Page 220 - Data Science Algorithms in a Week
P. 220
204 Abdulrahman Albar, Ahmad Elshennawy, Mohammed Basingab et al.
The expert has deep experience in the daily operations of emergency care
centers.
The expert has solid knowledge in staffing, performance management, healthcare
administration, patient flow analysis, and bed management.
To create a robust knowledge base for the proposed fuzzy system, a minimum of ten
experts are required who meet these qualifications. While discussing these experts here
for the purposes of analyzing their data, and elsewhere in this study, an assigned code
“HCE-k” will be issued for each participated expert, where HCE stands for Healthcare
Expert, and k stands for the expert number.
Database
This study adapts the indirect interval estimation elicitation method. Such a method
carries advantages such as allowing responses from multiple subject matter experts, while
not requiring knowledge of membership functions. Additionally, under this approach,
fewer questions may be used, and given questions may be easier to answer than those in
other approaches. To elicit the degrees of membership for a fuzzy class, let [ ,]
represent the interval values of the fuzzy class j that is determined by expert i. The steps
to elicit and analyze expert knowledge are described as follows:
- Determine all interval values for each j obtained from experts.
- Perform an intersection for j subset intervals to obtain expert consensus.
- Find ambiguous areas among determined intervals.
Fuzzy Rule Base
The fuzzy rule base is the other key part to the knowledge base, including the
database. It stores all derived fuzzy rules, which is intended to provide the fuzzy
inference engine with decision support information within each subsystem. To robustly
create fuzzy rules for each fuzzy logic subsystem, experts are given a form to assess the
consequences of each condition statement, developed from the permutation of each fuzzy
class for a given fuzzy subsystem. A total of 10 healthcare experts will participate in the
fuzzy rules assessment process. The total number of fuzzy rules to be evaluated by
subject matter experts for the fuzzy logic subsystems I, II, III, and IV are 12 (3×4),
2
9(3 ), 36(4×3 ), and 80(5×4 ), respectively. Therefore, the proposed three-level
2
2
hierarchical fuzzy expert system includes a total of 137 fuzzy rules, meaning that there
will be a total of 1370 fuzzy rule assessments from the ten experts. The process of
developing the fuzzy rules is detailed in the following steps: