UPLIFT: Project Update – August 2020

Sam Chapman - August 12th, 2020

uplift logo

In February 2019, The Floow started working on the UPLIFT project to investigate areas relating to the use of telematics within insurance propositions to inform risk understanding and policy pricing. This 21 month project, funded by the Industrial Challenge Strategy Fund, is designed to strengthen the UK’s position in the motor insurance sector by building upon telematics capability. 

The UPLIFT project aims to empower improved and fairer insurance products to make mobility safer and smarter for all. We do this by looking at areas in risk understanding that may contain bias, unfairness or a lack of transparency for end-users to understand and improve any areas that have any potential to be unfair. 

With this project, we are primarily seeking to enhance technologies for improved and fairer risk understanding as well as building the means to ensure new technologies deliver gains in certain areas, including:

  • Bias analysis, identification and reduction
  • Increasing transparency, inclusion, privacy and fairness for consumers
  • Setting new industry standards in ethical data handling

Our recent work on the UPLIFT project

Our work on the UPLIFT project is helping us to build a better future and we continuously investigate many areas relating to telematics and insurance in an attempt to make added gains to these areas. Over the last few months, we have been focusing our attention on three key areas in particular, they are: crash detection, driver/passenger recognition and data quality.

1. Crash Detection

The ability to detect crashes using telematics data is becoming increasingly important to insurers but when using data and algorithms, it is important to ensure that they are as accurate as possible to reduce the likelihood of non-crashes, such as an emergency stop due to another vehicle dangerously pulling out of a junction, being flagged as a crash and potentially impacting on a driver’s risk profile.

As a result, we have been looking at means to enhance high severity crash prediction to improve the detection capabilities of crash algorithms and their ability to triage accidents based on the characteristics of the crash detected. This research can be used to influence future work on crash detection algorithms which will continue to improve their predictability and ensure that they are utilised within all telematics insurance products as an essential safety feature.

2. Driver/Passenger Detection

As more insurers opt to use smartphones as the sole data collection method for their insurance telematics propositions, it is increasingly becoming important to refine understanding of when a driver has driven a journey versus when they are a passenger.

UPLIFT’s work in this area focuses on analysing anonymised driver data to look for enhanced features and patterns in available data from smartphone devices. This includes those immediately before, during and immediately after test journeys. By investigating this data and investigating new features from them we can uncover potential new insights that can help to accurately detect if a policyholder has driven their vehicle or not.

3. Data Quality

All of the work done on the UPLIFT project is underlined by a strict focus on data quality to ensure a clear and fair understanding of journeys, which fairly and correctly assess a driver’s risk based upon the real behaviours exhibited behind the wheel. Rare cases however may have seriously compromised data due to either lower quality devices or signal disruptions to them. To understand these potential impacts and mitigations to them, we study a large range of anonymised journey data and test cases. 

These include poor scoring journeys recorded on lower quality devices and those with extreme  signal disruptions. Investigations like these help to ensure scoring approaches are fault tolerant and where appropriate can help to prototype potential new mitigations making scoring smarter and safer, even for extreme and unusual data quality issues.

We also look to see if journeys more disrupted by lower quality data can affect the overall balance of driver scores and how drivers engage with and understand such scores. If poor quality data can negatively affect a driver’s score, it is not only unfair to the driver potentially impacting on their premium costs, but it may also affect their engagement and trust with a policy. 

All of this can potentially put barriers in place to stop drivers from interacting with telematics’ insights and feedback provided post-journey. Feedback and insights are designed to help drivers improve their driving behaviour, as well as overall road safety, therefore it is important that we understand how data quality can affect telematics and feed this research into future product development to ensure telematics remains an important and trusted tool for motor insurers and their policyholders.

Over the last 18 months, we’ve investigated a number of areas relating to telematics and its use in motor insurance, and how we can enhance this to ensure it is a tool free of bias and which is also transparent and ethical. 

During the remaining months of the project, we will continue to grow our fine-grained understanding of mobility and driver risk with the aim to enable fairer, personalised and easier to understand telematics scoring and insurance products for all drivers. The work we have undertaken, and continue to undertake, is influencing future product development which will benefit insurers and policyholders by providing a better and fairer understanding of driver risk.

For more information about the UPLIFT project, visit the UPLIFT page.

Share this article