SupplyChat

In the presence of COVID-19, it is crucial for hospitals to obtain essential supplies**. SupplyChat allows them to check inventories of their hospital and optimizes the search time for required supplies.

Inspiration

Our team wanted to provide a dynamic and user-friendly tool for health specialists aiding COVID-19 affected patients. We learned the benefits of being able to optimize the search for essential items for medical professionals, and through our research, we saw that one of the most important issues is the unavailability of inventory information in a user-friendly manner. Additionally, our team specialized in machine learning, and we decided that by utilizing dynamic machine learning algorithms, we'll be able to maximize our impact through our software tool. Through SupplyChat, we believe that we can provide a truly streamlined experience in the search for essential medical supplies.

How We Built It

We have developed a chatbot based on nltk and keras to enable true NLP. Our data has been retrieved from lists of essential items available at a hospital, and in this prototype, we have included the quantity of each item. Our training model for the chatbot is based on a 3-layer model: the first layer has 128 neurons, the second layer has 64 neurons, and the third output layer has the number of neurons equal to the number of intents to predict output intent with softmax. The number of times the model is trained is set at 20,000 to ensure maximum proficiency in the data. Furthermore, the training model is based on a stochastic gradient descent with Nesterov accelerated gradient to generate good results. Our chatbot GUI is based on the tkinter package and each visual representation of the chatbot has been backed by a python function based on keras.

Challenges

There were challenges that we faced, particularly with the stochastic gradient and developing the website. In terms of the stochastic gradient and the training model, we had a few bugs that we causing the program to either crash or not train the model correctly. Additionally, our team lacked experience on front-end development, but in the end, we were able to develop a beautiful webpage for SupplyChat that provides information about our product and the product itself in a very user-friendly manner.

Conclusion

We believe that with future developments and updated versions of SupplyChat, we'll be able to positively impact the health sector. Health specialists will be able to acquire essential supplies and medicines while hospital workers can find out where to obtain essential protective gear such as face masks, allowing medical professionals to aid their patients more often than worrying about obtaining medical supplies.

Built With

Share this project:

Updates