Deep Learning For Insurers

There is a lot of excitement in the insurance industry about the potential impact of deep learning on several areas of the industry including the customer experience, underwriting, distribution and claims processing.

You may wonder how deep learning will help insurers. Deep learning is a type of machine learning that uses multi-layered neural networks, this means, that machines can now learn new things as they operate, such as how to recognize images.

Can deep learning accurately predict if someone will end up in a hospital?

Deep learning in the last few years has not only begun to revolutionize computer vision, speech recognition and natural language process, it has already started to disrupt the insurance industry. With huge volumes of structured and unstructured data coming in from a variety of sources – including newly accessible third-party databases and the Internet of Things — insurers can now use deep learning not only to assess text and images, but to recognize patterns and draw conclusions from those patterns.
So won’t it be cool if someone could accurately predict whether someone will end up in a hospital soon? It would help oneself in preparing for the eventuality. The preparation can range from any possible preventive measures like medication etc. to simple mental preparedness to just face it! In fact this was the exact same (3 million US dollar) question that was posed by the healthcare industry for the machine learning community a while ago.

(Also check out our blog post on how one can improve their health with the help of Internet of Things. Click here to read more)

What are the challenges?

Human health is quite complicated. It depends on a lot of factors like human’s life style, genetic background, diet etc. Some of these factors – like human DNA – in themselves are very complex. Some of these factors are difficult to represent. The relationship between these factors and human health is difficult to capture. E.g. how human life style affects human health is quite intricate and is a research domain in itself.

Bridging the gap

However with the recent advancement of technology a lot of things have become a possibility. It is now possible to complete the DNA sequencing: collect and crunch a lot of data like continuous heart bit, pulse rate, facial expressions indicating potential medical condition etc: uncover correlations between common factors and health conditions. The simple wearables can now give real time analysis of one’s nightly sleeping pattern. The health conditions due to sleep deprivation can thus be monitored and even prevented real time. The image analysis in conjunction with self-teaching artificial neural network is helping in learning about heart problems. The sugar level in chronic diabetes patients can not only be monitored real time but also controlled automatically administering the corresponding insulin dosage avoiding any calamity.

Importance to medical insurance industry

If a patient is already using any of these auto monitoring or correcting tools, the insurance company can provide some discount towards the medical insurance policy premium. This discount is mostly due to reduced risk associated with the patient equipped with the tools. The reduced amount goes towards the cost of the equipment.

Insurance company would like to lower its patient cost as much as possible. Towards this end it monitors the waiting time of a patient at the doctor’s office. If the patient  leaves  due to  excess waiting  time and  is required  to return  later,  it might aggravate the patient’s condition and  increase  the  insurance cost.  It depends on the waiting time, patient’s condition at the time of the visit, doctor’s and patient’s calendars and the next available appointment.  Considering all these factors the system generates an alert when such a patient is likely to leave the office so that the doctor can attend to him immediately.

Health has the highest priority and we expect the highest level of care and services regardless of cost and are more emotional and ideological about the industry. With deep learning it is all about correct diagnosis, accurate predictions, and personalized medicine to administer the most effective medicine for an individual by avoiding any side effects he/she can have at the same time.

This can have an enormous impact in claims as deep learning applications can help insurers, not only to assess claims at high speed, but to identify anomalies indicating potential claims fraud. For example, If an insurance company, undersigning a medical policy for an individual, is aptly equipped with technology that can accurately predict whether the individual will end up in hospital, the insurer can rise his/her premium of insurance covering. At the same time it is crucial not to be pessimistic about the prediction and overcharge the patient. The accurate prediction comes to rescue here.

Comments 1

vinod borole

May 18, 2017

With recent Google IO 2017 and launch of Cloud TPU deep learning will have more focus where such problems would be easier to solve; It is observed that TPU is 15x to 30x faster than contemporary GPUs and CPUs