Offline reinforcement learning with uncertainty for treatment strategies in sepsis
This work addresses the challenge of overadministering treatments in sepsis care by enabling personalized recommendations, though it is incremental as it builds on existing reinforcement learning techniques with a specific domain application.
The paper tackled the problem of personalizing sepsis treatment by developing an offline reinforcement learning method that provides multiple treatment options with confidence estimates, and mitigated bias against aggressive interventions using subspace learning, resulting in more accurate policies across healthcare applications.
Guideline-based treatment for sepsis and septic shock is difficult because sepsis is a disparate range of life-threatening organ dysfunctions whose pathophysiology is not fully understood. Early intervention in sepsis is crucial for patient outcome, yet those interventions have adverse effects and are frequently overadministered. Greater personalization is necessary, as no single action is suitable for all patients. We present a novel application of reinforcement learning in which we identify optimal recommendations for sepsis treatment from data, estimate their confidence level, and identify treatment options infrequently observed in training data. Rather than a single recommendation, our method can present several treatment options. We examine learned policies and discover that reinforcement learning is biased against aggressive intervention due to the confounding relationship between mortality and level of treatment received. We mitigate this bias using subspace learning, and develop methodology that can yield more accurate learning policies across healthcare applications.