Our project consisted in three diferentiated parts.
The first one is a statistical analysis on the provided dataset where we checked the importance of the variables to predict the risks. We analized what variables where the most relevant to classify each group, in other words, what each group had more or less in common.
The second one is the development a web application to enable the medical staff to input the data related to a patient surgery, in order to try to predict the level of risk associated with its diagnostic.
If the medial staff have any doubt of where to proceed, they can ask for the help of our chatbot Ivan™ (Integrated Virtual Assistant Nurse) this chatbot is briefed on the several medical documents that the staff found important (in this case some of the portions of the documents of the dataset), this could be configurable to the user's use case.
The third part is the on the fly retraining. This feature came to the fact that we faced a problem that we assume Sant Pau hospital could also face when doing a similar analysis, the sample size is very small. To remedy this we leveraged the user process, where the users of the application can upload it's own data to benefit the crowdsourced machine learning model.
Built With
- angular.js
- fastapi
- pandas
- python
- scikit-learn
- tailwind
Log in or sign up for Devpost to join the conversation.