Skip links

Aidoc Clinical Usage – In conversation with Dr. Melissa Davis, Chief of Emergency Radiology at Yale Medicine

Where Always-on AI meets Radiology in Practice

I recently had the chance to sit down with Dr. Melissa Davis who is currently the Chief of Emergency Radiology at Yale Medicine. With so much going on in the field of radiology and with RSNA just around the corner, it was exciting to have this opportunity to have an open discussion about the current state of the field, how AI will be integrated currently and in the future, as well as Dr. Davis’s experience using Aidoc’s always-on AI within her department at Yale Medical.

Ariella Shoham: Can you share with me a little about yourself and what you currently focus on?

Dr. Melissa Davis: Sure. I am the chief of Emergency Radiology at Yale Radiology and Biomedical Imaging. I spend part of my time in a clinical setting, and part of my time administratively where I am the MD partner for an innovative group called Clinical Redesign.

I also work for an optimization team at Yale Medicine called Yale Clinical Optimization Services in which we go into departments individually and assess how they operate and make recommendations on how they can improve, as well as assisting in the implementation of these recommendations over a period of time.

Finally, I work for the Center for Outcomes Research and Evaluation where I help with two sets of measures that are focused on imaging and surgery, in which we work with CMS to maintain and evaluate those.

Ariella Shoham: Wow, fascinating! You’re definitely wearing many hats! How did you get into the field of radiology?

Dr. Melissa Davis: I like pictures. It’s very basic. I liked the people I met when doing my radiology rotations. I liked the idea of being able to diagnose and solve that problem of what’s going on.

Ariella Shoham: Can you tell me why you decided to look for an AI solution for your radiology department and how you got started with Aidoc?

Dr. Melissa Davis: Well, it was sort of a coincidence of events. At the time, we were looking for a new PACS, which also spurred a conversation about selecting something that could easily integrate with AI in the future and if so what would that look like. We were then connected with Aidoc through one of our fellows and we immediately saw that this solution was unique because it focused on the emergent setting, as opposed to other AI platforms we had encountered which focus on cancer or lung nodules, all of which are important but this seemed different because it can have an immediate impact on patient care which was important for me as an emergency radiologist.

The big question in radiology today is how are radiologists going to lead the next phase and how will AI affect radiology. Our department wants to be a leader in that conversation. We knew that this was a great way to engage in this technology.

Ariella Shoham: I’m sure you’re familiar with the fear shared by some in the community that AI will replace radiologists. Do you think this is still an actual fear or that most physicians are aware that this is the next step forward as something to help radiologists?

Dr. Melissa Davis: There may be a fear of that out there but the more we talk about it the more we can dissipate those fears. It’s not going to be that AI replaces radiologists; it’s going to be that radiologists that use AI replace the radiologists that don’t.

You (the radiologist) must understand how to leverage the technology around you. For example, Aidoc shows us where a potential hemorrhage in the head is right now, but it’s really us (the radiologist) that need to tell the clinician how big that hemorrhage is, how much mass affected is associated with it, and what the potential clinical pitfalls are to help guide clinical management. AI can help guide us in workflow management but the need for a human component should not be negated.

Ariella Shoham: What did you expect to gain from AI and what kind of expectations did you have?

Dr. Melissa Davis: The expectation was that we’d get something to improve our current workflow. We expected AI to help us identify a finding in head and cervical spine CTs faster and more efficiently, so we can relay critical information and findings to the clinicians in the emergency room more quickly. That was the baseline in what we were hoping to gain – specifically around workflow management – making sure that we see a critical case first. Our lists can be 30-40 deep, we want to see a critical case number 1 and not number 40.

Ariella Shoham: So, now that you’ve actually worked with Aidoc, what can you tell me?

Dr. Melissa Davis: Everyone who’s working with it, they like it. The usage is very high – it’s not just our emergency radiologists who are using it but also our neuroradiologists, residents, and fellows, and so that’s great.

When I first signed on I wasn’t sure how many people would interact with the system, and at the same time we have seen some data shared that’s shown improvements in our TAT for exams analyzed by Aidoc, more dramatically in the places where we implemented the software than where we didn’t.

We will need to do some configuration of the data and exclude some outliers but in general the fact that our turnaround times have improved dramatically at the implementation site when compared to other sites in our health system, which says that Aidoc has had an impact.

Ariella Shoham: You did mention that initially this would only be for emergency, because of the acute element, but you also mentioned it’s being used by other physicians as well. Why do you think other physicians are using it?

Dr. Melissa Davis: It’s a cool technology and it does pop up for them. It may not be as important in the inpatient setting, but in the outpatient setting it garnered a bit of attention. There was a case in which it helped us detect an intracranial hemorrhage the same night the patient came in for an outpatient, as opposed to reading it the next day or the next business day, which is typically what happens.

Once you start to have anecdotal examples, more people want to interact. In this case it’s been most beneficial for our outpatient side in which we have patients who can still come in and walk and talk and interact with the world, yet may still have serious findings.

Ariella Shoham: How do you see the future? Where do you think this is going to go in terms of AI and radiology? Or where do you think you’d want to see a tool like ours do?

Dr. Melissa Davis: Workflow management is a huge thing – How do we make sure we’re reading the right case at the right time.

Otherwise – there’s plenty of places.
One thing that could be interesting is to see how it could help with patient scheduling. Backroom, non sexy implementation of machine learning, makes patient lives better if we can figure out how to schedule them more easily.

In relation to imaging, it should go to a place where it removes the mundane tasks of being a radiologists so we can focus on high level interpretation of imaging as opposed to doing things like counting nodules or interval measurement of benign lesions. At the same time it can help make this type of work that much more accurate and decrease the inter-user variation that comes with it. It also impacts patient care. For example, if we measure something slightly differently every single time, we might be sending that patient into a cycle of follows ups that they don’t necessarily need. Tools like AI can bring more standardization across the field.

Ariella Shoham: Do you have any parting words you would like to share with me?

Dr. Melissa Davis: When you choose a company to work with you should chose it based off of the people because it’s a relationship that will last a long time. And I think we chose well in choosing Aidoc because everyone we have met so far has been very engaging and really passionate about the product and its quality.

Dr. Melissa Davis is the current chief of emergency radiology at Yale Medicine, spearheading Aidoc’s integration into their radiology department workflow.

Join the Discussion