Advancing digital pathology and informatics

Sept. 23, 2021

Why did you decide to move from the University of Chicago Medicine — where you were the medical director of pathology informatics, among other roles — to your current positions at the University of Michigan?

I decided to move to the University of Michigan/Michigan Medicine for three main reasons:

  1. While I was medical director of pathology informatics at University of Chicago Medicine, I had very little influence on decisions about informatics and central IT processes that affected the clinical laboratories. At Michigan Medicine, I would oversee a much larger, autonomous pathology IT group (pathology informatics) that has a direct impact on the operations of Michigan’s clinical labs and patient care.
  2. I would work with and learn from two other pathology informatics faculty members daily, which is very rare in our field, given that most institutions only have a single pathology informaticist (if they have someone at all).
  3. There would be expanded educational and research opportunities for me in clinical informatics at Michigan Medicine.

Will you describe the key duties of your role as associate CMIO and how your inclusion on this team benefits the clinical laboratory and pathology department?

I am the associate CMIO (ACMIO) for pathology at Michigan Medicine (one of 15 total ACMIOs). My key duties include serving as a liaison between clinicians and our pathology informatics team, representing the Department of Pathology at enterprise informatics/IT meetings, and acting as the primary contact for information assurance/cybersecurity risk and IT compliance issues for pathology applications. The key benefit of being part of the institutional clinical informatics team is ensuring that pathology and the clinical laboratories are “at the table,” and their voices are heard for major IT decisions and initiatives made on the enterprise level.

Since April 2021, the 21st Century Cures Act has required laboratories to release lab tests and pathology reports to patients promptly. What impact do you expect this requirement to have on the workflow at labs?

For most hospital labs where results are posted to the electronic health record (EHR) and its patient portal, there was little impact with Information Blocking Phase 1 since the primary burden was on EHR teams. That said, with Information Blocking Phase 2 (December 2021-December 2022), all ancillary clinical applications containing electronic protected health information (ePHI) will now be expected to have their data included in responses to information requests. This means clinical laboratories will have to assess not only their primary laboratory information system but also all middleware and other primary lab systems (e.g., for molecular, human leukocyte antigens [HLA], outreach/reference lab, etc.) to see if information from those systems is required to be included in responses to formal information requests. This may add additional work to both the IT and lab teams if they create standardized reports to handle these requests to achieve full compliance with the 21st Century Cures Act Final Rule.

In your opinion, what are the keys to implementing a digital pathology system successfully?

There are many different ways one could implement digital pathology, but in short, the primary keys are to do the following:

  1. Understand the use cases that are driving digital pathology forward at your institution.
  2. Create a realistic budget for digital pathology that includes not only the whole slide imaging systems but also the storage, network, staff, workstations, displays, and additional software required.
  3. Implement digital pathology in a stepwise fashion, so you can iron out any kinks in the process and evolve your laboratory and sign-out practices accordingly.

Will you describe what computational pathology is and how it can improve diagnostics and patient outcomes?

Computational pathology is best described as incorporating multiple sources of pathology and clinical laboratory data to create mathematical models that generate diagnostic inferences and predictions to make the best possible medical decisions. Over the past few years, computational pathology has become closely tied to the fields of machine learning and artificial intelligence (AI), with those terms taking over the healthcare informatics space. In general, AI, through the use of machine learning models, has the ability to improve multiple facets in pathology, from providing clinical decision support during the sign-out process to optimizing laboratory workflows. For example, current AI models show promise in standardizing diagnostic criteria for cancer synoptic reporting, creating computational “assays” to predict patient response to treatment and/or prognosis, and improving clinical laboratory operations through predictive and prescriptive analytics.