ASN's Mission

To create a world without kidney diseases, the ASN Alliance for Kidney Health elevates care by educating and informing, driving breakthroughs and innovation, and advocating for policies that create transformative changes in kidney medicine throughout the world.

learn more

Contact ASN

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

email@asn-online.org

202-640-4660

The Latest on X

Kidney Week

Abstract: FR-PO1027

Decoupling of Hemodynamic Instability and Patient-Reported Symptoms: A Deep Learning Analysis of 160,000 Hemodialysis Sessions

Session Information

Category: Dialysis

  • 801 Dialysis: Hemodialysis and Frequent Dialysis

Authors

  • Chiu, Yi-Wen, Kaohsiung Medical University Chung-Ho Memorial Hospital, Kaohsiung City, Taiwan
  • Lin, Ming-Yen, Kaohsiung Medical University Chung-Ho Memorial Hospital, Kaohsiung City, Taiwan
  • Kang, Yihuang, National Sun Yat-sen University, Kaohsiung, Kaohsiung City, Taiwan
Background

Intradialytic hypotension (IDH) is traditionally defined as a composite of blood pressure (BP) drops and patient-reported symptoms. However, frequent clinical discordance suggests these may be divergent phenotypes. We hypothesized that intradialytic symptoms and hemodynamic instability represent phenotypically distinct events driven by different pathophysiological pathways that require independent predictive logic.

Methods

We performed a retrospective analysis of 160,640 hemodialysis sessions from 418 maintenance patients at a tertiary center in Taiwan. Deep learning models (LSTM and 1D-CNN) were developed to independently predict hemodynamic IDH (defined as SBP <90 mmHg or ΔSBP >30 mmHg) and symptomatic distress (patient-reported complications including cramps, nausea, dizziness, etc.) using dynamic vitals and static patient traits.

Results

The models predicted hemodynamic IDH with high accuracy (ROC-AUC 0.92, PR-AUC 0.81). In contrast, predictive performance for symptomatic distress was significantly lower (ROC-AUC 0.77, PR-AUC 0.18). Feature importance analysis revealed that BP drops were driven by intradialytic fluid dynamics, whereas symptoms were primarily influenced by baseline patient fragility. Clinically, 84.5% of sessions with profound BP drops (SBP <90 mmHg) were asymptomatic. Conversely, 73.4% of patient-reported distress occurred without meeting IDH criteria. Even among sessions with major relative drops (ΔSBP >30 mmHg), 90.3% remained asymptomatic.

Conclusion

Our findings demonstrate a profound decoupling between objective hemodynamic instability and subjective patient perception. Current guideline-based composite definitions overlook over 80% of "silent" hemodynamic crashes and fail to account for the majority of symptomatic distress events. Future clinical frameworks should decouple these tracks to move toward more personalized, precision volume management.

Funding

  • Other U.S. Government Support