certify
Home4 Barriers to Adopting Artificial Intelligence in Healthcare and How to Overcome Them

4 Barriers to Adopting Artificial Intelligence in Healthcare and How to Overcome Them

The adoption of artificial intelligence in healthcare has been a hot topic and rightly so. AI innovation has already demonstrated significant promise in healthcare by reducing costs to providers and improving quality and access to patients.

Accenture predicts the healthcare AI market to be worth $6.6 billion by 2021 and experience a 40% CAGR. By the same token, the survey results of 200 healthcare decision makers conducted by Intel and Convergys Analytics indicated that 54% expect to see the widespread adoption of AI within the next 5 years. Of the same group of executive respondents, 83% agreed that AI will provide a competitive advantage.

Simply put, the time is now for artificial intelligence in healthcare.

While there are many powerful use cases of AI in healthcare, challenges still remain. At a high level, the key to successful AI adoption requires people, processes, and technology to work in harmony. Ensuring patients’ trust, upskilling talent, having a clearly defined digital strategy, along with ways to measure ROI are among top concerns that hinder successful AI adoption. Consider this your roadmap to overcoming the barriers of AI adoption in your organization.

Problem: Patients don’t trust artificial intelligence in healthcare.

Artificial Intelligence has disrupted multiple industries from marketing to financial services, to supply chain management. In fact, AI innovation is so embedded in our daily lives sometimes we don’t even notice it. While it may be perfectly harmless to have an AI algorithm make recommendations for what to watch next on Netflix, trusting technology to provide accurate health recommendations has far greater implications.

Ninety-one percent of healthcare decision makers surveyed by Intel and Convergys Analytics recognized the benefits of AI but 54% of them fear AI will be responsible for a fatal error. There have been numerous cases where AI has been less than perfect. The instances of Google and Microsoft’s AI going wrong were embarrassing while the accidents involving the self-driving Uber or Tesla were fatal. Taken together, it would make sense why patients would want an opinion from a human expert over that of a machine–even when they’re wrong. As the age old adage in medical ethics states, “First do no harm,” healthcare leaders are erring on the side of caution when considering AI adoption.        

Specific to healthcare innovation, consider IBM’s Watson for Oncology, an AI-powered supercomputer that promises to revolutionize the treatment of 12 cancers accounting for 80% of the world’s cases. According to a STAT investigation of the technology, Watson for Oncology has not lived up to its claims. Three years after IBM started selling this technology, STAT found that the supercomputer was still struggling differentiate between specific forms of cancer and received complaints from doctors outside of the US that treatment recommendations were biased toward American patients.

While IBM marketed Watson for Oncology as a cancer care product, there was a total lack of scientific publications showing how the technology changed doctors’ and patients’ experiences. To add to patients’ mistrust and confusion, when the machine offered a treatment recommendation, it couldn’t explain why the recommended course of treatment was credible because the machine learning algorithms were too complex for the average user to understand.

Solution: Be transparent. Inform patients about the benefits of AI in healthcare innovation and how it works.

In order to ensure patients’ trust in AI innovation, be transparent about the benefits of the technology and how it works. Unless the patient is an expert, AI and machine learning are inherently complex; help patients understand the benefits of AI and be clear about how the technology can support their care. When lives and treatment outcomes are at stake, it is imperative to ensure transparency for providers and patients.

Per the guidelines of the Clinical Decision Support (CDS) Coalition, developers and vendors of clinical support decision tools, especially those powered by machine learning must be transparent about what the product can and can’t do, its data sources, and potential drawbacks if providers were to use it. By educating healthcare providers on how a tool generates  recommendations, it helps protect patients and earns trust.

Problem: Employees worry about Artificial Intelligence jeopardizing job security.

According to a 2018 survey by MindEdge of 1,000 managers across multiple industries, 42% of them believe that AI automation and robotics will eliminate jobs. In addition, 40% of these leaders said their employees lacked the skills needed for AI adoption.

AI automation and robotics will cause a shift in skills needed in the workforce. The results of a 2018 McKinsey study indicate that there will be an increased demand for technological, higher cognitive skills, and social, and emotional skills as widespread AI adoption continues. Specific to artificial intelligence in healthcare, 21% of workers were concerned about their job security due to AI automation and robotics within the next 12 months.

Solution: Foster a digital culture and augment your staff with innovation.

Consider your employees the foundation on which your AI transformation will be built and foster a digital culture. Engage your staff by adopting the trademarks of digitally mature organizations –the willingness to take risks, experiment, and invest heavily in upskilling talent. According to a study by the Boston Consulting Group, 9 out of 10 successful digital transformations focused on culture.

According to Infosys, the most successful enterprises recognize employees are key to driving successful AI adoption. Also, the most forward thinking executives believe that AI innovation will create more opportunities for employees rather than eliminate them.

Consider the idea of augmented intelligence, instead of deploying AI to replace your staff, consider using it to “amplify their capabilities.” It’s not a question of innovation or humans, its about combining the two to form a better long term solution. To bolster this notion, Amazon CEO, Jeff Bezos said in an interview with Geekwire, “I think health care is going to be one of those industries that is elevated and made better by machine learning and artificial intelligence.”  

As it stands now, half of US health executives are investing in AI to help their organizations achieve significant cost savings. While AI has the potential to aid in the fight against cancer, it can address immediate needs that don’t require making a clinical diagnosis–namely, administrative overload.

Problem: AI is overhyped and won’t live up to expectations.

While the survey results of 200 health executives by Intel and Convergys Analytics indicated that vast majority of them expected widespread AI adoption within the next five years, over half of them are skeptical about AI and think it will be implemented poorly or won’t work properly.

Solution: Have a clearly defined digital strategy.

We’ve all heard the saying, “failing to plan is planning to fail.” When it comes to successful AI adoption, it is crucial to collaborate across functions and verticals to create a robust vision upfront.

Whit Andrews, research vice president and analyst at Gartner cautions against following a trend because everyone else is doing it. Instead, start with defining what AI really looks like in your industry, how you can use it to differentiate yourself among your competitors, and how you’ll measure ROI.

Make AI adoption everybody’s business by seeking the input of all departments across functions and verticals. Defining a digital strategy upfront ensures your AI transformation will help you rise above the competition.

Problem: What if I don’t see ROI on my AI transformation?

According to the PwC 2017 Global Digital IQ Survey, only 27% of health executives agreed that they effectively utilized all captured data to drive business value.  

Solution: Lay the groundwork for data interoperability and quality while driving value by increasing revenue and reducing costs.

Artificial intelligence can deliver value by automating redundant human tasks, identifying trends in historical data, and improving decision making.

ROI metrics can be refined to include increasing revenue and reducing costs. However, before the AI can deliver meaningful solutions, the algorithm must learn from a vast pool of data that is standardized, labeled, and free of anomalies. Integrate AI insights back into your workflows by enabling the seamless exchange of data from one source to another. According to PwC research, 59% of health executives agree that big data will be improved with AI.      

Filter the value of AI in healthcare innovation through the lens of increasing revenue and reducing costs. Ensure your insights are meaningful by training your AI with standardized, un-biased, data and enable the seamless exchange of those insights back into your workflows.

Conclusion:

The AI disruption of the healthcare industry is imminent, as executives predict its widespread adoption over the next five years. Successful integration means coordinating people, processes, and technology.

Let’s recap what we’ve learned:

  • Gain patients’ trust by ensuring transparency in all aspects of the technology. Be clear about the benefits and limitations to patients. Consider using AI to automate simple tasks before deploying it for more complex ones.  
  • Foster a digital culture by being willing to take risks, experiment, and upskill talent not only to engage your staff but also ensure faster AI adoption. The most successful AI enterprises see AI as a way to augmenting your staff’s capabilities.
  • Define your AI-strategy upfront by involving all stakeholders across all functions and verticals in your organization. Have a clear vision of what success looks like and define value through metrics of increasing revenue and lowering costs.
  • Ensure the quality of your AI insights by making sure you’re feeding the algorithms with unbiased, standardized data. Enable the seamless exchange of data and flow of AI insights into your workflows by investing in interoperability.

The time is now for the AI transformation of healthcare. While fears of AI-innovation hold patients, staff, and even executives back, its adoption becomes increasingly widespread. Consider the steps outlined in this article to overcome the barriers to AI to let it set you apart from your competition.

READ
Addressing Shortages in the American Healthcare System
banner
Adsense