Health Gecko

Inspiration

With the given prompt, we started working on finding the problems within the industry.

What it does

Our system automative abnormality detection, risk assessment, and predictive healthcare by analyzing real-time patient data and genetic mutation.

How we built it

Challenges we ran into

So firstly we tried the SVM model to train our data, initially starting off with 500 generated values. However, after training the model with this data we noticed that the accuracy rate was not only low but skewed towards the Middle value and after doing research into this it was found out that there were significantly more entries for the Medium value than Highs or Lows so we thought maybe leveling it out would help solve the problem. So, we generated more data for each category so that each label appears an somewhat equal amount of time so that we can get better accuracy results. At first, however, due to the library sklearn we soon learned that this library classified the labels into Low, Very Low, Medium, High and Very High. So we spent time trying to categorize these labels hoping that it worked well with this model, however, that was wrong we soon learned that due to the ranges the model was still predicting the Medium values. So, we had to change this labeling to just Low, Medium and High to try to get the best predictions this did seem to be the best option, however, it did not magically solve our issues. Now with over a 1000 data set we suggested that we would try a different model so we started trying Random Forest a supposedly better model to help us model outr data. However, we only saw increases in the range of .02 which really isnt that much which but us back onto the drawing board and looking for another solution which then lead us to learning more about the XGBoost model which also did not help. But we did notice that the Random Forest was the best so we came back to this method and tries to optimized it the best increasing the numb er of branches.

Accomplishments that we're proud of

That we successfully integrated the Genetic Mutation

What we learned

We learned how small data can lead us to the big solution.

What's next for Health Gecko

Built With

• ai
• apis
• css
• figma
• flask
• javascript
• joblib
• matplot
• ml
• next.js
• pandas
• python
• scikit-learn
• seaborn
• tailwind