operationalise the benefits of big data, taking the data and analysing it in a way that reduces the complexity, as well as delivering alarms and alerts that have already been verified to meet specific decision criteria. Predictive analytics, on the other hand, can be utilised by businesses to turn uncertainty about the future into manageable probabilities.

Yet only by implementing fraud management systems that combine both approaches, will the healthcare industry be able to provide the best of both solutions, using previously analysed suspicious patterns to enrich the rules knowledge base. This will eliminate repetitive and time consuming decisions for analysts when a new suspicious pattern is found, in order to stop fraud before it happens and take action before it’s too late.