Utility of a ML analytics on real time risk stratification and re-intervention risk prediction on AV access outcomes and cost

Background: Vascular access is the lifeline for patients on hemodialysis. The average survival rates of dialysis dependent patients have been improving over the last 5 years and hence their dialysis access needs longevity for uninterrupted optimal dialysis. With the lack of genomic vascular access failure predictors, there is an unmet need for predicting an event and the appropriate approach to mitigate recurrence of the event that could have cost and outcome implications. Methods: We performed a single center experience that extracted relevant clinical (access flow, laboratory data and CKD details), access intervention (prior interventions, type & location of lesion, type of balloon used, use of stents etc.) and demographic (age, vintage on dialysis, sex, social determinants, other medical conditions) data in real time and feeds it into validated ML algorithms to predict risk of reintervention. (Plexus EMR LLC). Results: About 200 prevalent hemodialysis patients with a AV graft or AV fistula were included for this analysis. Need for re-intervention and use of stent/ flow reduction/new access creation were the outcomes analyzed. Plexus EMR is a licensed Azure based platform. R software was used to develop the ML algorithms. Regression factors were developed to assess and test the validity of individual attributes across all the data attributes. Each patient had a real time risk calculator available to the interventionalist on risk of reintervention/ year. Of the 200 patients, 148 had a AV fistula and the remaining 52 had a AV graft. Mean interventions in the year prior to analysis was 1.8 in patients with AV fistulas and 3.4 in AV grafts which decreased to 1.1 in AV fistulas and to 2.4 in AV grafts (p < 0.01) post tool deployment. There were 62 AV graft thrombectomies done in the observation year and 62% of those were repeat thrombectomies. Stent utilization increased to 37 (22 in AV grafts and 15 in AV fistulas) and 2 patients had AV access flow reduction surgery. The cumulative cost (predicted) preintervention was $712,609 and decreased to $512,172 post intervention. Stent utilization increased by 68% in the evaluation year and 89% of the stents used were PTFE coated stents. Conclusion: Utilizing AI with ML based algorithms that includes clinical, demographic and patency maintenance variables could become new standards of care to optimally manage AV accesses and lower cost of care.