Apricity's Senior Data Scientist Chloe He discusses the importance of safety in Artificial Intelligence (AI) and the approaches of her own research with Imperial to tackle this task.
The introduction of AI into many areas of our daily lives, including healthcare, carries essential implications for our privacy and safety as consumers. While AI is becoming more commonplace in the fertility sector, from imaging and surgical tools to aids in clinical decision-making, it drives urgency to ensure the safety mechanisms for these tools.
The benefits of AI are clear - an ability to deliver the best, personalised treatment for every single patient and maximise the chances of treatment success. Traditional approaches can be prone to human interpretation and high inter-operator variability, especially in the fertility sector, and we are lucky to be able to use the knowledge derived from AI that allows us this standardisation of patient care. But how do we know that AI is indeed working effectively and, most importantly, safely?
Living in the age of information and having access to big amounts of data enables us to draw conclusions from it - but the problem is that there is too much of it and without AI we simply could not use it in any meaningful way. AI is fantastic at spotting patterns in all of this data and building on its learnings based on these observations. On the flip side, this makes AI increasingly vulnerable to the quality of the training data (think teaching false information). Since AI learns through reward & punishment, if, for example, you teach AI that having high blood pressure during pregnancy is a positive outcome, it could introduce ways to increase your blood pressure if no safeguards are put in place - clearly something we do not want.
As a result, Chloe’s research includes introducing strategies to ensure potential drifts are avoided and that the AI we use is safe for patients. At Apricity, we believe that we can deliver a better fertility experience through artificial intelligence, empowering clinicians and making the the most out of fertility data.
Learn more about our groundbreaking work on AI in fertility.
Submit your email to join our newsletter