July 20, 2023Read More
From June 14-16, I had the pleasure of attending the 2023 Society for Imaging Informatics in Medicine (SIIM) Annual Meeting in Austin. The conference drew a wide range of medical imaging professionals including physicians, informatics leaders, researchers and scientists. This year’s program focused on three pillars: educating, innovating and belonging. Attendees were afforded opportunities to enhance their knowledge, foster innovation and engage with the expansive SIIM community, creating a rich and collaborative experience.
I participated in a panel moderated by Nina Kottler, Associate Chief Medical Officer for Clinical AI at Radiology Partners. I was joined by fellow industry leaders from Canon Medical, Merge by Merative, Blackford and Nuance Communications. We shared ideas on transformational imaging technologies during an #AskIndustry roundtable titled ROI & Value Props: AI is Transformational But Can You Build a Business?
This panel, along with educational sessions and conversations with other leaders, produced many valuable insights. Below are three main themes I observed during the conference:
SIIM 2023 featured many insightful sessions on enterprise imaging and its benefits to healthcare providers. One discussion, An Evolved State of Enterprise Imaging, centered on establishing a common IT infrastructure that serves enterprise stakeholders and follows industry standards.
Although imaging is a large part of diagnosing patients, the infrastructure for accessing, sharing and exchanging records is still not meeting the needs of providers and patients. Health systems use PACS but they face data sharing barriers and rely on manual stop-gap solutions like CDs, which are often unreliable and lead to delays in care.
Effective enterprise imaging strategies fill the gaps in existing systems by creating efficiencies across the enterprise ecosystem. A session on New Methods to Seamlessly Integrate Disparate Radiology Systems and Workforces presented solutions and strategies to connect siloed radiology workforces and IT systems.
One approach is thinking holistically about IT infrastructure to focus on solutions that break down network barriers and connect disparate systems without added friction. This can be achieved by removing reliance on networks and placing patients at the center of their care. With a no-network solution and a patient-driven workflow, providers can seamlessly integrate radiology practices and IT systems to create organizational efficiencies.
As ChatGPT continues to dominate AI discourse, industry leaders are considering ways to integrate AI into their workflows to improve patient outcomes. Throughout the conference, many sessions focused on AI use within radiology and the larger healthcare ecosystem. Leaders discussed how AI can safely and effectively supplement existing workflows.
A session on Multimodal AI for Empowering Healthcare: Advancing Diagnostic Radiology and Beyond demonstrated the potential of generative AI to transform radiology and improve patient experiences. Another session, Leveraging Machine Vision, AI, Knowledge Graphs, and Modern Web Tools to Unify, Simplify, and Amplify Radiology, examined current applications of machine vision and AI within radiology.
Currently, one-third of hospitals and imaging centers use AI for patient care and business operations. Among those organizations, 27% use AI to improve their workflows. Radiologists can quickly search MRI images and detect neurological changes over time. Interventional physicians can also use AI to analyze CT scans and identify the source of a stroke and its location. For many providers, AI creates workflow efficiencies while reducing misdiagnosis and analytical oversights.
While these tools integrate with radiologist and physician workflows, they aren’t necessarily solving for patient workflows and improving their experiences within the imaging process. Patient engagement is key to retaining and attracting patients and AI presents a unique opportunity to increase engagement.
Interoperability issues also affect a provider’s ability to access imaging required to develop AI tools. Another potential barrier to adoption is AI bias. According to Trishan Panch, Heather Mattie and Rifat Atun’s 2019 article Artificial intelligence and algorithmic bias: implications for health systems, AI or algorithmic bias are “the instances when the application of an algorithm compounds existing inequities” toward marginalized groups.
When developing AI applications, providers must use training data that is reflective of the population they serve to ensure they’re combating, rather than perpetuating, existing health inequities. Overcoming these potential barriers will be critical for both providers and patients to realize the improved efficiency and patient care benefits of AI.
Data interoperability was a central discussion point among participants. From sharing out-of-network to working with multiple systems to relying on manual stop-gap solutions, many providers face significant roadblocks when sharing and exchanging data. A roundtable session on Data Interoperability and the Path to Federated Learning discussed interoperability challenges and how they create friction and prevent organizations from harnessing the power of data.
Imaging data is often stored on different PACS and is not easily accessible. Providers turn to CDs or VPNs to fill in the gaps but these manual solutions produce inefficiencies, ultimately preventing data from getting where it needs to go quickly. As a result, patient data is stored on multiple systems, creating complex workflows, staff burnout and delays in care.
Providers encounter difficulties finding and reconciling this fragmented information, leading them to potentially overlook critical data that can help them make more precise diagnoses. Limited data access and availability restrict access to patients’ data, making it difficult to realize the big impact of data-driven insights. Data fragmentation also makes it challenging to identify patterns within patient data and produce new learnings.
Providers must view patient data holistically and incorporate it into their systems mapping to ensure that they and their key stakeholders can easily access and draw actionable insights. Providers can also adopt patient-driven, network-agnostic solutions that streamline workflows and connect seamlessly with their existing systems. These solutions create a solid infrastructure for effectively analyzing and interpreting imaging data.
Continuous dialogue on siloed workflows, enterprise imaging, AI and data-driven insights have the potential to transform radiology and medical informatics. Reflecting on the takeaways from SIIM23, the innovation in the radiology and imaging space is exciting and much needed in order to address the major challenges that providers face.
For the latest radiology news, insights and resources, subscribe to our Radiology Roundup newsletter.