Covid-19 Risk Assessment

Inspiration

Coronavirus, or COVID-19 is an extremely contagious disease caused by the SARS-CoV-2 virus. Most people infected with Covid experience mild to moderate respiratory symptoms, while some experience severe illness. Those individuals who experience harsher symptoms often possess certain characteristics that put them at a higher risk of contracting and recovering from covid. It is of importance to use an individual's health characteristics in order to assess their covid risk level. This information will allow high-risk individuals to know to take greater precautions and seek medical care early when presenting covid symptoms. This information can also assist medical practitioners and hospitals in predicting medical supply usage and treatment plans.

What it does

This website helps users asses how at risk they are with COVD-19 contraction using their medical data. We gathered the data from the user and pass it through our model to predict whether that patient would be at high risk of contracting COVID-19, and provides preventative guidelines if so.

How we built it

Using scikit-learn and Google Colab, we created and trained a model based off of previous patient records from Mexico, using primarily logistic regression to provide our most accurate results. Using Next.js as our frontend to develop our website and connected it to our endpoint, developed using Flask. We utilized this to gather the data from the user and send it to the model for processing, which is then returned to the website to display the results of the model.

Challenges

Interpreting the data was one of the biggest challenges, seeing as there were many missing values and insignificant columns which lowered the accuracy of our results. Weighing each of the columns was also neglected, due to the fact that we could not assess whether each factor had a higher impact on contraction of the disease and pass it through our model. We also had to use outdated data points, which could have significantly affected our trained model's end product.

Accomplishments

We utilized machine learning for the first time, demonstrating our understanding for the topic and algorithms that were used to train certain models. We also created a modern/minimalist website which we believe appropriately caters towards the audience and the goal of the product, overall enhancing user experience.

What we learned

Each of us come from different backgrounds of study, so we learned how to utilize and maximize our various skillsets to develop different aspects of the product. COVID is still a prominent disease, and this model could help minimize its impact on individuals.

What's next

We want to train the model to provide more accurate results, so we would need to gather clearer, more recent data. Utilizing all of the data to play a factor and weighing risk of each would be another goal of our application as some factors have a greater impact on the risk of contracting the disease.