A Convolution Neural Network Based Framework for Similarity Learning in Healthcare
Krishan Kumar Saraswat1, Sandhya Tarar2, Sandeep Gupta3
1Krishan Kumar Saraswat, Research Scholar, Department of Computer Science and Engineering, Shri Venkateshwara University, Gajraulla (Uttar Pradesh), India.
2Sandhya Tarar, Department of CSE, School of ICT, Gautam Buddha University, Greater Noida (Uttar Pradesh), India.
3Sandeep Gupta, Department of CSE, JIMS Engineering Management Technical Campus, Greater Noida (Uttar Pradesh), India.
Manuscript received on 26 March 2019 | Revised Manuscript received on 03 April 2019 | Manuscript Published on 12 April 2019 | PP: 94-99 | Volume-7 Issue-6C April 2019 | Retrieval Number: F90370476C19/2019©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Foreseeing patients’ danger of building up specific ailments is a vital research theme in social insurance. Precisely distinguishing and positioning the similitude among patients in view of their chronicled records is a key advance in customized social insurance. The medical data for two different diseases are unpredictably examined and have fluctuated tolerant visit lengths, can’t be specifically used to quantify quiet similitude because of the absence of a suitable portrayal. Also, there needs a powerful way to deal with measure quiet similitude on electronic health records. In this paper, we propose a novel significant equivalence learning frameworks which in the meantime get the hang of understanding depictions and measure pairwise similarity. We utilize convolutional neural system (CNN) to catch nearby vital data in electronic health data and after that feed the scholarly portrayal into triplet metric similarity learning. In the wake of preparing, we can get pairwise separations and likeness scores. Using the closeness data, we at that point perform patient grouping. Test results demonstrate that CNN can more readily speak to the longitudinal medical data groupings, and our proposed systems can efficiently cluster patients into different disease groups.
Keywords: Convolutional Neural Network; Patient Clustering; Personalized Healthcare; Triplet Loss Metric Learning.
Scope of the Article: Patterns and Frameworks