E1757. The Proof’s in the Pudding: Using Artificial Intelligence to Improve Root Cause Analysis (RCA) and Failure Mode and Effect Analysis (FMEA)
Authors
Tanya Moseley;
The University of Texas MD Anderson Cancer Center
Romual Perard;
Maimonides Medical Center
Mary Guirguis;
The University of Texas MD Anderson Cancer Center
Megha Kapoor;
The University of Texas MD Anderson Cancer Center
Miral Patel;
The University of Texas MD Anderson Cancer Center
Flavia Posleman Manetto;
The University of Texas Medical Branch
Beatriz Adrada;
The University of Texas MD Anderson Cancer Center
Background
When the imaging-to-diagnosis process is portrayed as a "one-stop shop," radiology departments often have high patient volumes and fast-paced work environments. Consequently, human error can occur at any point of the process of providing health care. Placing sole responsibility for errors and near-misses on individuals as opposed to procedures may result in errors and near-misses not being reported. It is essential to recognize the significance of detecting faults and learning from them. The processes of root cause analysis (RCA) and failure mode and effect analysis (FMEA) should be taught at all levels of health care organizations to reduce or eliminate the possibility of adverse outcomes. Adding artificial intelligence (AI) to these processes would foster analyzation and contextualization of the information to trigger actions autonomously without inherit human biases.
Educational Goals / Teaching Points
This educational exhibit’s goals are to encourage a high-reliability organization safety culture that learns from errors, teach the principles of why and how to perform RCAs and FMEAs, provide the participants with the context needed to conduct RCAs and FMEAs in tandem with AI, and equip participants with the ability to perform RCAs and FMEAs immediately after viewing this exhibit.
Key Anatomic/Physiologic Issues and Imaging Findings/Techniques
RCA is a technique used to determine the origin of negative events that have already transpired. Within 45 days of a sentinel event happening in a clinical department, an RCA is now required by the Joint Commission. FMEA is a technique for proactive risk assessment that identifies potential weak spots. FMEA creates error prevention and mitigation procedures. Despite its military and industrial roots extending back to World War II, FMEA is utilized less frequently in medicine. The combination of AI, RCAs, and FMEAs could make it easier to analyze and contextualize the data, hence minimizing the impact of human bias.
Conclusion
The purpose of utilizing comprehensive systemic analyses such as RCA and FMEA is to study medical errors, determine the underlying causes or grounds for such failures, learn from them, and attempt to prevent their recurrence. Despite their important differences, FMEA and RCA cannot be separated and true sensemaking in patient safety requires both processes. FMEA attempts to determine the impacts of all conceivable cause sets. RCA tries to determine the causal set of each conceivable effect. FMEA is the temporal reflection of RCA in the present moment. AI in tandem with RCAs and FMEAs may simplify data analysis and contextualization, reducing the influence of human bias.
