Innovating The Insurance Claims Processes Thru AI And Machine Learning

Original article was published on Artificial Intelligence on Medium

Innovating The Insurance Claims Processes Thru AI And Machine Learning

Insurers, across the globe, face new challenges each year that make it tougher to maintain the bottom line while supporting high standards of customer satisfaction.

Insurers were traditionally dependent on the information collected by investigators and provided by claimants. In the new era, technology has transformed the collection and analysis of data completely.

Digital technologies like Internet-of-Things (IoT), Machine Learning (ML), Robotic Process Automation (RPA), Drones, to name a few, have opened new avenues to collect real-time data, to better-predict future events, to access hazardous regions without risking human lives, and to expedite the response time, leading to enhanced customer experience and better business outcomes.

In today’s era, digital maturity is pretty low and only a few companies are fully digital when it comes to claiming operations. Only some IT leaders consider their organizations to be digitally progressive. Companies need to take a deep dive and embrace digital and AI technologies to survive in today’s digital era.

Top challenges in claims management

Insurers face the need to resolve these challenges to optimize their claims operations:

Find efficiencies in the claims registration process

Typically, the claims registration process is data-intensive and repetitive, leading to operational inefficiencies.

Improve claims adjuster’s allocation inefficiencies the assignment of claims adjusters is dependent on availability, workload, and location data, resulting in inefficiencies.

Handling claims estimation in real-time situations

A significant amount of time is lost in preparing the data as opposed to analyzing claims. Machine Learning and AI can significantly help in reducing this gap.

Early detection of fraud

Digital & AI Technologies can help in early detection of fraud and help eliminate the manual labor and intensive claim processes activities which result in delayed claims

Inspection of hazardous and dangerous locations

In dangerous and hazardous locations where inspections are necessary to gauge damages, claims inspectors may potentially be exposed to safety risks and false compensation claims.

Handling the challenges with modern technologies
The insurer willing to embrace the latest technologies and willing to go the extra mile will stand out in the competitive market place and be the winner. However, not every company operates in the same way; hence, the solutions to the challenges cannot be the same and will vary depending upon the specific operating model in place. The goal for all insurers is to improve revenue and reduce costs. They need to adopt innovative technologies and embrace new ways of performing business.

Depending upon the goals, both short-term and long-term, an insurer may want to consider one of the following proposed solutions, or have a customized mix of solutions, to achieve the optimum results. Some sample applications are:

Internet of Things (IoT) to improve efficiency in the claim registration process; automobile telematics to improve service and risk assessment

Machine Learning (Artificial Intelligence) to improve efficiency in claim loss reserving and fraud detection; sales optimization and client targeting

Robotic Process Automation (RPA) to improve efficiency in assignments and investigation efforts; highly useful in claims processing and policy origination

Robotic Process Automation

Robotic process automation can enable improved claims processing in many ways.

RPA enables execution of transactional and administrative tasks that are rule-based, repetitive, and frequent. This fits in well with the need of insurance companies to gain operational efficiencies with back-office operations such as claims processing, data entry, and policy application handling. RPA can be adopted in routine processes for streamlining.

Robotic process automation can optimize claims processing in some important ways:

Automating the assignment of claim adjusters

Administrative and repetitive tasks such as assigning claims adjusters can be easily performed by RPA based on availability, workload and location data

Processing of bulk claim payments, such as in catastrophes

Repetitive processing of functions such as bulk claim payments in case of catastrophes requires speed and efficiency which RPA can provide. Large backlogs of claims data can be easily processed by adopting automation techniques

Enhanced fraud detection and management

RPA can be leveraged to run fraud detection algorithms to enhance detection and management of fraud activity

RPA can reduce the need for utilization of human resources such as claim adjusters in claim processing.

Internet of Things (IoT)
Smart devices that you use in your daily life can communicate with each other and share valuable information. This information, if exchanged between the insured and insurer, can improve performance in various areas ranging from the quick and easy claim notification to providing exceptional customer service.

Virtual Customer Service Agents (VCSA)
As Internet usage increases year-over-year, with most of the Internet usage driven through smartphones, live chat-based customer support is the most effective medium for providing customer service. Insurance companies can take customer service a step further and introduce Artificial Intelligence-driven virtual agents to provide customer support wherein each customer receives a tailored service based on the policy/claim information related to the insured, thus reducing the wait time and effort by the insured to contact insurance companies.

Customers can initiate a chat from the smartphone/web with the VCSAs to notify a claim, enquire about the claim status, or even seek suggestions on risk management. With the advancements in areas like Artificial Intelligence, Natural Language Processing, Optical Character Recognition (OCR), VCSAs can provide personalized and close to human-like support to the customer.

The VCSA can ask intelligent questions to get the relevant information from the customer, specific to the claim instead of a generic questionnaire. The claim will be processed faster by reducing the waiting time to receive information from the claimant.

However, it is still expected that a conversation is initiated by the user manually using a smartphone or web unless virtual assistants are deployed to sense the environment and notify the insurer automatically about a pre-configured invent that suggests a possible loss.

Virtual Personal Assistants
Users can interact with virtual assistant technologies, like Google Home or Amazon Echo, to request task execution such as making online reservations for services, making phone calls, etc. Such devices, with customized configuration, can send alerts of situations that may potentially lead to a loss. For example, the device can send an alert to the insured and insurer when the smoke alarm goes off and no action is taken for a pre-defined time. This functionality enables the insured and/or insurer to take the necessary steps to prevent or reduce the loss due to fire or other perils or may even initiate notification of loss event. The majority of today’s customer engagements can be supported by bots.

Wearable Devices
Wearable devices are becoming popular in the market and its usage is increasing day by day. This increased usage of fitness trackers, smartwatches, smart shoes, and even smart clothing, has created an enormous amount of data that can be used for analysis. The exploitation of this data can help build predictive models that can alert the insured of potential adverse impacts on health due to shifts in lifestyle. Sending such alerts to the devices of the insured along with suggestions will help avoid adverse impacts. The application of such a model can help prevent and/or reduce the loss in the area of health coverage and Workers’ Compensation.

Drones
In their search for finding new ways to improve operational efficiencies, insurers have started exploring the use of drones (unmanned aerial vehicles) for claims investigation and processing.

Drones can be very effective in situations where the safety of the adjusters and rapid deployment of adjusters is needed, such as in the case of a catastrophe, environmental impairment, and accident-related claims. Similarly, in cases of auto accident claims, deploying drones to survey the accident site, provide images of the accident site, or aid in accident reconstruction can prove to be very effective.

The benefit realized by the insurers would be lowered costs since the cost of deploying drones would be cheaper than deploying claim adjusters. An additional benefit realized to insurers would be a reduction in workers’ compensation claims made by claim adjusters since drones would be able to go into areas such as catastrophe areas where the safety of claims adjusters would be of prime concern.

Another area where drones could be very effective is the surveillance and monitoring of disability claims. Insurers could provide adjusters with discreet ways to gauge the physical activity of claimants claiming disability, and then either, verify or disqualify disability claims. This would help insurers in reducing fraudulent claims, as well as the cost of hiring special investigative units for surveillance and monitoring.

However, one key issue that insurers could face using drones is the privacy of claimants. To address this concern, a few states have passed laws that talk about restrictions on taking images and videos of individuals. Insurers would have to deal with regulations before utilizing drones for surveillance.

Machine Learning
With increasing requirements from insurance regulators to report and maintain good solvency status, insurance companies need to report close to actual estimates of the claim liability that the company holds in terms of claims reserves.

Adopting machine learning for estimating claims reserves, the periodicity can be reduced to a point where it is close to real-time estimation, thereby allowing better reporting of estimates and reduction in time spent on data preparation for reporting. Since machine learning depends heavily on learning from the decisions made along the way, it is possible to account for factors like inflation, changes in actual claim settlement amounts, changes in claim data due to major economic events, or even changes in actuarial philosophy of company for reserving adjustments.

Today the process of adjusting the claims reserves is manual and one cannot expect the claims management team to review each claim and update the reserve periodically on a timely basis. Machine learning, if integrated into the core claims management system, can perform real-time claim level adjustments to reserves and additionally perform adjustments at a pre-determined period, so that they more accurately reflect the potential ultimate value of the claim then.

In the process where a system is trained about the reserving patterns, it can also be trained with potential fraud indicators with an additional feed from predictive models to improve the accuracy of flagging of potentially fraudulent claims. With an average assumption of 10% of claims being fraudulent, machine learning can help improve the efficiency in the claim process by identifying more accurately, thereby saving cost and time in deploying claims investigators.

Machine learning, as a system, learns from the behavior of the user over the period and thereby develops a trend of possible actions in different scenarios. Hence, one of the constraints to implementing machine learning is that it needs to learn from the decisions made by actuaries that can lead to a longer learning duration. However, if the system is trained to study the historical actions taken to adjust the reserves and input is provided of the reasons for adjustments, the duration of learning can be reduced.

In addition to new technologies, operationalizing information to improve claims outcomes requires a shift in leadership mindset, talent models, organization design, company culture, and the skills needed for success. The disruptive impact of new technology and new claims processes is fundamentally re-writing the way claims organizations operate.

Conclusion
The advancements in areas like machine learning, robotics, and IoT devices will bring a drastic shift in the way claims are processed, managed, and reported today, thereby resulting in better operational performance and improved customer satisfaction. The use of robotic process automation will allow claims organizations to involve the right resources at the right time and improve the claim outcomes. Drones will enable the efficient processing of claims. Introduction of machine learning in claims reserving and loss estimation processes will help improve the frequency and accuracy of the reserving, thereby providing opportunities to improve regulatory reporting and investment strategies through the maintenance of appropriate capital ratios.

Fortunately, the use of several emerging technology tools can support efficient claims processing, which will be a key to better customer satisfaction, thus better customer retention.

Inmediate is an insurtech startup from Singapore that is using the latest technology such as Artificial intelligence, Distributed Ledger, and NLP, making insurance processing fast, cheap, and flexible. That gives for better processes, lower costs, improved time to market, and new revenue opportunities.