Page 250 - Data Science Algorithms in a Week

234 Khaled Alshareef, Ahmad Rahal and Mohammed Basingab

1. Categorization: Cases were classified in one of three solutions categories- 1)
optimization, 2) crowding, and 3) new designs/methodologies problems.
2. Paths taken by patients within the Emergency department (from admission to
checkout) take into consideration different plan layouts and processes and the
originating station used by the patients upon entering the ED. The existing paths
describe the patients’ movements while inside the ED. The literature identified
four different paths depending on the patient point of entry into the system.
As shown in Figure 3, the first path (Path 1) is the most commonly used path.
Patients arrive to the ED through the entrance station. Then, move to the triage
station where a triage nurse will perform the necessary process. After which,
patients with levels 1 and 2 (of a 5-level triage scale) skip the registration and
move to either the treatment station or the hospital depending on their conditions,
while other patients will need to register prior to proceeding to the treatment
station and receive the needed treatment. The lab station where services including
x-rays, CAT-SCAN, or any other tests are made available to the patients. Finally,
patients leave the ED through the exit station. The other three paths include
different permutation of the same services and stations.
3. The third attribute includes the available resources performing treatments in the
ED including physicians, specialty doctors, and nurse practitioners that treat low
acuity patients in some EDs.
4. The fourth attribute includes the number of nurses and their classification such as
triage nurses, emergency nurses, and regular nurses. These two attributes are
initialized at one “1”, since all EDs will have at least one doctor and one nurse.
5. The fifth attribute includes the number of lab technicians in the EDs, and the
number of workers in the lab station.
6. The last attribute includes the number of staff in the EDs including all workers in
non-medical and administrative jobs in all stations. Upon indexing the cases, the
case-base will be populated as shown in Table 2.

The Retrieval Engine

The literature shows several techniques and algorithms used to create retrieval
engines for the CBR methodology. Examples of these techniques include nearest
neighbor, induction, fuzzy logic, database technology, and several others. The most
commonly used techniques are nearest neighbor and induction with decision trees
(Watson, 1999).