Methods and apparatus for the application of reinforcement learning to animal medical diagnostics

A method for the application of reinforcement learning (RL) to the treatment of animals, the method comprising:

• establishing a policy and state value function for an assessment RL agent based on historical assessment outcomes;
• establishing a policy and state value function for a plan RL agent based on historical diagnostic and treatment recommendations for a predetermined set of assessments;
• determining a set of assessments, using the policy and the state value function for the assessment RL agent, based on a set of observations;
• determining a set of plans based on the determined set of assessments using the policy and the state value function for the plan RL agent;
• storing the determined set of assessments and the determined set of plans in a historical log storage device;
• receiving feedback on the determined set of assessments in order to update the policy and the state value function for the assessment RL agent;
• and
• receiving feedback on the determined set of plans in order to update the policy and the state value function for the plan RL agent independent from the update of the policy and the state value function for the assessment RL agent.

A system configured to apply reinforcement learning (RL) to treatment of animals, the system comprising:

• a classification artificial engine that takes as input radiographic images, and outputs classifications for various conditions of an animal;
• a subjective biological data storage device that stores subjective biological data for the animal;
• an objective biological data storage device that stores objective biological data for the animal;
• an assessment RL agent which takes as input the classifications for the various conditions of the animal from the classification artificial engine, the subjective biological data for the animal from the subjective biological data storage device, and the objective biological data for the animal from the objective biological data storage device, and outputs a determined set of assessments for the animal based on the inputs using a policy and a state value function for the assessment RL agent;
• a plan RL agent which takes as input the determined set of assessments for the animal from the assessment RL agent, and outputs a determined set of plans for the treatment of the animal based on the determined set of assessments from the assessment RL agent using a policy and a state value function for the plan RL agent;
• a historical log storage device which stores the determined set of assessments and the determined set of plans;
• wherein the system:
• establishes the policy and the state value function for the assessment RL agent based on historical assessment outcomes;
• establishes the policy and the state value function for the plan RL agent based on historical diagnostic and treatment recommendations for a predetermined set of assessments;
• receives feedback on the determined set of assessments in order to update the policy and the state value function for the assessment RL agent;
• and
• receives feedback on the determined set of plans in order to update the policy and the state value function for the plan RL agent independent from the update of the policy and state value function for the assessment RL agent.