1. Ada
  2. Editorial
  3. Business & industry
  4. Putting AI to work: How can we overcome barriers to AI adoption in healthcare?

Putting AI to work: How can we overcome barriers to AI adoption in healthcare?

From futuristic visualization techniques to speedier diagnosis, artificial intelligence (AI) is transforming healthcare. But while investment in health AI continues to boom, uptake in healthcare settings could be too slow to meet ever-growing demands.

Incorporating revolutionary technology into a field as closely regulated and complex as healthcare is a real challenge. There are basic logistical concerns as well as more complicated barriers to adoption that need to be addressed.

So what can we, the health AI community, do to help advance the industry and speed up the adoption of AI solutions?

Let’s take a look at some of the challenges we face and how we can overcome them.

Regulation

It goes without saying that we follow medical device regulations to the letter. But as it stands, these regulations tend to be written with physical devices in mind. We need to go further than following the rulebooks. We need to work with regulatory bodies to create new frameworks that are purpose-built for AI. Constructive feedback streams for post-market surveillance will help us develop and refine AI solutions.

Privacy

People share the most intimate details about their lives with health providers. That’s an enormous responsibility we take on. As digital health companies, we can never do too much to protect our users’ data. Where privacy laws aren’t yet tight enough, we need to collaborate with legal institutions and bodies to fortify them.

Representation

Health AI should work for everyone. But the health needs of a population are as diverse as the individuals that make it up. We can’t expect AI solutions to work for everyone if we don’t build them using datasets that represent that diversity. Neutral data doesn’t exist, but we can strive to make sure the data we create and use represent the diversity of the populations we aim to serve.

Sustainability

Our tools need to be safe and effective – users expect no less. But we face a unique challenge in AI. Fundamentally, we want to help as many people as possible. However, we need to strike a balance between the benefit to the individual and the health system as a whole. For instance, AI tools that screen otherwise healthy people for certain conditions could cause an unmanageable influx of people at clinics. We need to face this challenge head-on by continuing to undertake research and publish robust clinical evaluation studies. We need to work with health systems and physicians to ensure they’re equipped and funded to manage any increased demand.

Transparency

Physicians need to understand the logic behind the AI they use. So we need to provide clear explanations wherever we can. That transparency is crucial for building trust in AI. As algorithms get more complicated – neural networks, for instance – decisions can become increasingly difficult to explain. Where this happens, we need to be explicit about it so people can make informed choices about the tools they use. We can minimize how much complexity impacts transparency by maintaining human control throughout development.

Sociocultural

Medical knowledge doubles every few months. That’s a lot to keep on top of. Unfortunately, it means most physicians don’t have time to follow developments in AI unless they have a particular interest. On top of that, emerging technologies usually aren’t covered in detail at medical school. This lack of familiarity means some physicians are hesitant to use new technologies. We need to work to make sure digitization, innovation, and emerging technologies are covered at university and during early training. That way, we can build trust and familiarity from the start of physicians’ careers.

It’s important to remember that resistance to change in healthcare doesn’t come from a place of fear or technophobia. Stakeholders across healthcare are primarily concerned with protecting patients and users. But with the new pressure of a global pandemic, we can’t waste any more time. People need these tools and innovations now more than ever.

It’s our responsibility, as a health AI community, to work collaboratively and make that happen.

Keep reading. Take 8 minutes to learn more about the people behind our products and review enterprise solutions that could help us improve health and care together.