Insurer turns to AI to streamline application process
Zurich Insurance is planning to use artificial intelligence tools designed with University of Technology Sydney to help process the mental health side of insurance applications.
Zurich Insurance is planning to use artificial intelligence tools designed in conjunction with the University of Technology Sydney to help process the mental health side of insurance applications in Australia.
The move, which comes after more than six months of work with UTS Rapido, a research and development hub at the university, aims to speed up the processing of insurance applications by people who have disclosed a mental health condition, removing the need for them to visit a doctor for a report on their condition.
“Some 25 per cent of applicants for life insurance disclose a mental health condition, which is the biggest reason that we ask them to get a doctor’s report,” Zurich Australia’s head of retail Jacqui Lennon said in an interview with The Australian.
“It is the biggest driver of making the process a lot longer than it needs to be. It can take as long as 22 days.
“We can use machine learning tools to help make a decision and give the applicant an outcome straight away.”
While Zurich insists that human beings will still be involved in the decision-making process on insurance applications, it says the AI tools, which have been developed from anonymised data from seven years of past applications, will allow much faster processing of the data that people wanting insurance provide on application forms.
Mr Lennon said using the AI models could reduce the time taken to process an application from 22 days to less than a day.
She said removing the need for the applicant to see a doctor also reduced potential privacy concerns the person may have.
“Mental health is becoming more prevalent from society’s point of view and is a big driver of our claims,” she said.
“Thirty per cent of the claims in some of our policies are for mental health reasons.
“We wanted to do something around how we can make mental health an easier condition to underwrite.
“We asked UTS to help us understand the predictors of different decision outcomes focusing on mental health.
“It’s given us a greater ability to reduce the time taken to get to a decision and improve the customer experience by allowing them to maintain more privacy around their mental health information.”
The process has developed six predictive models that can be used to assess information provided by applicants who disclose a mental health condition when they apply for insurance.
The models reveal a number of conditions that are associated with people more likely to have a mental health problem.
These include participation in recreational activities (with lack of participation in recreational activities more likely to be associated with mental health issues), ongoing health issues including chronic pain, cancer and heart disease, time spent in hospital or having medical treatment in the past five years, and having parents or siblings with health conditions such as diabetes, cancer, mental health issues and heart disease.
While insurance companies can reject applications based on prior health conditions, they can also choose to issue policies with specific exclusions or loadings.
“The process can allow us to make better decisions so we can protect more Australians,” Ms Lennon said.
Lennon says a human being will oversee the final decision on policies.
“We have a philosophy of having a human in the loop to look at the outcome to make sure it makes sense,” she said.
Herve Harvard, the executive director of engineering and IT solutions at UTS Rapido, who worked on the project for Zurich, said insurance application forms could involve significant amounts of data, including a person’s medical history, family background and personal circumstances.
“Machine learning allows you to process thousands of data points which would not be possible for a human being to do,” he said.
“With machine learning you can find trends and correlations and develop a machine model.” Mr Harvard said UTS became involved in the project because of its long history of AI research.
It established the Centre for Artificial Intelligence in March 2017 as part of its School of Computer Science in its faculty of engineering and IT. Now known as the Australian Artificial Intelligence Institute, it has 35 academic staff and more than 200 PhD students.
Mr Harvard said the insurance process did not involve generative AI.
He said the next step could include the use of broader data sets to develop predictive models around the potential for mental health claims.
The move comes as the insurance industry is grappling with a rise in mental health claims in Australia.
Mental health is the third most common cause for a claim with Zurich, accounting for 20 per cent of claims and resulting in $255m in payments in 2023. For some of its retail life insurance products, mental health has become the No.1 or 2 reason for a claim, making up some 35 per cent of claims.
Ms Lennon said there was potential for Zurich to use AI-developed models for more parts of its business, but it had begun with mental health because it was a growing issue.
“Mental health disclosure rates and the need to avoid having to go to the doctor is one of the biggest problems and the one we wanted to tackle first,” she said.
She said the models could help develop programs to promote mental health among customers.
The insurance industry is keen to streamline the application process as it has seen a reduction in the number of people buying life insurance-related products in recent years.
Ms Lennon said the sharp fall in the number of financial advisers in Australia, from some 28,000 in 2019 to around 15,700 today, had meant that fewer Australians were talking to advisers and considering taking out insurance.
She said the rising cost of living led to people reducing their insurance cover to save money.
“We want to improve the customer experience and reduce the friction (in the application process),” she said.
“We want to make it easy to provide more cover for Australians.”
More Coverage
