UPMC

Steven L Blumer

Clinical Informatics Pathfinder

As the associate medical director of radiology informatics, what are your key roles and responsibilities at UPMC?

We perform imaging within the organization. We are introducing AI algorithms into some of our in-house development of imaging applications to optimize imaging information technology systems. My role is to provide physician expert input regarding IT-related issues to improve the process while conducting substantial workflow testing to develop new PACS systems for radiology.

What challenges do you face in the pediatric healthcare space, and how does UPMC help address these challenges?

To overcome the prevailing shortage of pediatric radiologists in the country, we are dealing with large volumes of radiology studies at UPMC to develop AI algorithms specifically for pediatric radiology patients. Incorporating this advancement will help alleviate the staff shortage by automating the process to provide an assistive imaging diagnosis for pediatric patients.

Can you talk about the technology that you are using to make your recent project successful?

Currently, we are implementing a bone age assessment algorithm study in pediatric radiology to enhance pediatric radiology practices. For instance, we have developed AI algorithms to improve radiologists' workflow, ensuring that the correct study is routed to the correct radiologist. These algorithms also help with radiology scheduling and can generate reports from other parts of the radiology department using natural language processing. However, incorporating this AI algorithm into an organization is technically challenging since most of the algorithms are yet to get 510(k) clearance from the FDA, and healthcare professionals do not have the necessary experience in implementing the products.

What will be your advice to your fellow peers in the industry?

This is a health equity issue, and we have to figure out a way to improve the development of algorithms specifically for pediatric patients. As it is permitted to implement artificial intelligence to some extent within pediatric radiology without FDA approval, some institutions are using AI algorithms to provide quality care for pediatric radiology patients. They have done research projects to ensure that they work for the betterment of the pediatric population. The federal government has to get involved in creating incentives for using AI algorithms in pediatric radiology, as it has done for pharmaceuticals. Pediatric healthcare institutions need to share data among industry partners to address the shortage of data sets required to develop the algorithms for pediatric radiology.