Enterprise imaging’s role in AI

enterprise medical imaging

IDG recently released the results of a CIO survey and the findings are telling. The report, titled The CIO Tech Poll: Tech Priorities 2018 explores the upcoming technology spending plans and budgets for organizations across a variety of industries.

There’s a clear trend toward increased spending on artificial intelligence (AI), machine learning and predictive analytics. Overall, 43 percent of respondents plan to spend more on AI, 44 percent plan to spend more on machine learning and 47 percent plan to spend more on predictive analytics.

These tools are particularly sought after in healthcare because of their potential to improve population health, reduce hospital readmissions, cut operational costs and enhance the accuracy of clinical predictions. However, these technologies aren’t effective without a robust and accurate data set.

AI can’t learn without enterprise medical imaging

To achieve optimal results, healthcare providers must first build a comprehensive patient portrait that includes discrete, semi-structured and unstructured medical information. Therefore, enterprise medical imaging and enterprise content management technologies that connect disparate semi-structured and unstructured content to enterprise systems like EHRs are essential to properly feed AI, machine learning and predictive analytics initiatives.

An example of the key role enterprise imaging solutions play in machine learning is illustrated by a recent research endeavor at Yale New Haven Health. The provider has developed APIs that allow head and spine CT imaging data to be pulled directly from its vendor neutral archive (VNA) and run through AI algorithms to determine ways to improve workflow efficiency in the ED, according to Matt Zawalich, Director of Imaging Systems at Yale New Haven Health. (To watch Zawalich’s video, click here, scroll to the bottom of the page and click on the tab “Leveraging enterprise imaging for machine learning in the real world.”)

This process allows for large volumes of imaging data to be analyzed rapidly. With the VNA interface, CT scans are randomly sampled, aggregated and anonymized in an automated fashion. If staff members had to complete this step manually, it would take them five to six hours to pull and aggregate each study for the AI tool.             

The standards-based vendor-neutral nature of enterprise imaging solutions also helps support broader population health initiatives by providing a platform that facilitates image sharing among different locations within a health system as well as with other healthcare providers within a region. This capability not only streamlines continuity of care for a patient, but also helps to ensure regional population health initiatives are infused with the most comprehensive set of patient imaging data possible, regardless of origin. Breaking down imaging silos enables more robust data sets, and more robust data sets provide more accurate results.

Interested in talking more about enterprise imaging and AI? Come see me in booth 5743 at HIMSS18!

Ken Congdon has expertise in the healthcare technology industry and has been a contributor to the Hyland blog.
Ken Congdon

Ken Congdon

Ken Congdon has expertise in the healthcare technology industry and has been a contributor to the Hyland blog.

... read more about: Ken Congdon