Embracing artificial intelligence (AI) technologies is a top priority for many enterprises across different industries. Financial services group Prudential taps AI in insurance whereas Ping An Insurance’s subsidiary Ping An Technology uses AI in healthcare, finance, and automobile insurance.
Prudential: AI in medical claims
Prudential plc’s group head of AI, Michael Natusch, shed some light on how AI can be applied in traditional financial services companies like Prudential at the recent Cloud Expo Asia. He shared how AI helps to make hospital claims easier, faster, and round the clock.
Natusch said it normally takes nine days to process an individual hospital claim in Hong Kong, starting from receiving a claim in paper format from a policyholder, evaluating it by a claims assessor, approving the claim to issuing payment by the finance department.
Prudential built an AI system with a chatbot to deal with insurance claims management, enabling customers to submit their hospital claims electronically and get instant claims status updates. “We reduce the claims processing time from nine days to 2.3 seconds,” he said.
Customers can use the chatbot to scan and upload their claims documents like hospital receipts and medical records. The documents will then be sent to Prudential’s claims decision engine, which assesses whether the claims are valid or not, and notifies customers of their claims status in real time. If customers want to talk to an agent, the chatbot can bring the agent into the conversation.
Natusch said the AI system was trained in less than two weeks to make character recognition of hospital claim forms reach over 80% accuracy rate. Hospital claim forms usually contain handwriting data in English, Chinese and numeric digits.
“We built a neural network for handwriting recognition,” said Natusch. “On average, whether it’s Simplified or Traditional Chinese characters, we have about 85% accuracy rate. For English characters and digits, the accuracy is around 89% and 98% respectively.”
Ping An: AI in automotive, medical & finance
Ping An Technology CEO Ericson Chan shared how AI is embraced in automobile, medical, and financial areas at the Internet Economy Summit 2018.
Ping An Technology uses AI for vehicle damage assessment and claims. Chan said the company accumulated millions of photos of damaged vehicles including internal damaged parts, and used them to train its AI system.
If a customer’s car got into an accident, the customer can use Ping An app to contact Ping An call center. The call center can verify the vehicle owner identification based on the incoming phone number and the caller’s voice recognition. The customer can also use the app to upload a photo of his damaged car. The photo is then cross-referenced with the company’s database of millions of photos of damaged vehicles to assess the damage level of the car, and recommend the insurance claim amount.
“Sometimes cars get damaged in a typhoon. Customers can get the insurance claim in their bank account before a typhoon left,” said Chan.
Disease prediction and treatment
Through the use of AI and big data analytics, Chan said Ping An is capable of forecasting infectious and chronic diseases such as flu. Governments can use the forecast information to warn the public and take preventive measures.
“We make traditional forecast method more precise,” he said. “With our multi-layered trend analysis, we provide disease trends prediction and prevention suggestions for a specific region or population.”
Take flu as an example, AI can forecast the risk level of flu infection in different districts. Like weather forecast, a flu warning heat map can be created to indicate geographic spread or estimate flu activity.
On the diagnosis and treatment front, Ping An uses AI to help with medical image reading for healthcare providers, auxiliary diagnosis and treatment services for patients.
Chan said Ping An scanned, read, and learned plenty of historical X-ray images such as lung modules to build an accurate lung module engine. Based on the position, type, and size of lung modules and combined with medical history, the engine can analyze the possibility of lung disease for a patient and provide treatment advice to doctors.
“We help doctors to scan through x-rays that normally take 30 minutes but now shrink into one second,” he said. “Hong Kong healthcare ecosystem is under stress. AI can release some stress from doctors and nurses.”
Ping An developed various cognitive AI technologies including facial, micro-expression, and voice recognition to help with identity verification such as opening bank accounts, applying personal loans, and mutual funds. This helps prevent frauds while speeding up processing time.
“We’re handling one million phone calls in our call center every day, so there’re enough voice records to train our AI system to get smart,” he said. “We hit the world record with 99.8% accuracy on facial recognition last year. We can also recognize 54 micro movements on a face and determine a person’s mood.”
Take personal loan application as an example. If a customer applies personal loan on his mobile phone, a bank’s agent can use video chat to capture the customer’s facial expression to quickly determine his mood at that moment. For instance, if the customer is nervous according to the micro expression detection when asked whether he works in a specific company, the agent can ask further questions to verify his identity.
“This is how we leverage AI. It delivers value to an organization and protects customers,” he said.