A Fuzzy-Ontology Based Diabetes Monitoring System Using IoT
293
HasAlarm(?p, ?a), HasServiceMessage(?a, “You are in danger, Call a doctor”),
HasActuatorState(?e, “Device is swiched On”)
Rule 2: Patient(?p), HasAge(?p, Old), HasBMI(?p, OverWeight), HasPC(?p,
HighPC), HasPP(?p, HighPP), HasPF(?p, LowPF), HasDCP(?p, MLUnAcceptableCDP), greaterThan(?Old, 65), HasActivity(?P, LightAC), Recommandation
(?R), greaterThan(?OverWeight, 38), lessThan(?OverWeight, 40), greaterThan
(?HighPC, 65), lessThan(?HighPC, 100), greaterThan(?HighPP, 20), lessThan
(?HighPP, 100), lessThan(?LowPF, 25), greaterThan(?LightAC, 8), lessThan
(?LightAC, 30), greaterThan(?MLUnAcceptableCDP, 50), lessThan(?MLUn
AcceptableCDP, 200)-> HasDietStatus(?p, UnHealthy), HasRecommandation
(?p, ?R), HasServiceMessage(?R, “Do more exercises, Eat More Fat, Less Carbohydrate, Less Protein”), NeedFood(?p, LowCalorie).
The proposed system is implemented on java. Fuzzy components are implemented using jfuzzylogic plugin to calculate the output variables, which are
health condition and diet status. The ontology is evaluated based on the
querying-answering approach using DL and SPARQL queries and is then integrated into the java application using Jena API. The evaluation includes: (1)
Technical evaluation to check the consistency and the coherence of the ontology
using Jena reasoner as well as the response time that the system takes to execute user queries and display the result. Results show that the time consumed
depends on the number of inputs that the query needs to calculate the output.
The more the inputs are, the more the response time is. (2) Functional evaluation, which evaluates the efficiency of the system and the accuracy level of
the decisions for which the system was queried. Results demonstrate that the
accuracy of our system can reach 100% for diet status related queries, 96% for
health condition related queries and on average 94% for queries related recommandations. Therefore our system is capable of acting more similar to human
expertise.
6 Conclusion
This paper proposes a fuzzy-ontology based diabetic monitoring system using
IoT technology. The fuzzy-logic is adopted to infer the health condition and the
diet status for the patient and then presents the result to the ontology to generate
convenient recommendations. Evaluation indicates that the performance of the
system is increased considerably and the system gives results with more accuracy
compared to the system using classic ontology.
293
HasAlarm(?p, ?a), HasServiceMessage(?a, “You are in danger, Call a doctor”),
HasActuatorState(?e, “Device is swiched On”)
Rule 2: Patient(?p), HasAge(?p, Old), HasBMI(?p, OverWeight), HasPC(?p,
HighPC), HasPP(?p, HighPP), HasPF(?p, LowPF), HasDCP(?p, MLUnAcceptableCDP), greaterThan(?Old, 65), HasActivity(?P, LightAC), Recommandation
(?R), greaterThan(?OverWeight, 38), lessThan(?OverWeight, 40), greaterThan
(?HighPC, 65), lessThan(?HighPC, 100), greaterThan(?HighPP, 20), lessThan
(?HighPP, 100), lessThan(?LowPF, 25), greaterThan(?LightAC, 8), lessThan
(?LightAC, 30), greaterThan(?MLUnAcceptableCDP, 50), lessThan(?MLUn
AcceptableCDP, 200)-> HasDietStatus(?p, UnHealthy), HasRecommandation
(?p, ?R), HasServiceMessage(?R, “Do more exercises, Eat More Fat, Less Carbohydrate, Less Protein”), NeedFood(?p, LowCalorie).
The proposed system is implemented on java. Fuzzy components are implemented using jfuzzylogic plugin to calculate the output variables, which are
health condition and diet status. The ontology is evaluated based on the
querying-answering approach using DL and SPARQL queries and is then integrated into the java application using Jena API. The evaluation includes: (1)
Technical evaluation to check the consistency and the coherence of the ontology
using Jena reasoner as well as the response time that the system takes to execute user queries and display the result. Results show that the time consumed
depends on the number of inputs that the query needs to calculate the output.
The more the inputs are, the more the response time is. (2) Functional evaluation, which evaluates the efficiency of the system and the accuracy level of
the decisions for which the system was queried. Results demonstrate that the
accuracy of our system can reach 100% for diet status related queries, 96% for
health condition related queries and on average 94% for queries related recommandations. Therefore our system is capable of acting more similar to human
expertise.
6 Conclusion
This paper proposes a fuzzy-ontology based diabetic monitoring system using
IoT technology. The fuzzy-logic is adopted to infer the health condition and the
diet status for the patient and then presents the result to the ontology to generate
convenient recommendations. Evaluation indicates that the performance of the
system is increased considerably and the system gives results with more accuracy
compared to the system using classic ontology.