Inspiration
-What inspired me to create such a chatbot is an article on New York Times that I read in my COR 100 class. The article explained how hopsitlas in NYC are facing extreme traffic, causing people to wait months at times for an appointment. Therefore, with a chatbot that is able to answer frequently asked questions, some of that traffic can be cleared and a better health care system can exist in my state.
What it does
-The project is a way to help hopsitals and pharmacy respond to frequently asked questions and provides health guidance to the user, assisting with what popular medicines are and tips on treating certain health issues.
How we built it
-This chatbot uses data in a json file by storing it in a python dictionary. This data is then looped through and tokenized, where a list contians each individual word from different possible inputs the user can give to the chatbot, with the corresponding tag in another list. Furthermore, the words are sorted and stemmed, ensuring they are not case sensetive, improving the accuracy of the chatbot. The list consisting of 0s and 1s that trains the model, gets transofmred into numpy arrays and that determines what is written in the pickle file. The probability is calculated using tflearn, which contains regression, etc. The chatbot after being fitted is able to provide accurate responses that provides health guidance.
Challenges we ran into
-Some of the challenges I ran into was attempting to incorporate exception handling in order to organize the chatbot and save space and memory, by not training the model everytime the main.py file is ran. This caused the code to function inproperly and the neural network to overflow, requiring problem solving.
Accomplishments that we're proud of
-An accomplishment I am proud of is being able to feel like I am making a difference when working on something that could benefit my community and even the state on a larger scale. This sense of importance, motivated me even farther in order to present this project completed and working optimally.
What we learned
-I learned that Machine Learning can make our lives drastically better, by having the ability to create a model that is simply trained using data, to become stronger and stronger and provide high accurate responses, as if it is a human speaking.
What's next for Pharmacy/Hospital Chatbot
-Actively testing the chatbot within my network of friends, by gathering feedback and improving the model’s accuracy, with plans for deployment to serve as an alternative for traditional appointments and contribute to an effective healthcare system.
Log in or sign up for Devpost to join the conversation.