Practical AI advice for top health care leaders

There's a lot of hype about the future of AI in health care. But what does AI really mean? And how can CIOs and IT leaders use it to start solving real business problems?

Tushar Mehrotra, SVP of analytics at Optum^®, helps bust the AI myths and offers practical advice for how to approach AI most effectively.

AI will become more relevant, but ignore the hype

AI is not a passing trend. The growing amount of data and computing power are making these technologies increasingly relevant in the health care space. But when you get rid of the buzzwords and hype, AI is simply a set of new technologies that can help you meet business goals.

Often this means automating processes so you can free up capacity and let humans intervene when it's most needed. AI can be impactful in many situations, such as determining if a skin lesion is cancerous or if a claim is likely to be paid on appeal.

Technically speaking, AI is the ability of a machine to perform cognitive-like functions that you normally associate with human minds, such as perceiving, problem-solving or learning. So how can a health care CIO use these technologies?

Ask the right questions first

When thinking about using AI for your business, it's essential that you don't start with the technology and then force that into a solution. The key is to partner with the business leaders to understand their most important needs, and then develop a technology strategy to address those needs.

Once you have those conversations and understand the business needs, then you can then start to ask and answer a few more questions:

• What are the AI technologies that can enable your business solution?
• What is the right talent and skillset that you need? (And should you try to hire, train or partner?)
• How do you access and curate the data you need?

The key is to start with the business use cases and work backward from them to determine where AI would be most valuable, and how CIOs and other IT leaders can lead the way toward a business solution.

AI can do some things very well (but it can't fix all your problems)

There are a few myths about what AI is and what it can and can't do. Here are some of the top misconceptions.

1. AI won't replace everyone's jobs. It's no different from other advances: it can help humans become more effective and make processes become more efficient. It might change some current roles and create entirely new job categories.
2. AI algorithms won't make accurate predictions with messy data. The quality of data is more important than the actual algorithm. The most important input is the right data, relevant to the specific business problem.
3. AI can't remove human bias in decision-making. When human observations and data-collection processes are not consistent from one observer to the next, algorithms are going to have problems analyzing the data, learning, and making predictions. This can result in, for example, misinterpreted medical prognoses or distorted financial models.

Data science is the core of AI

Large data sets are required to continually detect patterns, and data from disparate sources needs to be standardized and organized. Your algorithm is only going to be as smart as the data you're putting into it. Keeping the data clean and consistent is critically important to being able to solve your business problems.

When you look behind the scenes, using AI to solve business problems is really data science. For example, how do you improve your ability to identify trends and patterns? How do you use increasing amounts of data to make decisions much more rapidly?

It's important to work with a data set that you're comfortable with, that's clean, that has reliable sources, and that you've structured in a way so the output will be relevant to the problem you're trying to solve. That's why the right data-science talent and skillsets need to be part of your solution.

There's a lot of competition for data scientists

It's quite challenging to acquire the right people because there's high demand for this skillset. Your organization needs to be very thoughtful about how, when and where you're going to invest in these resources. You can't just turn a switch and hire 100 data scientists overnight. Because of the demand for people with this skillset, they have a lot of options, inside and outside of health care.

It is critically important to think about what you're going to use them for, when you're going to use them, and how you're going to use them. Hiring the talent before understanding the business use case is a recipe for failure. The best approach is to start with the business use case, connect it to your overall analytics and AI strategy, and then find your talent.

Where a CIO can start

There are a few things CIOs and IT leaders can do to move forward with AI.

1. Educate yourself and be prepared to have a dialogue with other business leaders. Your peers have heard about AI and seen some of the hype. But they'll look to you as the technology thought leader to tell them what's real and what's not.
2. Create a very clear vision of what you want to achieve as an organization. You'll need to lay out a roadmap for how you want to get there, and you'll need close partnerships with your business leaders and the CEO to make AI a commitment for your organization.
3. Think about how you're going to achieve your vision in terms of the talent you have or need. Be thoughtful about what kind of talent and skillsets you'll need, and whether you hire, train, or partner, or all of the above.

Learn more about artificial intelligence, data science, and how Optum is improving health care with these technologies at optum.com/cio.

This is the first article in the Optum "AI talent/skillset" series. Read the second article for insight into hiring vs. training. Read the third article for tips on implementing AI in the business.

© Optum, Inc. All rights reserved.