Last week I was excited to launch our new ‘The AI opportunity’ video series which will surface novel insights from our most advanced AI projects.
The AI renaissance: why it’s taken off and where it’s going
The first video focused on using AI in high-trust applications, using deep learning algorithms to automate visual inspection on a manufacturing line. This is a requirement we see from clients across diverse industries.
This week I want to focus on another hot topic - how to overcome a lack of quality data to train deep learning models.
In the video below, I share insights from working with a global medical device company to invent deep learning algorithms to amplify clinical trials data by 30x. This allowed their data science team to achieve greater levels of performance from their AI systems, which will directly improve the experience for patients with chronic disease.
I hope you found this interesting. If you’d like to discuss some of the techniques we are able to develop for projects similar to this, or any other business challenge needing a unique AI solution, please get in touch at email@example.com
The next video, out on 19 May, will discuss how to ensure your investments in AI pay-off. With many of our clients making significant investment decisions on the back of AI, it’s critical to get right.
You can find more about our work in AI and Analytics here.
Today, I want to talk about the request many of our clients bring us – how to overcome a lack of quality data to train deep learning models.
We recently worked with a global medical device company and invented some world-first algorithms to amplify clinical trials data by 30x. This allows their data science team to achieve greater levels of performance from their AI systems which will directly improve the experience for patients with chronic disease.
I’m Tim Ensor, and I lead the AI Capability at Cambridge Consultants.
This is the second of a series of short videos on ‘The AI opportunity’. I’m sharing our most recent AI developments, the challenges they address for industry and the opportunities they are creating for the ambitious businesses that have deployed them.
The data science team at our client were working hard to develop algorithms which would calibrate their devices for the range of effects cause by normal variances in the manufacturing process. Collecting real-world data from clinical trials is costly and time-consuming and so, as is often the case, the team didn’t have enough clinical data to manage all the variables they were trying to control.
To solve this issue, the client team started to explore the use of advanced generative AI to amplify the training data they had.
Working alongside the client’s research team, we created new tools to generate over 30 channels of synthetic timeseries biomarker data to accurately represent the target population of patients and sensors, including variations in human behaviour and product aging.
As part of the project, we also delivered training tools and material to enable the team to onboard new members into working with these advanced AI methods.
By taking a completely new approach, we demonstrated that we could amplify their clinical trials data up to 30x and enhance the performance of their algorithms and products. This also allowed the client to completely rethink the design of their clinical trials. It means that in future, they can be much more efficient and targeted in collecting this critical, but expensive real-world data.
If you’d like to talk more about using advanced AI for massively amplifying algorithm training data or any other business challenge needing a unique AI solution, please get in touch.
You can find more about our work at cambridgeconsultants.com/ai.