Problems
Healthcare System is complex and hard to manage in many hospitals. Patients always need to wait long time for making appointments with doctors, doing operations and getting the treatment payments from insurance company. Healthcare System is related to different intuitions and people, it is hard to do interviews each task in this system. The goal of this research is to capture, understand, and document how clinical work is actually done, including constraints of context and information resources, and then analyze how the efficiency and quality care could be measurably improved, with an emphasis on health IT as the means. The project will model and analyze clinical care work flows, networks and decision making.
User Experience Research
Research Process
Survey and Interview
UW Medicine Hospital is a great clinic hospital in Washington State. We sent online survey to paitients who have done clinics in the hospital. In addition, we also interviewed doctors and nurses in the hospital to ask their normal daily work and how to get patients' medical insurance record. Finally, we called some large insurance companies to ask them how to deal with every patients treatment bill.
Online Survey for Patients
Interview nurses and doctors
Build Model
Model Building is the core part in this research. In this process, we analyzed the interview and survey result; filtered the important tasks in healthcare workflow; transfer to workflow model and evaluated the model. This process is an agile process and we repeated the process four times, because we need to double check with hospital officer several times.
Discussion, Sketch, Model Workflow and final workflow diagram based on Visio
Quant Research
After, finished the workflow model, we do the model simulation by MATHSim, and get the result data in order to find out the problem (waste time, useless) part to upgrade our model.
Simulation Tool (part) based on workflow
According to the simulation, we found that the patients workflow model is very similar to Poisson and Normal distribution in different tasks. Thus, we decided to use t-test and linear regression model to find out unnecessary this model (Confidence Interval range: 95%).
Data Analysis Visualization
Result Optimization
After data analysis, we simplified the workflow model based on the analysis result. The final optimized model is much easier to manage and record. In addition, we sent a report to UW Healthcare IT to ask them implement the new system and we published several papers related to this research in internal conference meeting.
Related Publication
Model Checking for Verification of Interactive Health IT Systems, Keith Butler, Eric Mercer, Cui Tao. Conference: AMIA 2015 Symposium, At San Francisco
Using Conceptual Work Products of Case Management to Design Health IT, Andrew Berry, Keith Butler. Journal of Biomedical Informatics (Impact Factor: 2.19). 11/2015; 59.