Reinforcement Learning with Balanced Clinical Reward for Sepsis Treatment.
Zhilin LuJingming LiuRuihong LuoChunping LiPublished in: AIME (1) (2024)
Keyphrases
- reinforcement learning
- medical treatment
- real patient data
- acute myocardial infarction
- clinical diagnosis
- therapy planning
- clinical studies
- ischemic stroke
- disease progression
- mental health
- clinical information
- clinically relevant
- temporal abstractions
- function approximation
- state space
- reinforcement learning algorithms
- eligibility traces
- treatment planning
- cancer patients
- medical doctors
- cancer treatment
- markov decision processes
- clinical data
- electronic medical record
- model free
- initial stage
- clinical decision making
- reward function
- evidence based medicine
- medical staff
- temporal difference
- dynamic programming
- physiological processes
- radio frequency ablation
- optimal control
- optimal policy
- drug resistance
- breast cancer patients
- medical care
- medical data
- average reward
- action selection
- clinical trials
- clinical practice
- learning algorithm
- long run
- multi agent
- clinical applications
- treatment plan
- action space
- partially observable
- lung cancer patients
- total reward
- function approximators
- radiation therapy
- policy gradient