A recent report in the US; The State of Insurance Fraud Technology revealed some worrying numbers. In particular, only 14% of insurers are using advanced technological fraud detection techniques. Given the rise in sophisticated insurance fraud rings – this seems a little disturbing. However, it may just be that the industry isn’t aware of what can be done. With that in mind we’ve put together a quick guide to the kind of technology that can combat insurance fraud.
Link Analysis
Whilst this might sound like a Search Engine Optimization technique; it’s not. Link Analysis is the use of modelling software to map complex relationships between large sets of insurance data. It can go beyond the borders of the insurer and link to data from solicitors, hospitals, garages, etc. to see if there are patterns of behaviour that indicate fraudulent behaviour.
Predictive Modelling
This is software the takes previously identified patterns of behaviour and extrapolates them in real-time against insurance data. The idea is that predictive modelling lets you know when fraud is taking place, preventing payments to the fraudulent parties and allowing insurers to take a long, hard look at suspicious claims as and when they arise.
Text Mining
Not all data is neat. Much of the data that insurers collect, for example call data from call centres, is unstructured. Being able to analyse that data for patterns requires the use of text mining; this provides the ability to wade through huge volumes of unstructured information and still identify useful data or patterns within that data.
Data Visualization
Massive data sets can quickly become overwhelming even to the most skilled analytical teams. Software analysis is generally very good at recognizing patterns of behaviour, which have already been identified at some point in the past. It’s not as good at making the intellectual leap to identify new patterns of suspicious behaviour. Data visualization allows huge complex data sets to be represented in visual format; this allows analysts to search for patterns that simply wouldn’t be visible in the original raw data sets.
Geographic Data Mapping
Geographic data sets are often hard to correlate without mapping them on a visual overlay. Insurance fraudsters often have to organize in tight areas of geography in order to get a consistent approach to making fraudulent claims. The ability to see trends by geography – gives insurers an effective means of finding these rings of fraudsters early and cutting them off before they cause too much economic damage.
Insurers have never had as many tools at their disposal to combat fraud as they do today. Investment in these tools needs to increase in order for them to reach their maximum effectiveness but the message should be clear; those that do invest are likely to find that the 2.1 billion GBP lost each year in fraud by the British Insurance Industry can be dramatically reduced.