SMARtCARE: Privacy-Preserving Agentic AI for Bounded-Autonomy

Computer Science > Artificial Intelligence

arXiv:2609.31763 (cs.AI) — Submitted on 24 Sep 2026

Title

SMARtCARE: Privacy-Preserving Agentic AI Systems for Bounded-Autonomy Clinical Decision Support

Authors

Srini Ramaswamy, Deveeshree Nayak

Publication Venue

Accepted to, and to appear in, the 2026 IEEE HealthCom Conference

Subjects

• Primary: Artificial Intelligence (cs.AI)
• Emerging Technologies (cs.ET)
• Software Engineering (cs.SE)

Cite as: arXiv:2609.31763 cs.AI] · [v1 · DOI: 10.48550/arXiv.2609.31763

Abstract

Long-context clinical AI systems can miss relevant patient history when prior admissions fall outside the active reasoning context. In ICU monitoring, this can cause early vital-sign drift to appear nonspecific—even when it resembles a prior deterioration pattern. SMARtCARE addresses this gap through a four-state clinical decision-support architecture: Stable, Meta-cognitive, Assisted, and Regulated (Revoked).

Rather than automatically retrieving prior records, SMARtCARE uses a lossy six-channel fingerprint of the patient's prior trajectory. When current drift matches that fingerprint and the prior record is absent from context, the system raises a Meta-cognitive escalation for clinician review; full retrieval occurs only through clinician action in the Assisted state. A patient-identity guard is designed to enforce correct attribution across data loading, logging, and audit layers.

Evaluation combines:

1. A synthetic Monte Carlo study that validates the state-transition logic and estimator stability—not clinical performance.
2. Real-data runs on both the MIMIC-III and MIMIC-IV Clinical Database Demos.



On MIMIC-III, one prior-pattern recurrence was identified among 14 two-admission patients. On MIMIC-IV, the same pipeline produced no fingerprint matches among 9 two-admission patients—illustrating a key limitation of a fixed canonical pattern library.




Across both runs, all logged decisions were fully traceable and correctly attributed. The results support SMARtCARE as a traceable, privacy-aware mechanism for surfacing middle-context risk; they are not a clinical efficacy claim.




---




Key Takeaways



• Problem: Long-context clinical AI can lose sight of prior admissions, causing early ICU deterioration signals to be misread as nonspecific.
• Approach: A bounded-autonomy, four-state agentic architecture (Stable → Meta-cognitive → Assisted → Regulated/Revoked) that surfaces risk without automatically retrieving patient records.
• Privacy mechanism: A lossy six-channel fingerprint of the prior trajectory that triggers clinician-reviewed escalation only when current drift matches and the prior record is out of context.
• Evaluation: Monte Carlo validation of state logic plus real-data runs on MIMIC-III and MIMIC-IV demos; traceability and attribution held across all logged decisions.
• Limitation: A fixed canonical pattern library produced sparse matches (1 of 14 on MIMIC-III; 0 of 9 on MIMIC-IV).
• Claim boundary: The work demonstrates a traceable, privacy-aware risk-surfacing mechanism—not clinical efficacy.

via ArXiv AI