| Abstract |
Sepsis is an acute and often life-threatening condition necessitating Intensive Care treatment. Intensive Care clinicians must anticipate patient care needs in a fast-paced, data-overloaded setting.
However, despite intensive treatment and monitoring, prognosis remains poor. Current treatment protocols are not sufficiently personalized. Thus, a need exists for better decision support tools.
The secondary analysis of intensive data is a critical step toward. The widespread availability of electronic healthcare data allows for new investigations into evidence-based decision support,
where we can learn when patients need a given intervention. Modeling sepsis treatment requires frequent and recurrent decision making that cannot be adequately captured by supervised models.
Reinforcement learning offers a potential innovative solution. in this work, we focus op guiding a treatment balance between two treatment options: fluid resuscitation and vasopressor dosing using Deep Q-learning. |