Welcome to Part 2 of our article examining what the future of Digital Claims Processing holds.

If you missed Part 1, in which we covered personalised policies, FNOL, assessment and investigation; click here and we’ll see you back shortly.

Otherwise, let’s continue on with the next stage of the digital claims process – Adjudication.

Adjudication

All claims benefit from some automation at this stage. Once the investigation has taken place and the required information collected, the customer expectation is that this stage happens quickly. The key first step is to get the assessment to the appropriate person with the authority to make the decision. Secondly if further authorisations are required, before an offer is made, this should happen digitally. Decision Automation for small claims can be rules based. LEAP can automate authority, audit and exception management. Also, FLOvate can put an digital framework around the cycle time; automatically escalating before timescales become an issue. This ensures that customer expectations are met – even if adjudication is carried out manually in the case of large claims.

Negotiation

Sometimes, the parties involved in the claim don’t accept the initial offer. Your organisation may have a formal appeal process. This may be as simple as a process of offer and counter offer. Alternatively, the claimant (or their representative) may need to add information for further adjudication.

Regardless, technology can help make this stage go more smoothly.

 

How FLOvate LEAP helps with claims negotiation:

Data Collection App technology allows additional evidence to be requested and seamlessly uploaded to the process. Therefore, supporting a claimants’ request for a review of the amount offered.

 

Data collection screen

Data collection view example

Settlement

When the decision has been made further delay is unnecessary. To be efficient, it’s important to automate most of the settlement stage. And that applies even where humans are required. For example, authorisations.

Using digitisation, we can ensure that authorised parties have immediate access to the relevant part of the digital file. Approving that specific element is fast and straightforward, with all the information to hand. Finally, to communicate the settlement, we can simply automate this as a task.

 

Authorisation required illustration

 

How FLOvate LEAP helps with digital claims settlement:

FLOvate LEAP low-code comes with advanced digital authorisation built in. Simply configure your required authorisation levels and groups. For convenience, you can also configure all necessary authorisations in parallel. Are external parties involved? No problem. With LEAP, authorising a task is as simple as clicking on an email. Digital authorisation is also blind. So, the person making the request cannot predict who will be asked to authorise it.

All organisational processes – and that includes claims – involve actions and decisions. In a digital environment, most actions are automated. Or, they are heavily assisted by technology. As a result of data collection technology and third party portals, we pretty much eliminate re-keying.

That leaves us with decision automation.

Digital Decision Making

Decisions occur at all stages in the claims process. Some are explicit and others occur manually, based on experience.

Explicit decisions are easy to model and implement in most platforms. These range from simple triage rules implemented post FNOL to authorisation rules when setting reserves or making payments. Typically they involve 2-5 dimensions (fields) and one or two nested decision branches.

Sometimes, we know the algorithm for a decision. However, we choose to hide it in training materials. As a result, we compound the expense of training and manually making these decisions by heavily auditing them.

 

Machine learning algorithm

 

So What’s The Answer?

There is a tried and tested workflow for converting these informal algorithms into automated routines:

  • Analyse and implement the implied algorithm; start with decision assistance highlighting the most likely decision to field test it.
  • Refine the algorithm based on the correlation between the suggested decision and the one taken by the claims professional.

This method improves efficiency and consistency of decisions and can lead to full automation.

The alternative is to snapshot the data (fields) used in the decision, plus the manual decision outcome into a dataset at the point of the decision. You may have to implement this as a double blind decision or add “second pair of eyes” (SPOE) controls to ensure you eliminate human errors. Once you have sufficiently large set of independent variables (fields) and (error free) dependent variables (outcomes), you have a good candidate for an AI generated algorithm using machine learning (ML).

The Data Curation Challenge – Too Many Variables?

Those with experience at the coal face of AI and ML will tell you that the main challenge is data curation. All the following challenges need to be handled, usually by scarce data scientist resource:

  • Which independent variables (input fields) relate to the dependent variable (outcome)?
  • Have the independent variables been collected at the point of decision?
  • Is this the same as the proposed point the AI generated algorithm will run?
  • Many ML routines don’t handle gaps in the data. Is the dataset complete?
  • How do you intelligently infill missing data?
  • Is the data structured or unstructured?
  • Which ML routine (there are many) will work best for the data I have?
  • How do you split between training and test data cohorts?

 

How FLOvate LEAP helps with digital decision making

FLOvate LEAP low-code supports configuration of explicit decision algorithms (rules/triages). It also supports embedding of implied decision algorithms derived using machine learning (ML).

LEAP uses Microsoft Azure Cloud hosted ML routines to generate AI algorithms. In 2020, LEAP will also include guided AI routines that automatically manage data curation. What’s more, it can produce training and test datasets, and select the most likely ML routine for the type of data under consideration.

The selected ML routine will run periodically and report the correlation achieved. Promoting for deployment of the AI generated algorithm when acceptable levels of correlation are achieved.

Digital Claims and FLOvate LEAP: In Summary

Digital claims is not an end in itself…

  • Claims digitisation benefits your customers when they can access and input information in a simple way, coupled with regular updates leading to faster settlement.
  • It works for your organisation when it increases process effectiveness, efficiency and retains your customers.

What’s next after digital claims? SMAC (Social, Mobile, Analytics, and Cloud)? To keep ahead of the digital technology transforming claims, make sure to follow FLOvate on LinkedIn, Facebook and Twitter.


About the author

Edwin Harrell is a technology entrepreneur specialising in digital transformation and process design. Edwin worked as part of the team that founded and later listed WNS Holdings, a global business process outsourcing organisation, on the NYSE. He is the founder of FLOvate; a technology business specialising in low code and digital business process management software.