Optimization Instability in Autonomous Agentic Workflows for Clinical Symptom Detection

TL;DR

This paper investigates optimization instability in autonomous workflows for clinical symptom detection, revealing failure modes where performance degrades paradoxically and demonstrating that retrospective selection can stabilize and improve results.

Contribution

It characterizes a critical failure mode of autonomous AI systems and shows that retrospective selection outperforms active intervention for stabilization in low-prevalence tasks.

Findings

Validation sensitivity oscillates between 1.0 and 0.0 across iterations.

At 3% prevalence, the system achieved 95% accuracy with zero positive detections.

Retrospective selection outperforms active guiding in preventing catastrophic failure.

Abstract

Autonomous agentic workflows that iteratively refine their own behavior hold considerable promise, yet their failure modes remain poorly characterized. We investigate optimization instability, a phenomenon in which continued autonomous improvement paradoxically degrades classifier performance, using Pythia, an open-source framework for automated prompt optimization. Evaluating three clinical symptoms with varying prevalence (shortness of breath at 23%, chest pain at 12%, and Long COVID brain fog at 3%), we observed that validation sensitivity oscillated between 1.0 and 0.0 across iterations, with severity inversely proportional to class prevalence. At 3% prevalence, the system achieved 95% accuracy while detecting zero positive cases, a failure mode obscured by standard evaluation metrics. We evaluated two interventions: a guiding agent that actively redirected optimization, amplifying overfitting rather than correcting it, and a selector agent that retrospectively identified the best-performing iteration successfully prevented catastrophic failure. With selector agent oversight, the system outperformed expert-curated lexicons on brain fog detection by 331% (F1) and chest pain by 7%, despite requiring only a single natural language term as input. These findings characterize a critical failure mode of autonomous AI systems and demonstrate that retrospective selection outperforms active intervention for stabilization in low-prevalence classification tasks.