Object-Centric Process Mining of Laparoscopic Cholecystectomy Workflows: Identifying Bottlenecks and Optimization Opportunities

Background: This study applies object-centric process mining (OCPM) techniques to analyze laparoscopic cholecystectomy (LC) procedures. Traditional process mining techniques are limited in analyzing workflows involving multiple interacting entities (patients, surgeries, anesthesia physical status, etc.). Objectives: To construct an Object-Centric Event Log (OCEL) from LC procedures, discover multi-entity process patterns, identify workflow bottlenecks, and analyze how patient complexity affects perioperative dynamics. Methods: An OCEL representing 1,186 LC cases (1,108 performed, 78 cancelled) from University of California, Irvine Medical Center (UCIMC) was analyzed, using PM4Py to obtain object-centric directly-follows graphs (OC-DFG) and variant explorer. Statistical comparisons examined ICU admission and ASA classification effects on perioperative duration using Mann-Whitney U and Kruskal-Wallis tests. Results: Process discovery revealed a 92% adherence to the reference clinical pathway. Intensive care unit (ICU) admitted patients (n=353, 31.6%) demonstrated significantly longer perioperative durations than non-ICU patients (n=765, median 13.83 vs. 8.85h, p<0.001, Cohen's d=0.81). American Society of Anesthesiologists (ASA) rating showed no significant effect on total perioperative time (p=0.824). Major bottlenecks included booking-to-operating room (OR) transfer (median 7.2 hours), pre-anesthesia preparation delays (28 minutes in 5.6% of cases), and post anesthesia care unit (PACU) discharge (median 19.5 hours, IQR: 6.3-42.1 hours). Pathway completeness was 99.1% with minimal documentation errors. Handoff efficiency varied substantially, with OR-to-PACU transfers occurring rapidly (median: 6 minutes) but PACU-to-discharge transitions exhibiting extreme variability. Conclusion: OCPM enables multi-perspective insights invisible to traditional case-centric approaches. While intraoperative phases function efficiently, preoperative scheduling and postoperative discharge represent primary bottlenecks. The high ICU admission rate (31.6%) likely reflects institutional case mix and data classification practices rather than true critical care needs. Targeted interventions addressing preoperative scheduling optimization, discharge bottlenecks, and real-time monitoring could substantially improve surgical throughput.