April 26, 2018

Spotting insurance claim and eye anomalies using GPU-powered AI


What does a fraudulent insurance claim and eye problem have in common? Well, both can be identified, thanks to powerful NVIDIA GPUs deployed by AIDA Technologies, a Singapore-based artificial intelligence (AI) startup which began in 2016.


Staffed by a team of award-winning data scientists from leading research institutions in Singapore, AIDA (short for AI-driven analytics) focuses on using its strong analytics capabilities to develop innovative machine learning solutions for the banking, insurance and healthcare sectors. These customisable solutions bring a new ease to developing high precision analytical solutions for each sectors.

The startup adopts an approach that seeks to understand customers’ challenges and deliver a very good deep learning solution to tackle those problems. Its solutions are aimed at helping banks and insurance firms increase revenue, mitigate risks and ensure compliance by identifying risky behaviour.

To do this, it relies on NVIDIA GPUs powering Amazon Web Services’ cloud.

When you talk about heavy duty computing, there’s only one company out there who can help us, which is NVIDIA. I think they are making such a big inroad in deep learning which is a very important tool used in analytics,” said Dr Tan Geok Leng, CEO of AIDA.

The company leverages GPU computing in two areas – text mining for the banking and insurance field and eye imaging in the healthcare industry.

 Identifying good customers
In banking, we help banks identify who a good customer is,” said Tan.

AIDA’s deep learning solution uses predictive analytics to determine whether a customer will stay on or move away. In the highly competitive market, this is vital because it costs banks lots of money to attract customers and customer retention is a priority.

Our solution can also predict which customer will stay or go. If a person is likely to go, we can also predict which incentives can be offered to hold him back using a profitability simulator,” said Tan.

Recognising fraudulent claims
Insurance companies are another of AIDA’s target market with some of the biggest names in Singapore among its customers.

AIDA helps insurance firms to automate their claims processing in a short period of time. One of its customers, Prudential, announced in November that it can shorten claims from days to mere seconds with help from AIDA’s deep learning solution.

The solution can also help to identify fraud. With 100,000 or 200,000 claims a year, it can be difficult for insurance firms to know which cases are unusual and possibly fraudulent.

We can do what’s called cohort comparison and look for outliers such as those who seem to be claiming a little bit more or coming back more frequently. We also identify agent misconduct such as those who have a disproportionate number of clients who make claims or claim too fast,” said Tan, whose company has customers in Singapore, Malaysia, Indonesia, Thailand, and India.

The solution can also spot hospitals that take longer to treat patients or those that have a higher rate of people returning for more treatment.

Accelerating eye screening
Eye screening is a big issue as it is a time-consuming process. Compounding this problem is the lack of specialists who can do the screening.

A healthcare company in Singapore is using AIDA’s GPU-powered solution for remote eye screening to expedite this process and get more people screened. The solution can help detect defects faster so that patients can see an eye doctor before their eyes get too bad.

Only three in 10 persons have an eye problem. If we don’t use a GPU to do the eye screening, it takes 15 to 20 minutes to read an image because the image is very complex. With deep learning, we can read the image in seconds,” said Tan.

Certain standing
AIDA is a member of NVIDIA’s Inception Program, which helps accelerate startups pushing the frontiers of AI and data science.

Tan is pleased that his company is part of the program and could participate in the GPU Technology Conference in San Jose in March.

When you tell people that you are in the Inception Program, people know that you have a certain standing,” he beamed.

0 comments:

Post a Comment