The case against making AI disruptive in healthcare

In technology, disruption is often considered the crowning achievement. Transform an industry, and you’ll succeed. Be like Uber, and you’ll probably win.

By nature, AI is disruptive. It learns like a human, can improve on its own, and will require new business models, new approaches to regulation, and entirely new ways of thinking. According to Accenture, growth in the AI healthcare market is expected to reach $6.6 billion by 2021— a compound annual growth rate of 40 percent. The 2018 GE Global Innovation Barometer – a survey of more than 2,000 business executives across 20 countries – validated the excitement. For both ‘hype’ and ‘impact,’ executives from most countries rank AI among the top innovations.  

But calling AI disruptive to healthcare is a bit of a misnomer.  Yes, AI will transform the industry, and I am a firm believer in the improvements it will bring to cost, quality and access. But its success won’t be found in solutions that are rushed to the market, lacking necessary quality and regulatory measures. And AI won’t be adopted if it disrupts the flow of those who it intends to help – clinicians, technicians and administrators.  

Unless AI is embedded into the devices and workflows already used today, it simply will not work.

The funny thing about disruption is that it’s mischaracterized as something that happens fast and changes everything. But disruptive technology doesn’t happen overnight. Netflix’s original business model involved mailing rental DVDs. Amazon launched as a website that only sold books.

The best solutions evolve, incorporating customer feedback and strategically progressing into the mainstream with high quality, reliable products.

This is the way GE sees AI entering healthcare, and we’ve committed to taking that iterative, pragmatic approach. The results of taking methodical steps, so far, have been incredibly promising.

With NVIDIA, we’re bringing the most sophisticated AI to GE Healthcare’s 500,000 imaging devices globally, accelerating the speed at which imaging data can be processed. For example, our GPU-powered LOGIQ E10 ultrasound increases the focus area of the image by more than 200 percent and increases the images generated per second by 74 percent, helping detect liver disease earlier, track disease progression and better determine proper treatment.

In addition, our AI-powered Command Centers provide real-time, predictive insight in hospitals by surfacing data from multiple sources that is otherwise buried, so staff have access to the information they need, when they need it, and can react immediately to save lives. These ‘control rooms’ act as a nerve center for smarter decision making – using advanced analytics, predictive tools and cross-function knowledge to make faster, better clinical decisions.   

At Johns Hopkins Hospital, since the Command Center opened in February 2016, there has been about a 60 percent increase in the ability to accept complex patients, an over 25 percent reduction in emergency room boarding, and a 60 percent reduction in operating room holds. 

These are examples of AI addressing critical problems, integrating seamlessly into existing tools, and delivering value in real, tangible ways. This is the AI we need in healthcare – elevating rather than disrupting the industry.