This paper discusses on finding meaningful outliers by hybridizing Rough_Outlier Algorithm with Negative Association Rules (NAR). By using NAR rules, it is expected that the rules generated will give a better perspective of outliers detection for medical diagnosis on Heartdisease dataset. To improve the computation of outliers, a Binary Particle Swarm Optimization algorithm is used. Only strong and interesting rules are generated from Positive and NAR rules, hence suitable measurements for support and confidence are used in the research work. Results showed that PAR and NAR rules are able to give better perspective on identifying outliers and in giving meaningful outliers for predictive medical analysis in case of Heartdisease. For further work in future, a test for accuracy will be applied by applying the false positive and true negative measurements. In conclusion this work is significant for giving different perspective of medical diagnosis of Heartdisease, hence assisting experts in their field for better decision making.