Efficient Conversational AI Agent to Improve Rural and Urban Healthcare
Nischita N. J1, Mylara Reddy C2
1Nischita N. J, Reva University, India.
2Mylara Reddy C, Reva University, India.
Manuscript received on 05 June 2019 | Revised Manuscript received on 14 June 2019 | Manuscript Published on 29 June 2019 | PP: 160-164 | Volume-8 Issue-5S, May 2019 | Retrieval Number: E10340585S19/19©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: Conversational AI agents are software programs which works exactly like humans, they interpret the users and accordingly react to the inputs given by them. These agents are built considering the medical interventions required to improve the overall health of the society. The AI agent designed acts intelligently during the process of the interaction between the humans and itself. It allows the user to use the interface by asking interactive questions then it processes them and responds relatively. Conversational agents are not only web based but they can also be used on other platforms like mobile phone or any other mobile devices. Despite all these a user shall be satisfied if and only if the software is easy to use and obtains the exact results with all of the queries being answered. The main concern with this model is to give that ease to the user to interact with the agent thus solving the queries related to the symptoms suffered by the patients and hence predicting the disease at an early stage by maintaining the accuracy. There are around 100000 diseases in the world according to WHO. Most of their symptoms overlap as well hence by using this agent its possible by it to think insightfully and predict the early symptoms of the disease. In this paper we have designed a user interface and this interacts with the user to take the necessary inputs. This data is fed to the advanced Natural Language Understanding (NLU) to provide the personalized prediction based on the user interaction. The predictions done by the model uses the classification algorithms of Machine Learning. The accuracy of each of these algorithms varies. Therefore instead of considering only one algorithm and hoping it gives the best accuracy, we can use the Ensemble learning method to improve the overall prediction rate. This method gives better predictive indications as it combines many models results thereby improving the overall precision. Here we train our model using various algorithms and ensemble them to get the final results based on the technique of voting. This paper presents a front-end interface for common man using HTML and Angular JS, NLU for text pre-processing using Tensorflow method and ML model as a classifier, for the prediction which uses various machine learning algorithms like SVM, Decision Tree, Random forest etc and combines them all in a majority voting ensemble for balanced results. Therefore this model interacts with any patients be it from the rural or the urban and based on their symptoms predicts and ranks the most probable disease accurately and reliably.
Keywords: Conversational Agent, Artifical Intelligence, SVM, Decision Tree, Random Forest, Ensemble Learning, Tensor Flow word Embedding.
Scope of the Article: Healthcare Informatics