Tag Archives: machine learning

ML for Commercial Property Insurers

For years, the preparation and management of data have exposed themselves as two costly and critical challenges for commercial property insurers. These challenges are hampering production and efficiency and inhibiting growth and profitability. The flow of submissions and the preparation of statement of values are laborious and time-consuming to agents, brokers, insurers and anyone else in between. Without a solution to meet the changing market needs to manage these complex data sets, commercial property insurers’ ability to quickly respond to markets and aggressively price business is hindered.

The inability to address these issues has obstructed the process, making it prone to error and hard to scale, especially in today’s market. In turn, this obstruction limits the speed and accuracy of commercial insurers’ decision making and debilitates businesses’ potential to grow. The gap between data preparation, screening, prioritization, analysis and pricing steepens, and companies find themselves stagnant and looking for answers. There is yet to be a commercially viable solution focused exclusively on automating the operational preparation and processing component of commercial property insurance data so companies can better meet the growing need of customers and markets and handle the substantial work that is required.

A company’s inability to respond quickly can affect the relationship with the producer, leading to a higher chance of being selected against. These types of companies are more likely to take on more complex characteristics, along with riskier business as the expectation of long processing times is already set.

But we’re starting to implement machine learning into problem-solving tools to address these challenges. These tools enable commercial re-insurers to take their raw data sources and harmonize them with next-gen technology that analyzes, reviews and writes business submissions to provide companies with the competitive edge that’s been sought after for years.

Making the most of your data 

On average, commercial property insurers can only process a portion of the submissions they receive. Typically, managing and preparing results in inconsistencies surrounding labeling, coding and more, which create downstream issues with pricing, modeling and aggregation. Critical amounts of data are lost through the process, and information is not consistently accessible, hindering the ability to make crucial decisions. The only way to solve this and manage business expectations is by hiring additional skilled labor, but this increases the acquisition costs, hurts profits and isolates information among the skilled experts.

Using machine learning, data integration and analysis offer the ability to make data mapping suggestions based on learning algorithms. Manual adjustments are then fed back into the decision-making model, transforming complex, big data into actionable insights that are accessible, in real time, to the entire organization. This allows teams to spend their time on business-generating activities and acting on insights from data, instead of the constant back and forth editing spreadsheets.

See also: How Machine Learning and AI Reduce Risk  

Potential opportunities to grow the business are lost today because of the acquisition costs for new business, but machine learning allows insurers to get from point A to point B by enabling them to screen and prioritize submissions. Today, submissions can be prepared one at a time, but, with machine learning, employees are able to triage multiple submissions at once, including new submissions, enabling the underwriter to focus on the key deals and negotiating terms.

A solution for the enterprise

Giving users the ability to gain access to all commercial property data gives them a wider, more detailed view of the market as well as an understanding of the risk profiles that producers are sending. By providing an automated process to ingest and prepare data, insurers are afforded a more efficient and flexible way of consolidation that essentially helps eliminate errors, cuts costs and promotes growth as companies can now allocate resources to address other areas of the business. Ultimately, automation and machine learning provide insurers with the ability to process submissions at a much higher rate of around 80%.

While giving data access to individuals within the company is beneficial, expanding that access in the form of outsourcing can create a number of different security concerns. Many insurers are operating and sharing data globally, making security and compliance with regulations like GDPR an absolute necessity. Outsourcing is nearly impossible under GDPR due to the heightened risks in sending and having external sources manage large amounts of customer data. Insurers need to show due diligence in not only securing their own data but their customers’, as well. In place of outsourcing, we are now seeing data management and storage platforms incorporating heightened security and data integrity into the design, ensuring these tools meet security standards such as ISAE 3402, SSAE 16, AES at rest and SSL/TLS in transit and ISO 27001. Meeting the standards not only helps to prove compliance with regulatory requirements, it also shows customers that insurers are taking their data privacy demands seriously.

Looking ahead

For the commercial property insurer, it is of the utmost importance to have the ability to prepare and manage complex data sets with an easy, quantifiable solution. Emerging solutions across the industry will enable insurers to make fast, appropriate decisions required to address the always-changing market and expand the business.

See also: How Machine Learning Transforms Insurance  

With the introduction of technology such as artificial intelligence and blockchain, combined with machine learning, the realm for new directions provides the insurance industry an unprecedented opportunity to collaborate. While these changes will continue to bring us new and improved methods to get things done faster and more efficiently, one thing is certain, ambitious commercial property insurers are already discovering collaborative initiatives to establish concept cases.

Three Key Takeaways

  • The current processes are putting insurers behind their competitors in the commercial property market because they typically process 20% to 30% of the submissions received.
  • Machine learning is allowing insurers to triage, screen, prioritize and score submissions much faster and with a higher output rate.
  • The result is more completed submissions, which leads to the ability to be first to market.

How Machine Learning Halts Data Breaches

Although we hear a lot about major cybersecurity breaches in non-insurance organizations – Target, Experian, the IRS, etc. – there have been breaches in the insurance industry, too, albeit less publicized. Nationwide faces a $5 million fine from a breach back in 2012. Horizon Blue Cross Blue Shield is still the defendant in a class action suit over a 2013 breach that affected 800,000 of its insured. 

As hard as organizations try to secure their data and systems, hackers continue to become more sophisticated in their methods of breaching. This is why innovation in risk management and insurance is so important.

Can Machine Learning Improve Cybersecurity (and Vice Versa)?

The short answer is yes. Because machine learning can collect and process huge amounts of data, the technology can analyze historical cyber-attacks, predict types that may occur and set up defenses against them

Here is a very simple example:

An on-site employee has decided to use his computer to access some shopping sites during his lunch break. One of those sites has elements that alert the machine of a potential security threat. The security team is notified immediately. It is then possible to block access permission from that computer to any data that could be useful to hackers until a full investigation can be completed. 

See also: How Machine Learning and AI Reduce Risk  

This may be a rather far-fetched example because most organizations limit private use of their computers in advance. But consider the Horizon breach – two laptops were stolen from a facility, and access was obtained. Or the case of Target, where a third-party contractor did not have appropriate security in place, and hackers were able to access the company’s systems through this third-party. Machine learning can help to reduce these threats through a proper alert system, and remote shutdowns can then occur.

Common Types of Data Breaches that ML Can Help Thwart

1. Spear Phishing

Company employees receive emails every day, in their company inboxes. Some of these, from sources that may not be known, can include malicious links. 

There are now ML algorithms that can identify and classify language patterns – email subject lines, links, body content/communication patterns, phrases and even punctuation patterns. Anomalies can be flagged, and security analysts can investigate, even catching the emails before they are opened, if the system is set up correctly. Some of these emails, for example, may be very poor translations from foreign languages, certainly not professional translations from services like The Word Point. Poor translations will alert machines that spear phishing is a possibility.

2. Ransomware

Most everyone is familiar with this security threat. Users’ files are “kidnapped” and locked. Users must then pay up to get an encryption key that will unlock those files. Often, these files house critical client data, other proprietary information or system files that are necessary for business operations. The other type of ransomware attack will simply lock a user’s computer and not allow access until the demanded amount is paid.

To train a machine to identify potential ransomware requires some pretty deep learning. Data sets of historical ransomware files must be loaded, along with even larger sets of clean files, so the machine can learn to distinguish between the two. Again, so-called micro-behaviors (e.g., language patterns) are then classified as “dirty” or “clean,” and models are developed. A ransom file can then be checked against these models, and necessary action taken before files are encrypted or computer access locked.

3. Watering Hole

Employees, especially insurance agents who are out in the field, may have their favorite spots for coffee or lunch breaks. Or, suppose, a group of employees have favorite food joints from which they frequently order food for delivery or takeout. Whether they are using the Wi-Fi in that watering hole or accessing that business’s website to place an order, there is far less security and an ideal place for hackers to enter a user’s access/credentials through that backdoor. 

Sometimes this is called “remote exploitation” and can include a situation like what occurred with Target – a third party is used as the “door” to get in.

ML algorithms can be developed that will track and analyze the path traversals of an external website that employees may be accessing on devices they are using either on- or off-site. Users can be directed to malicious sites while they are “traveling” to a destination site, and this is what ML can detect.

See also: How Machine Learning Transforms Insurance

4. Webshell

A Webshell is nothing more than a small piece of code. It is loaded into a website so that a hacker can get in and make changes to the server directory. The hacker then gains access to that system’s database. Most often, hackers look to take banking and credit card information of customers/clients, and this type of attack occurs most often with e-commerce websites. However, medical practices and insurance companies are certainly at risk, too, because they house lots of personal data. When the insured set up automatic payments from their bank accounts, the activity is even more attractive to these hackers. Payments are simply routed somewhere else.

Machines can be trained to recognize normal patterns of behavior and to flag those that are not normal. Machines can also be used to identify webshells preemptively and then prevent them from exploiting a system.

The Requirement? Machines and Humans Must Work Together

Will machines ultimately eliminate the need for in-house or contracted cybersecurity experts? Highly unlikely. At this point, machines cannot engage in the deeper investigations that analysts perform once they are aware of potential breaches or once aberrant behaviors have been detected. But innovation in risk management and insurance should certainly include machine learning. Humans simply cannot gather and analyze data as fast as machine algorithms can. Incorporating ML as a solid part of cybersecurity just makes sense.

Realistic Expectations for Insurance in 2020

Visualize a meter that ranges from No Change (1) to Total Transformation (10). I expect the actual changes to the 2020 Insurance Industry meter to register somewhere between 1 and 2.

Thinking about insurance industry trends for the next year was always a fun exercise whether I was at the META Group, Financial Insights (IDC) or Ovum (now Informa Tech, I believe). Each trend captured the opinions from our team of technology-focused insurance industry analysts concerning what we thought would occur over three to five-plus years for each specific issue. Once the trends were finalized by the team, our trend report drove a significant part of our research agenda for the following year.

Instead of trends, I decided to publish my realistic expectations for the 2020 insurance industry:

  1. The League Tables (ranking of insurance carriers) for each major insurance line of business will look the same at the end of 2020 as the tables look at the end of 2019.
  2. There will continue not to be any (statistical or otherwise meaningful) correlation between investment levels in startup insurance firms and any measurable impacts on incumbent insurance firms specifically or the insurance industry generally. (Hype does not equal reality regardless of how much PR digital ink is spewed by the startups!)
  3. Insurance firms will continue in their grand tradition of exhibiting “magic bullet” syndrome: believing that the latest technology or technology application can resolve their major business objectives and can be implemented by using minimal company resources.
  4. Insurance firms, particularly in the U.S. and Europe, will continue to struggle to rationalize the large multiplicity of each of their core administration systems (i.e. policy administration, billing, claims management systems).
  5. Independent agencies (and broker firms) will continue to sub-optimize their operations by not acting in the reality that they are joined at the hip with each of the carriers they conduct business with.
  6. Although insurance firms will continue to recognize the absolute criticality of data, the firms’ various data elements will collectively behave more like useless sludge than a clean and useful resource.
  7. The lack of clean, standardized data will continue to hinder (stop?) insurers from successfully deploying customer-facing (and other market-facing, including producer-channel-supporting) initiatives.
  8. Most insurers will continue to give lip service to providing world-class customer service.
  9. The number of independent insurance agencies and insurance broker firms will continue to decrease as M&A continues in the producer channel, but the number of agents/brokers will remain stable.
  10. 5G, immersion technologies (AR and VR) and enterprise streaming will join the never-ending parade of technologies/technology applications in 2020, already chockablock with other “supposed insurance firm immediately must haves” that include leveraging social media, offering increased functionality on mobile devices, virtual agents/chatbots, interactive video for client onboarding and customer service, IoT, big data, cognitive computing, deep learning and machine learning – all of which technology firms will use as door openers as they reach out to insurance CIOs and CTOs.
  11. Cyber risks will continue to cascade through any device connected to the Web used, owned, leased or otherwise in the possession of society (families, individuals, businesses, federal/state/local governments and the military) adding more pressure on insurers to decide whether or how to profitably offer protection or services.
  12. I’ll continue to hope, in vain, that increasingly more insurance firms will realize the importance of using geospatial solutions as critical components of decision-making, whether the geospatial data comes from terrestrial or Earth Observation sources.

See also: Are You Ready to Fail in 2020?  

How Machine Learning and AI Reduce Risk

Risk management is integral to insurance, but it’s traditionally been an inexact science. Thanks to recent technological advances, however, risk management is about to get a long overdue upgrade.

If an eyebrow is raised, it is likely because the insurance industry has been slow to adopt technology, but artificial intelligence (AI) and machine learning are making headway. The appeal in using data to predict outcomes, drive efficiency and reduce costs has sparked intrigue and curiosity. Tack on the ability to make jobs easier and facilitate claims faster, and even the biggest skeptics, those most resistant to change, are curious about how AI can be applied.

Despite the aversion to tech or potentially costly, time-consuming operational overhauls, AI systems already have been put to work in some of the world’s largest insurance organizations, where they are used to address highly specific issues that have plagued different sectors for years. Now, the time has come to consider how AI can help with risk management.

New Data, New Insights

Much of the information that risk managers value in making assessments is not readily accessible to them today. Data in claim notes, documents, images, even injured worker sentiment requires someone manually poring through files because this type of information can’t be entered or sorted in conventional systems easily. But new, AI-based systems can incorporate and analyze these forms of unstructured data. They make it much simpler for employees — even the least tech-savvy employees — to find and interpret the elements that will be the most crucial to their decisions.

See also: How Machine Learning Transforms Insurance  

Additionally, the more that AI-based systems “read,” the faster and better they learn and understand. Models that leverage unstructured data yield more accurate and detailed analysis, and, by enabling adjusters to make more informed decisions based on data, organizations can reduce the severity and frequency of claims. This makes everyone happy. The industry can move light years forward by delivering this kind of data and analysis to risk managers’ fingertips whenever they need it.

Group Analysis

Another way in which new AI-based systems can help risk managers is by analyzing data across groups. It’s far more efficient to grasp what is happening across a portfolio or set of claims when a machine generates a report vs. reading file after file to formulate an opinion. With new tools, risk managers easily can look across very large datasets to see what’s happening collectively. They can determine the macro impact instead of relying on an isolated view of a single claim. In addition to the time and resource advantages, AI-based software spots trends and outliers that cost money unnecessarily.

Collective View Vs. Limited Project Basis

AI models also are able to draw on a wealth of historical information — information that is constantly updated. This stands in contrast to the way the world of risk works today, where most analysis is conducted on a project basis. The project ends; so does data collection. Important information is often lost in the lapse between projects. Modern AI systems solve the issue by persistently refreshing to ensure updated reports can be ready on demand. The result is a much richer and more realistic picture of what is happening in an organization’s claims.

Power of Prediction

The gold for risk management, however, lies in AI-based solutions’ ability to predict outcomes. AI applies science to risk management based on an incredible number of data points that should be considered in helping teams prepare for the future. Modern systems show risk managers the behaviors that need to change, assumptions that are incorrect and what things will look like if they continue to follow the present course. This information is so important because every customer or risk manager has observed different behaviors, which shape their views and how they conduct their jobs. AI systems parse all of this behavior to give a far more comprehensive view. Systems then can alert users to adverse trends that are developing so that teams can adjust accordingly. This not only decreases the lifespan of claims but potentially can save millions of dollars.

To gain the best predictions, however, it is necessary to use a platform solution that lets users easily gather insights and create models that learn from the entire industry, not just their own data. They then apply that information to a specific customer’s data. The more data a system can analyze, the more patterns come up, yielding more precise and valuable predictions.

See also: Key Challenges on AI, Machine Learning  

Armed with an abundance of data that is simple to access and interpret, claims managers can do their jobs faster and more easily than ever. This can make a potentially huge positive impact, not only on their own organization but also on the larger sector. As machine learning and AI-based technologies mature and are more widely adopted, the industry will become more exact. Costs will drop, and efficiency will improve, ultimately helping to transform the insurance industry.

As first published in WorkCompWire.

Leveraging Data Science for Impact

As insurers strive to become more relevant to their customers and more efficient, they have embraced the strategic importance of their data. Insurance companies have been using various data streams to predict property damage and loss for generations. But while they have been collecting increasingly large stockpiles of consumer data, until recently they have lacked the tools and talent to operationalize it — particularly with the level of transparency required by regulatory bodies — to drive better products and services and operational efficiencies.

Advances in AI and machine learning have enabled insurers to improve the customer experience and boost policyholder retention while cutting claims handling time and costs, eliminating fraud and protecting against cybercrime. These new tools and platforms have generated increased interest in using data science across the industry, and insurance companies have been investing accordingly.

According to a recent study, 27% of large life/annuities insurers and 35% of large property/casualty insurers are expanding their data science efforts to some degree, while 13% of large life/annuity insurers are piloting an initiative. Midsize insurers are similarly active in the space, with 20% of life/annuity carriers and 24% of property/casualty carriers looking to expand their data science efforts.

See also: Turning Data Into Action  

But while investments in AI are growing, insurance organizations are often finding that their existing analytics and business intelligence technology and talent aren’t capable of meeting their current and expanding needs. Challenges in resources, technology infrastructure and the ability to operationalize models quickly and efficiently can prevent insurers from fully leveraging AI and data science to drive business impact. To overcome these challenges, and maximize the ROI on AI investments, insurance companies must look to innovative solutions such as data science automation.

While data science is becoming a valuable tool in the insurance industry, implementing a data science program is not easy. A typical enterprise data science project is highly complex and requires the deployment of an interdisciplinary team that involves assembling data engineers, developers, data scientists, subject matter experts and individuals with other special skills and knowledge. This talent is scarce and costly. This is neither scalable nor sustainable for most insurance organizations.

Data science automation platforms fully automate the data science process, including data preparation, feature engineering, machine learning and the production of data science pipelines – enabling insurance organizations to execute more business initiatives while maintaining the current investments and resources. Data science automation allows data scientists to focus on what to solve rather than how to solve. End-to-end data science automation makes it possible to execute data science processes faster, often in days instead of months, with unprecedented levels of transparency and accountability. As a result, insurance organizations can rapidly scale their AI/ML initiatives to drive transformative business changes.

There are several key areas where data science automation can make a big impact in the insurance industry. For increasing operational efficiency, AI-based automatic underwriting and claims management will be a major trend that we will see in coming years. In customer relationship management, AI will be used more frequently to help profile customer behaviors, helping insurers to get a better and deeper understanding of their customers’ wants and needs. This, in turn, will help to drive revenue growth.

See also: Role of Unstructured Data in AI  

In the near future, data science and AI will be widely implemented in the insurance industry, and the barrier to adoption for data science and AI will become low. Once this happens, accumulated critical use cases will be key differentiators for insurance companies implementing these technologies. Data science automation accelerates the data science process, enabling insurers to explore 10X more use cases than with the traditional method of data science. Early adopters have already started to leverage automation to scale their data science initiatives.