Innovations in Data Science

At MNC we truly value the skills our data scientists and actuaries with data science skills, bring to the table. We know that their work transcends the entire value chain from lead generation through to claims stage and all the areas in between. There are many examples of innovative uses of data science techniques in insurance that emerge daily, here are just a few of them.
Pilot Bird

1. Pilot Bird

Pilot Bird – uses social data points to gain insights into lifestyles, building personal profiles that inform the sales, underwriting and claims processes. In the sales process they are used to identify people who have (a) recently had a baby or become pregnant and (b) people who have recently been victims of cybercrime.

In the claims process they use lifestyle data to reduce fraudulent claims. For example, they look at social data points to identify discrepancies between policyholders disclosed lifestyle and their actual activity and also to identify people with second jobs who are not disclosing them whilst claiming disability income benefits. Find out more >
JDMC logo


In Japan, there is a smartphone app being developed by the Japan Medical Data Center (JMDC) that instantly measures a person’s “health age”. The user takes a photo of their health check-up data, that data is digitized and sent to JMDC, where it is checked against a database of 1.6m data points on health exams and medical prescriptions. The app then instantly calculates the health age which is described as an indicator of a person’s overall well being and provides advice on how to become healthier. Find out more >

3. Lemonade

Lemonade uses a fully digital claims process that captures a large amount of high-quality data. Customers log claims by recording a video explaining what happened. Lemonade’s AI (Jim) analyses the videos, looking for signs of fraud and then makes an immediate decision to either pay the claim instantly or divert the claim to a human for further inspection. The claims video can be re-played multiple times for human claims adjudicators to inspect carefully, unlike traditional claims.

This method of handling claims is both efficient for the customer (fast pay-outs) and protective for the company (flagging suspicious claims). But most importantly it captures all the metadata easily missed by humans whereas with traditional insurance companies, the data identifying past fraudulent claims is generally not fed into a system that gets smarter and better at handling future claims, but the AI method does. Find out more >
MNC logo

4. MNC

MNC has been working on its own innovation in recent months. We have been developing a machine learning model that can classify free text claim causes to a set of analysis-relevant categories. This significantly reduces the time taken to clean claims data. Our model currently has a testing accuracy of the model is currently 84%, with precision and recall of 78% (% of true positives) and 68% (% of all positives – true or false) respectively. The final model was developed by fine-tuning on large transformer-based language models. Future improvements to imbalances between classes will be solved as the model receives more data and this will improve the accuracy too. Find out more >

Come and talk to us at the Data Science Seminar or contact us via the website