By Karley Yoder, Vice president & General manager AI, GE Healthcare
There’s no doubt Artificial Intelligence (AI) will change the way we work in healthcare. The opportunities to speed up patient diagnosis and treatment planning, to ease the routine burden on radiologists and to improve clinical confidence are without question. Precision Health is on the horizon. Yet, as with every technological revolution, there comes an element of fear and uncertainty.
Trust and transparency at this and every stage of AI’s march into mainstream healthcare application is vital. We need to be ready to answer questions to alleviate doubt. From where has the data come to build AI algorithms? Is the data diverse enough to represent all patient populations?
Safety and regulation are at the forefront of healthcare; they need to be. And ethics and conduct are not new - the Hippocratic oath was written nearly 2500 years ago and to this day steers the moral principles of our physicians. The opportunities are too great and the risks are too real for us to not get AI right, which is why GE Healthcare has set out its guiding principles of AI.
Dispelling AI cynicisms and myths
First and foremost, AI is not here to replace clinicians. It will automate the mundane and repetitive elements of the role: prioritising which scans to read first; ordering sequences; or flagging anomalies that need a closer look by the physician. It will empower the radiologist to refocus attention on high-value tasks and complicated cases, improving the patient experience and standard of clinical care.
Remember when we had X-ray film and lightboxes? Like our forefathers going from horse and carriage to the motor car, migrating to digital was a leap into the unknown. But look how we’ve managed and normalised to new ideas. Innovation in new healthcare technologies and ways of working such as EMR and PACS over the past 20 to 30 years have enabled us to cope with growing patient numbers and stretched resources. Therefore, we must embrace AI not with doubt but with optimism - it is the tool to futureproof healthcare.
Committed to diversity and transparency
We acknowledge the caution that surrounds the development of AI for healthcare purposes; this is about human lives, not simply designing a decision-making system for a manufacturing process.
The first challenge will be how to obtain the data needed to develop the algorithms that underpin AI. Getting data out of the silos of healthcare mega-repositories and validating it into meaningful insights is key.
Then guarding against mistakes and biases during the development stages is critical as these data sets will define the decisions of the future. Where information comes from and how it is labelled is very important.
The diversity of this data at development stages is also a must to ensure AI mirrors the global populations that it will serve. We live in a multi-cultural world, where diverse data sets and diverse validation partners are essential to neutralising potential bias and best serving the racial and ethnic differences that exist in our patient populations.
Transparency, the accountability of data, is also an important part of GE Healthcare’s guiding principles of AI. When data enters our intelligence platform, Edison, it is sourced and then traced throughout development. This makes algorithms easily traceable, reproduceable and fixable, should problems arise.
The success of healthcare AI means all stakeholders working together
Of course, no one should be acting alone in the development of AI. Ongoing dialogue and mutual respect among academic institutions, hospitals, regulatory entities, governments, industry associations and technology partners will ensure we work toward a common goal. This will also ensure practicalities, policies and methodologies are standardised and that the benefit, safety and privacy of the patient remains at the core.
Recently, the European Commission appointed 52 experts from across academia, civil society and industry to steer a high-level expert group on AI. Its general objective will be to support the implementation of European strategy including recommendations and consultation on policy development and on ethical, legal and societal issues, including socio-economical challenges. Individual governments also have their own committees and steering groups working with industry and relevant stakeholders to ensure a smooth and beneficial path to widespread adoption and regulatory approach.
Ensuring that AI will support future generations
Defining the ethics behind AI and using such principles to drive our work is the key to ensuring that a culture of trust and transparency surrounds the innovation. We all want to move forward to the many benefits it will provide on clinical and operational fronts. Open dialogue, clear understandings and keeping a moral code is the drum we must beat.
Keeping our oath, fulfilling our promises and steering our integrity will cement the future of AI clearly and positively now and for generations of patients to come.