Garbage in, garbage out: Avoiding the common pitfalls of AI in healthcare

The promise of artificial intelligence (AI) in healthcare is limitless.

The emerging technology is being positioned to revolutionize diagnostic and treatment strategies for some of the most common and complex disease states, with the potential to continuously learn and improve upon its medical expertise overtime. With this, it’s no surprise that many are suggested that the technology may soon even replace providers.

But are we jumping the gun too quickly on leaving our clinical (and not-to-mention human) specialists for the world of “meaning learning?” As vast amounts of data continue to accumulate, the industry is suddenly eager to put all of its eggs in one basket with the expectation that AI will be able to flawlessly predict the future.

What has been lost in the AI and machine learning gold rush, however, is the immutable law of computing: Garbage in, Garbage Out. The algorithms that AI relies on use accurate and unbiased data to better inform care decisions and to predict the future clinical result. Therefore, when you train AI on inaccurate or biased data, you obviously get inaccurate or biased outputs.

To avoid this common misstep and to take full advantage of the promise of AI in healthcare, here are a few suggestions to consider:

1. Don’t be blinded by the shiny new object

It’s easy for organizations to become overexcited when a new AI capability is first introduced. However, hoping that the capability will magically solve complex business challenges with unproven technology and without a plan is a fool’s errand. Organizations must remember that no amount of marketing hype will be able to replace a solid strategy and the hard work of data science. Holding this belief also ties your AI strategy to a particular vendor, and can leave you locked into a single solution.

The right way to view AI systems is that it is simply another set of information technology infrastructure that should be modular to update components easily. Likewise, the right data strategy needs to be in place to support the aggregation and normalization of healthcare data out of various systems to support the building, testing, and deployment of machine-learning algorithms across the organization. This approach will enable organizations to take advantage of a number of industry innovations, while reducing the risk of a machine learning-system becoming obsolete and resulting in costly custom integrations.

2. Treat AI technologies as the pupil - not the master

At its core, AI and machine learning technologies replicate human cognition and learning capabilities, only faster. Therefore, if organizations don’t have a sound learning strategy in place, using AI or machine learning won’t be helpful. It sounds simple, but believe it or not, this is a common key component that is often overlooked.

Mayo Clinic Chief Information Officer Cris Ross described it best by stating that AI is, in fact, “still pretty dumb.” He didn’t mean this in a derogatory way. Today’s best artificial intelligence capabilities are still entirely driven by the relationship of words to one another and resulting understanding of language. So, the only way for these learning technologies to work is by giving them mountains of data to plow through with the hope that they will find statistically significant meaning. The industry is still in its infancy of AI; think of it as a 2-year-old child just learning to speak, walk and interact with the world. And while AI may not be able to cure cancer or solve world hunger, it can make its human counterparts more efficient at processing and analyzing the right data. It just needs the training wheels to be put on correctly before we can expect it to peddle on its own.

3. There is no substitute for good data

Most importantly, organizations must remember that there is no substitution for good data. Algorithms do not provide a shortcut to magical outputs when data is unsound. As John Bruno at Forrester recently wrote on the implication of Salesforce’s new AI, Einstein:

“The future analytics-driven sales processes is bright, but the path ahead is not without its challenges. Current and potential Salesforce customers should be mindful that intelligent recommendations require a large volume of quality data. If poor data goes in, poor recommendations will come out. Cleansing data and iterating the fine-tuning of recommendations will be vital to long-term success.”

Looking ahead, the healthcare industry must establish an effective data strategy in order to reap the full benefits of AI and machine learning. That means moving beyond the EHR and data warehouse, and making sure that the right underpinnings are in place to manage ALL of the data. Then, and only then, will the full promise of AI in healthcare be seen.

By Todd Winey, Senior Advisor of Strategic Markets, InterSystems

The views, opinions and positions expressed within these guest posts are those of the author alone and do not represent those of Becker's Hospital Review/Becker's Healthcare. The accuracy, completeness and validity of any statements made within this article are not guaranteed. We accept no liability for any errors, omissions or representations. The copyright of this content belongs to the author and any liability with regards to infringement of intellectual property rights remains with them.

© Copyright ASC COMMUNICATIONS 2017. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.

 

Top 40 Articles from the Past 6 Months