Unlocking the power of AI to make babies
BJSS is a leading digital consultancy with 10 years’ healthcare experience designing and delivering a range of critical NHS systems and services, including NHS Spine, the e-Referral Service and the NHS App.
Founded in Leeds in 1993, BJSS has grown into a team of over 2,000 engineers, analysts, designers and consultants, based out of 26 offices globally, including Glasgow, Edinburgh and Aberdeen.
We work with national NHS bodies, regional and local health and care providers, third sector organisations and a number of health tech companies, both established and up-and-coming.
If a project is interesting, challenging and supports the NHS and the health and wellbeing of the population, we’re usually interested!
Here is just one example of the type of work we do at BJSS…
Introducing artificial intelligence into IVF laboratories
Most patients undergoing IVF treatment, hoping to have a baby, have several embryos which the embryologist must choose between.
The embryos often look very similar. Selecting the ‘best’ embryo for transfer remains pivotal to the success of a patient achieving a pregnancy when proceeding with IVF.
CARE Fertility, a leading UK fertility clinic group for IVF treatment, designed an embryo selection tool based on time-lapse image capture, identifying key events in the developing embryo that are used to score each embryo’s potential for live birth outcome.
These key events are currently identified and annotated by manual inspection of thousands of time-lapse images by trained embryologists.
CARE approached BJSS to explore the potential for artificial intelligence (AI) to transform the embryo selection process.
AI has the potential to save time, improve accuracy and repeatability, and is already used effectively in several fields of medicine – why not embryology?
BJSS was engaged to investigate whether AI could be used to annotate with the same or better accuracy as expert embryologists’ manual annotation.
BJSS developed a robust deep learning model, applying modern techniques in computer vision and time series analysis.
The model draws from a significant historic data set of over 500 million images and 2 million manually annotated events from eight UK CARE clinics.
The Automated AI annotation model has been developed, validated and tested. When applied to CARE’s embryo selection tool, it performed as well as manual annotation.
The CARE/BJSS team is now working towards productionisation and deployment to CARE’s 14 IVF laboratories across the UK and Ireland.