Authors:
Rui Miguel Dos Santos Patornilho
1
and
André Vasconcelos
2
Affiliations:
1
Instituto Superior Técnico, Portugal
;
2
Instituto Superior Técnico, Instituto de Engenharia de Sistemas e Computadores and Investigação e Desenvolvimento, Portugal
Keyword(s):
Diagnosis, Recommendation, Health, Interactivity, Reliability.
Related
Ontology
Subjects/Areas/Topics:
Applications of Expert Systems
;
Artificial Intelligence and Decision Support Systems
;
Biomedical Engineering
;
Data Engineering
;
Enterprise Information Systems
;
Health Information Systems
;
Information Systems Analysis and Specification
;
Knowledge Management
;
Ontologies and the Semantic Web
;
Society, e-Business and e-Government
;
Software Engineering
;
Strategic Decision Support Systems
;
Web Information Systems and Technologies
Abstract:
Today’s health has a determinant role and it is a subject of concern by society. Diagnosing a
disease or obtaining a medical specialty, given a set of symptoms, is not a trivial task and different
decisions and approaches can be adopted to solve and handle this problem. Expert systems advise
patients about a possible diagnosis, associated diseases, treatments and more concrete information
about a disease considering simple symptoms. However, most systems don’t have the recommendation
component of a medical doctor, which will be the differentiating factor of this research. The aim of
this paper is to develop an algorithm capable of determining the medical specialties associated with a
set of symptoms and diseases, and based on the medical specialties obtained, recommend the most
suitable specialists. The algorithm is divided into two phases: Health Screening and Health Professional
Recommendation. Health Screening has the purpose of determining and computing all the medical
specialties probabilities, given a set of patient symptoms and applying a statistical model based on all
the relations symptom!disease and disease!medical specialty. Health Professional Recommendation
has the purpose of recommending the best health professionals, given a set of patient preferences,
applying a weighted mean average, where each weight of a health professional feature is given by a
patient according to his preferences. This algorithm was evaluated through a set of test cases, having a
database with information about symptoms, diseases and medical specialties. This algorithm was later
compared to other systems that have the same purpose, to access its quality. The comparison result
between the algorithm and WebMD system indicates that the diseases found by the solution are in 80%
of all the cases equal to the diseases found and pointed by WebMD system.
(More)