Code4Thought, that offers software engineering and AI advisory services and the global leader in manufacturing software to monitor aged care facilities, Kepler Vision Technologies, have announced the results of an independent screening of the existence of prejudices performed by the former. The results confirmed that Kepler’s Artificial Intelligence (AI or AI) algorithm follows an unbiased approach to caring for and overseeing patient wellbeing in a professional care environment, regardless of the sex of the patient.
Based on PyThia technology, Code4Thought examined the Kepler Night Nurse (KNN) algorithm for the possibility of bias in its decisions/notifications. Based on the detection of human activity using machine learning, the innovative KNN algorithm, immediately alerts the staff whenever they detect a patient’s fall in the care unit or when they feel discomfort, in order to receive the care they need, at the time they need it. This process eliminates the need for constant checks on patients and the lost time associated with false alerts provided by other monitoring systems.
As the world’s first fall detector based on machine vision, and categorized as a medical device, Kepler’s deep learning algorithm, Night Nurse, requires careful testing to ensure that objective and accurate results are produced for each patient it attends.
For this reason, the company Code4Thought was selected, whose PyThia technology is based on the standard – ISO-29119-11 designed to control AI systems to ensure their correctness and accuracy. The result of the analysis proved that the KNN algorithm is able to produce correct and fair predictions regardless of the patient’s gender. Code4Thought’s evaluation was completed with advice provision on how Kepler Vision Technologies can ensure the use of datasets that promote protection against prejudice and inequalities in healthcare, which consists of already vulnerable demographics.
“As the importance of decisions made by AI systems increases, the need to ensure that they operate objectively and transparently increases. Two properties that are vital especially in the field of medical devices, where crises of AI tools can, literally, be a matter of life and death.
The European Commission’s proposal for legislation on AI systems makes it clear that companies using deep learning algorithms with high complexity and opacity can take further steps to build trust in AI systems. By working with Code4Thought, Kepler Vision confirms its dedication to improving the lives of all patients to whom its technology is applied, regardless of the individual differences it that they have with each other.”, said Harro Stokman, CEO of Kepler Vision Technologies.
“Especially in the medical field, algorithms that have been configured in the wrong way lead to inaccuracies and prejudices that can have potentially devastating consequences. For this reason, for any company using AI systems, ensuring that its algorithms meet the highest level of independent control should be a priority.
Working with teams and organizations like Kepler that value developing reliable AI technology that businesses and individuals can trust creates joy and satisfaction. It is encouraging to see that behind a high-quality AI system, there is a corresponding high-quality team of engineers working with the professionalism required by data science.”, commeneted Yiannis Kanellopoulos, founder of Code4Thought.
