*This article is based on an educational exhibit from SIR 2024.
The integration of artificial intelligence (AI) solutions into the health care domain has been met with both enthusiasm and skepticism. Our institution, Temple University Hospital, has shown how AI impacts patient care and team dynamics when integrated into the pulmonary embolization response team (PERT) of an academic medical center. In this commentary, we explore the lessons learned from this novel technology and evaluate its impact on patient outcomes, workflow efficiency and clinical decision-making.
AI, specifically AiDoc, is designed to enhance diagnostic accuracy and speed in identifying pulmonary embolisms (PE). It represents a significant advancement in critical care. PE is where one or more arteries in the lungs become blocked by thrombus. This requires rapid detection and treatment to prevent serious morbidity or death. The traditional diagnostic process, which relies on the radiologist and treating physicians to interpret complex imaging studies, can be time consuming and may be subject to human error. These findings may be further delayed by extensive backlogs of imaging cases. AiDoc's integration demonstrates promise in mitigating these challenges by providing a real-time, AI-driven analysis of radiological scans, which can flag potential PEs for immediate review.
Lesson 1: Speed and accuracy in diagnosis
One of the most tangible benefits observed was the reduction in time from imaging to diagnosis. AiDoc's AI algorithms analyze images faster than human counterparts (without fatigue), while also offering constant and consistent support. This accelerated diagnostic process potentially saves lives by facilitating quicker treatment initiation. However, this benefit also underscores the importance of ensuring that AI recommendations are accurately interpreted and acted upon by the diagnosing and treating clinicians. Notifications are sent directly to the AiDoc-integrated mobile app. All members of the PERT can immediately and simultaneously review these notifications.
Lesson 2: Workflow integration challenges
The integration of AiDoc into PERT's workflow highlighted several operational challenges. In the initial assessment, multiple AI solutions were evaluated with a focus on the most robust platforms for radiology. Initial integration of the software to the hospital RIS/PACS necessitated system-wide approval from numerous stakeholders. Adding the mobile app to the implementation and adoption process posed new challenges, including safeguarding patient data and ensuring provider support.
Lesson 3: Impact on team dynamics
The implementation of AiDoc positively impacted team dynamics, fostering a more collaborative multidisciplinary atmosphere. As team members became more comfortable using AiDoc's capabilities, they discussed best practices and protocol optimizations and utilized key in-app features, resulting in a more cohesive approach to patient care. This shift highlights the potential of AI to serve not just as a tool for enhancing individual tasks but as a catalyst for team development and process improvement.
Effect on patient care
From a patient-care perspective, the early evidence points to positive outcomes. Faster, more accurate diagnoses lead to timely interventions, which can significantly improve patient prognoses. Furthermore, the ability to rule out PE quickly makes patient management and resource allocation more efficient. To ensure that AiDoc benefits are extended across a variety of patient populations and clinical settings, it is essential to continue monitoring these outcomes.
Conclusion
AiDoc's integration into a tertiary academic medical center's PERT offers valuable insights into AI's transformational potential for healthcare. While the technological advancements promise improved patient outcomes and operational efficiency, the lessons learned highlight the importance of addressing human and systemic factors to fully realize these benefits. As AI continues to evolve, its role in augmenting health care delivery will undoubtedly expand—improving patient care, optimizing patient outcomes and potentially decreasing healthcare disparities in low-resource settings.
