ASN's Mission

ASN leads the fight to prevent, treat, and cure kidney diseases throughout the world by educating health professionals and scientists, advancing research and innovation, communicating new knowledge, and advocating for the highest quality care for patients.

learn more

Contact ASN

1401 H St, NW, Ste 900, Washington, DC 20005

email@asn-online.org

202-640-4660

The Latest on Twitter

Kidney Week

Abstract: PO0583

Machine Learning (ML) Driven CKD Care Navigation Confers Robust Value Through Adoption of Home Dialysis and Transplantation

Session Information

Category: CKD (Non-Dialysis)

  • 2102 CKD (Non-Dialysis): Clinical, Outcomes, and Trials

Authors

  • Srinivas, Titte, University Hospitals, Cleveland, Ohio, United States
  • Morales, Ray, Intermountain Health Care Inc, Salt Lake City, Utah, United States
  • Lee, Suji, Intermountain Health Care Inc, Salt Lake City, Utah, United States
  • Phillips, Michael, Intermountain Health Care Inc, Salt Lake City, Utah, United States
Background

The ESRD patient journey is an abyss of lost opportunity for home dialysis or transplantation (Tx). Despite known value of home dialysis (Home hemodialysis (HHD), Peritoneal Dialysis (PD)) and Tx, CKD patients default to in center hemodialysis (IC-HD) after crashing into ESRD. This picture exists absent upstream navigation with management of comorbidity and misaligned fee for service economic forces incenting IC-HD. We report early results of ML directed care navigation upstream of CKD and multidisciplinary co management (Primary care (PCPs) and Nephrology) driving greater adoption of Tx and HHD/PD in a large integrated health system in the intermountain West.

Methods

A custom-built ML algorithm identified chronic kidney disease (CKD) patients using synthetic data from multiple Electronic Health Record sources. Features used to identify CKD included but were not limited to eGFR, other laboratory values, ICD-10 codes, comorbidity clusters, CKD risk factors (DM, HTN etc.), scheduling data, DRG data, biopsy and imaging data. ML output triggered workflows of Kidney Care Navigators (KCNS) who, using a customer relationship management utility (CRM) co-managed comorbidity of CKD with PCPs and navigated patients to nephrology consultation aiming improved HHD/PD/Tx adoption rates

Results

Over 6 mo, among 12000 CKD records, the ML algorithm identified 1898 patients not seen by a nephrologist. KCNs interfaced with PCPs co-managing ESRD and CKD patients using CRM;1169 unique patients generated 1873 workflows (Fig, A); 58 PC of ESRD pts. adopted HHD/PD and pre emptive Tx vs. historic averages of 12 percent. Fig, B.; 7 year Net Present Value, $ 28,000,000 vs. Capital outlay of $2,00,000 (Fig., C).

Conclusion

1) ML-Driven CKD Care Navigation conferred robust value through five-fold increased home dialysis/transplant adoption in a large integrated health system.
2) Our approach is generalizable across EHRs and with synthetic data ML, allows multi-institutional collaboration or consortia to deliver value in CKD at scale.

Figure

Funding

  • Clinical Revenue Support