# Final Project Progress Report ### May 13 -------- #### Updates from Meeting with Arthur We concluded that looking at the patient flow through the ICU will be more interesting and insightful then just the patient release rounds process. My proposal has been modified to reflect the change. This model will focus on optimizing patient flow that avoids patient overflow (system break) and is robust to adverse unexpected events. System breakdowns will be measured when the amount of patients needing ICU care exceeds the total number of beds. --------- ## Agent Behavior Time-Step (hour) 1. Doctor sees patient - reduces acuity points based on experience 2. Nurse sees patient - reduces acuity points based on experience 3. If my acuity points are 0, I am ready for release 4. If my acuity points > 0 and TTL = 0 [ I die] ## System behavior Please look at the updated proposal for details. Since I have not modeled the system I do not know the outcomes. The obstacle I foresee here is finding a suitable reference pattern. Would reproducing the number of overflows be sufficient? ## Rationale for Agent Rules Based on my conversation with Arthur and my observations this past friday, patient overflow seems to be a sign of a system breakdown as it causes havoc outside of ICU walls. Also, I am assuming each patient has acuity points that each physician and nurse reduces if they are available. I would need data to make this abstraction. I could look to see if doctor patient and nurse patient ratios affect length of stay or outcomes. ## Model Output No model yet, but I think the conceptual model is becoming much more interesting and more sound. I will have to work to calibrate patient acuity distributions and so forth. ## Questions 1. Should I decompose my question into smaller parts? If so, what parts would be useful? From my understanding, patient velocity seems to be determined the the systemic processes. In other words, I think it may make more sense to to approach it holistically rather than from a reductionist prospective. 1. How should I got about visualizing the model for effective communication? Does it make sense to lay out rooms, etc? ## Next Steps I first want to make sure my direction is reasonable and the approach rationale. Then I will attempt to make a basic ABM of the system. Meanwhile, I will be collecting more data from the MICU which should inform the development of the model.