131. Process Mining Electronic Health Records of People with Multiple Sclerosis: A 5-year Longitudinal Study.
Invited abstract in session HF-4: Innovation 3, stream Sessions.
Thursday, 15:30-17:00Room: St Olavs, Kunnskapssenteret KA12
Authors (first author is the speaker)
| 1. | Märt Vesinurm
|
| Department of Industrial Engineering and Management, Aalto University |
Abstract
Chronic care delivery systems often rely on standardized outpatient pathways to ensure efficiency and consistency at scale. However, when these pathways fail to adapt to individual patient needs, system rigidity can result in delayed interventions and suboptimal outcomes. We apply process mining to a 5-year longitudinal dataset of 1279 people with multiple sclerosis (pwMS) to analyze real-world care trajectories and assess deviations from the intended patient pathway. Our method reconstructs patient-level event logs to identify common patterns, transitions, and escalations.
We find four phenotypes for patient trajectories: (1) those treated with only outpatient contacts, (2) those with outpatient contacts preceding escalation into the emergency department (ED) or inpatient care, (3) those with extended periods of no contact followed by escalation into the emergency department or inpatient care, and (4) ‘others’ with high variation within the service use patterns. We find that on the system level, there is little variation in proportion of different service utilization categories of outpatient, inpatient, ED, no contact. Additionally, annually roughly 70-80% of pwMS are treated with outpatient contacts only, with no escalation. With a significant proportion of costs of care stemming from the more expensive inpatient and ED services, our findings highlight the need for ‘flexibility triggers’ to allow prediction and more flexible reaction to possible care escalations.
Keywords
- Care Pathways
- Patient flow
- Process optimisation
Status: accepted
Back to the list of papers