A Surgical Process Mining Model to Improve Operational Efficiency in the Peruvian Healthcare System

Abstract

Surgical backlog in low- and middle-income countries (LMICs) is exacerbated by fragmented information systems and weak process governance. Existing process mining solutions often assume data environments and governance maturity absent in LMIC public hospitals. We propose and evaluate a governance-integrated process mining framework architected for resource-constrained, digitally fragmented settings. The framework translates five inputs—clinical data, process definitions, infrastructure, controls, and policies—into actionable outputs through a four-layer architecture. A proof-of-concept in a Peruvian tertiary public hospital (1,234 surgical episodes, 2023–2024) revealed a critical 15-hour bottleneck between admission and preoperative preparation (IQR: 11–19), accounting for 62.5% of planned preoperative stay, and 39 distinct process variants indicating extreme pathway fragmentation. Conformance against the institutional protocol was 75% (fitness=0.75). Expert validation (n=15) rated integration of information between areas (4.81/5) and active participation of medical staff (4.75/5) as critical factors. However, a significant gap emerged between high perceived usefulness (4.50/5) and willingness to use (4.63/5) versus lower implementation feasibility (4.00/5) (p=0.010, Cohen’s d=0.77). Reliability analysis yielded Cronbach’s alpha of 0.89. This study provides a transferable, empirically grounded framework that moves beyond aggregate indicators to diagnose surgical inefficiencies, offering a roadmap aligned with national backlog reduction policies in LMICs.