Audio Version: Press Play to Listen
Three top tips to ensure you do the right thing today for your AI needs in the future.
In our recent blog on the scalability of AI in imaging, we shared some thoughts on the need to manage the load on existing PACS infrastructure. But scalability is also about careful planning, so today we thought we’d share some advice on what to think about when making decisions around implementing AI today and in the future. Have a five-year vision It’s important to think beyond the single application you want to integrate today, and ensure that the approach you're taking allows you to expand without having to repeat effort. If you want one or two products today, but know that you’ll probably want 10 more in the future, it doesn't make sense to do 12 separate integrations. Consider doing one platform integration that can also manage all the subsequent integrations – it’s all about doing the work today that will give you the maximum flexibility in the future. Try before you buy – if you can There’s a lot of noise out there – more than a hundred imaging AI vendors appeared at RSNA. Some have FDA-cleared products, some have actual experience in a real hospital environment, some have an understanding of what price their product can maintain. And some don’t. It’s important to find a sustainable way that allows you to try-out and evaluate these applications, without risking everything on something that isn't ready or isn’t suitable. This can take a huge amount of assessment work for each application, or you can adopt a curated marketplace approach, where someone else does that work for you. A curated marketplace ensures that an application meets an appropriate level of performance, reliability and value. This should minimise the clinical and management time spent assessing whether a product is even worth trying, and should help you find applications that have a good chance of being useful in your specific environment. Research or commercial AI? The vast majority of hospital sites are not focused on research. They don’t build their own solutions and are focused on delivering results. However, there are research hospitals that develop and maintain their own AI algorithms. And the scalability strategy for these two models is quite different. A non-academic hospital needs finished AI solutions that add value to current practice. And, while this is also important in an academic hospital, there may also be a focus on building usable data and archives, and how to create an internal infrastructure for the development of AI. An AI strategy needs to be split between what is for research and development and what is for production use. It’s important that hospitals don’t spend their time building an environment for developing their own AI when what they really need is something that can deliver regulatory approved products that are ready to go. Like the points made above? Download our eBook to find out even more: