Artificial intelligence in healthcare – elusive, but close

With the deployment of electronic medical records now widespread, healthcare providers are contemplating what’s next in their ongoing adoption of information technology efforts. That’s why the development of artificial intelligence in healthcare was a primary focus at HIMSS19.

It’s no surprise, as artificial intelligence solutions are well positioned to help usher in a new healthcare ecosystem that is better equipped to deliver predictive healthcare information that drives improved outcomes. One great example is how AI detects skin cancer more accurately than dermatologists.

Analytics in healthcare continue to evolve

Gartner uses the graphic below to outline how analytics continue to evolve. Healthcare organizations should understand that moving into the AI opportunity phase requires additional investment in analytics tools and strategies.

An important step in this process is preparing your technology infrastructure by reducing the number of systems your organization uses. Providers should constantly evaluate how existing systems can be consolidated and extended to meet the needs of all key stakeholders within a healthcare enterprise.

Gartner Analytics Maturity Model

Source: Gartner Analytics Maturity Model

What are the major AI components needed?

Before a healthcare organization begins any AI project, here are the key pieces they need to have in place:

  • Appropriate data

Organizations need to identify both internal and external data; they also need to optimize the quality and accuracy of that data. Oftentimes, data is either in an incorrect format, not properly collected or poorly managed.

These data problems must be addressed first to ensure a clean resource pool for any AI initiative.

  • Extraction tool or method

The ability to extract, clean or stitch together data to make it useable is important. Since large data sets are required to use AI, it must be understandable, managed appropriately and available for easy consumption by an AI algorithm.

Effective technology tools are required to assure that any extracted data is actionable.

  • Analytical model

Once they appropriately identify the data, healthcare organizations need to determine which analytical model to apply. Also, the analytical model must be appropriately supported by a team of knowledgeable staff.

Any AI analytical model requires translation of data to actionable information and solution architects are required to ensure the ongoing use and management of the model. Data scientists are also necessary to verify data is properly modeled and all actions are geared toward the model’s success.

  • Healthcare expertise

The team established to implement an AI solution needs to have solid healthcare expertise. Many data scientists don’t have a healthcare background, so programs should be developed to nurture their healthcare expertise and knowledge.

AI initiatives that link business and clinical components together are often the most valuable, so working knowledge of clinical environments and workflows is essential.

  • Infrastructure components

As with any major technology infrastructure, system components should be consolidated to provide a focused enterprise solution where data can be easily accessed and managed. Multiple archives of clinical data (e.g. documents, images, legacy data, etc.) can inflate costs and create silos that require unique integrations to be applied in order to extract data for AI and machine learning purposes.

Time to play catch-up  

In conclusion, even after major investments in IT, healthcare still has a long way to go to address the needs of a new reimbursement standard and model of care. Providers shouldn’t underestimate the level of change that is occurring.

The impact of AI will be 3,000 times that of the industrial revolution.

In fact, AI is taking root across many industries 10 times faster and 300 times the scale of the industrial revolution, according to a recent report by the McKinsey Global Institute. Although McKinsey says that AI adoption in healthcare is a lower rate than other industries, the overall impact of AI will be 3,000 times that of the industrial revolution.

The healthcare industry will need to adapt its care delivery processes appropriately to successfully navigate this transformation. That’s where content services and enterprise imaging solutions can help. By connecting all your unstructured data, you create a single enterprise repository of patient-centered content to feed AI algorithms that improve outcomes.

Ready to learn more?

Phil Wasson, FACHE, is a healthcare industry manager and consultant at Hyland. His mission is to develop content and create alignment with healthcare organizations focusing on information management and imaging solutions so healthcare organizations can realize more efficient operations that improve patient care. Phil joined Hyland after a three-year stint at Lexmark Healthcare as a consultant, and later as a healthcare industry manager. Phil has more than 25 years leading healthcare IT functions as a CIO and holds a fellowship in Healthcare Management with the American College of Healthcare Administrators. He received his B.S. in Healthcare Management from Southern Illinois University at Carbondale, IL.
Phil Wasson

Phil Wasson

Phil Wasson, FACHE, is a healthcare industry manager and consultant at Hyland. His mission is to develop content and create alignment with healthcare organizations focusing on information management and imaging... read more about: Phil Wasson