They want to pave the way for sober traffic with AI tools

Between 25 and 30 per cent of road accidents in Sweden are alcohol- or drug-related. Breath tests are important for detecting drivers under the influence of alcohol, but to test for drunk driving the police still rely on old-fashioned blunt methods using a torch and a pencil as their main tools.

"They shine the torch to assess pupil size and ask the suspected drunk driver to follow a pencil with their eyes to see if their eyes twitch when they look away. As a police officer, you learn to make judgements, but you also rely a lot on external features, which means you may not always be as objective in your assessment," says Stefanie Najafi, former police inspector and founder of Sightic.

She and her partner Jenny Johansson were working together in the Swedish Security Service when they came up with the idea for a digital tool to detect the influence of alcohol and drugs. In 2021, they left the Swedish Security Service to found Sightic Analytics.

"Helped us dismiss leads"

At an early stage, Stefanie Najafi and Jenny Johansson contacted Olof Mogren, head of AI research at RISE, whom they had met at a workshop. The plan was to start a research project funded by Vinnova, which required a research partner. The project involved collecting and analysing data from people under the influence. This data was then used to build the AI software that would detect whether a person was under the influence.

"There were quite a few of us in the company at the time, just three or four developers. This is where RISE could contribute with their development expertise. Our developers experienced great technical benefit from working with the RISE team. The group also provided new input and ideas, helping us to discard tracks that didn't work and move forward with those that had potential. We moved a lot faster with their help," says Stefanie Najafi.

"During the project, Sightic started to build its own technical department, hiring a technical manager and several engineers. It was a great way for us to work, providing deep technical knowledge and experience while the client built the expertise they needed in-house. When the project was finished, the collaboration continued for a while until Sightic was able to stand on its own two feet," says Olof Mogren.

First up: Mining in South America

The Swedish Transport Administration's report "Action plan for safe road traffic 2022-2025" states that it should be possible for the police to carry out random checks for drug use in road traffic. It also stresses that drug testing requires reliable and effective tests. But it is not in Sweden that Sightic's AI tool will first be used.

"A South American mining company that had seen us online contacted us and was interested in our solution. The company had problems with employees under the influence of alcohol and cannabis and needed a flexible tool to test many people at the same time. That is why South America and the mining industry became our first market," says Stefanie Najafi:

"Our challenge was that the tool is a bit too sensitive, based on the laws that apply in each country. This has taught us that we need to adjust the threshold of our algorithm so that it is adapted to the organisation where it will be used. In Sweden, for example, the threshold for alcohol is 0.2 per mil, but in other countries it is 0.8. Acceptance of cannabis also varies from country to country.

In addition to its presence in the mining industry, Sightic has also approached the automotive industry, where it has developed a drug-impaired driving detection solution to be implemented in vehicle software.

Importance of RISE reference

Stefanie Najafi stresses the importance of trust when working with AI, and with something as sensitive as determining whether people are affected. 

"Being able to point to a well-known research institute like RISE when presenting the solution to potential customers adds weight," she says. Of course, Stefanie and Jenny's professional background is not unimportant.

"Many old colleagues have said, 'Why didn't I think of this?'. It is obvious to anyone involved in drug testing that there is no way to get away from subjective judgement with the current method and that a different tool is needed," says Najafi.