Combined Machine Learning Approach to Cluster and Analyze Intrinsic Medical Data Set

Recently Machine learning algorithms have received increased attention in Medical domain and they have been successfully applied to many medical applications. Machine Learning is a field of artificial intelligence which is employed to solve complex models and thus helps computers to perform themselves without human assistance. Multitask learning is the prominent field of machine learning which learns and solves multiple related tasks simultaneously and it is successfully applied in wide range of applications like disease prediction, disease progression, etc. Existing learning methods assume that tasks are related to each other and hence adaptclustering technique to improve the performance of all tasks by exploiting the commonalities and inconsistencies of each task. Thus, we focus on developing a combined and enhanced clustered multitask learning method for medical domain. Taking into consideration the nature of biological data, and to overcome the difficulty of analyzing the complex structure of data in medical domain and to help the physicians to forecast details about disease progression in patients, we intend to use a combined spectral multitask clustering approach.This collective frame work will first identify the similar tasks after which the representative tasks are identified and the cluster structures are learned simultaneously by multitask learning method. We collected data in medical domain which consists of similar feature structures and the experimental results obtained from this medical data set are compared with normal fuzzy and K-means algorithms.