Health Language Blog

Webinar Recap: Applying AI in healthcare: Challenges, opportunities, and emerging applications

Posted on 11/19/18

feb844d8b3aaf2b66ada73fd0de4263b

The noise about the promise of artificial Intelligence (AI) in healthcare is deafening at times. In fact, today’s clinical, financial and IT leaders would be hard pressed to devise a forward-looking strategy that did not at least consider the role of AI in such initiatives as healthcare quality, cost reduction and population health management.

 

The attention AI is receiving was recently underscored in an informal webinar poll of more than 700 healthcare professionals, where 60 percent of attendees said they have either implemented or plan to implement AI in coming months. Hosted by Health Data Management, the webinar “Applying AI in healthcare: challenges, opportunities and emerging applications" featured insights from Wolters Kluwer experts on the implementation challenges that limit AI advancement as well as solutions to common problems and emerging use cases. Senior Medical Informaticist Chris Funk, Ph.D., and Director of Product Management Sarah Bryan, both with Wolters Kluwer Health Language Solutions, joined Senior Data Scientist Krishna Srihasam to unpack AI best practices that are powering healthcare environments.

 

AI is a branch of computer science that emphasizes the development of intelligent missions, ultimately enabling computers to think and work like humans on various levels. While AI is not a new concept, Srihasam pointed out that healthcare has recently identified use of AI as an effective way to augment intelligence to support human decisions through two of its domains: machine learning and natural language processing (NLP).

 

AI uses are broad across the healthcare field, and the potential continues to expand. Bryan explored six common use cases where healthcare organizations are realizing value and return on investment from NLP and machine learning today.

1. Knowledge management: Catalog and create search indexes for articles and education content

2. Interoperability: Enable meaningful exchange of information through semantic interoperability

3. Reporting and analytics: Extract and normalize information for complete, accurate data aggregation

4. Clinical decision support: Extract patient data from unstructured text to design customized decision support

5. Clinician workflow: Automatically codify data, eliminating complex workflows

6. Financial management: Expand computer-assisted coding applications




The predictive power of AI is directly correlated to quality of data. Funk emphasized that clean data continues to be one of healthcare’s greatest challenges, as data lives in multiple places in different formats, creating data silos and barriers to meaningful data exchange. The result is inaccurate, incomplete and inconsistent data. When healthcare organizations fail to address use of non-standard local terminologies, specialized terminologies and free text without adequate mapping to industry standards, the potential of AI is limited due to difficulties with context, disambiguation and negation.

 

To improve this outlook, infrastructures must exist that can aggregate and normalize structured, semi-structured, and unstructured text to a standard that is recognized for the purposes of AI algorithms. The presenters provided insights into the common use cases explored earlier in the presentation, pointing out specific data quality issues that can arise and solutions that can overcome these challenges. They showed how reference data management (RDM) and other terminology management solutions, like those offered by Wolters Kluwer Health Language, are the most important foundational first step when implementing any AI technology.

 

For organizations ready to learn more about how AI can be implemented, presenters reviewed a few key characteristics to be mindful of when researching potential vendor partners. A strong vendor must have the following knowledge and capabilities: 

  • A strong healthcare terminology foundation that can support accurate and complete data aggregation and normalization
  • A platform that can handle the complexities and nuances of healthcare data
  • Availability of data scientists who continually advance AI platform capabilities as use cases expand
  • Ability to leverage the right AI methods to address the specific challenges of a healthcare organization
  • Ability to customize, as there is no one-size-fits-all solution

Healthcare is just getting started with AI, and the opportunities to impact quality outcomes and the bottom line are significant. We invite you to join the discussion and learn more about how to get your data ready for the AI revolution.

In case you missed the webinar, click here to view it on demand. 

Not ready for AI, but want to take the first step? Click here to learn more about Health Language Reference Data Management solutions. 

Topics: Reference Data Management, Natural Language Processing, artificial intelligence, Machine learning, NLP

About the Author

Ali Gilinger has over eight years of experience in the healthcare industry, with primary focus on product management and strategic product marketing. Prior to joining the Wolters Kluwer Health Language team, Ali was a Solutions Manager for the pharmacy automation division of Swisslog Healthcare. Ali is responsible for strategic product marketing of the complete Health Language solution portfolio.