Chatbot Disease Prediction

Inspiration

When people experience symptoms, they often search online to determine potential causes. However, this information can be overwhelming and inaccurate. Our chatbot simplifies the process by helping users identify diseases based on their symptoms and providing an initial assessment of whether the condition might be positive or negative.

What it does

The chatbot takes a text input from the user where they describe their symptoms in natural language. It processes this input using natural language understanding (NLU) to detect keywords and patterns that correlate with specific diseases. The chatbot then predicts the most likely disease based on the user's input and assesses whether the diagnosis is likely to be positive (requiring further medical attention) or negative (minor or easily treatable).

How we built it

Natural Language Processing: We integrated OpenAI's GPT-4 API for processing user inputs and generating predictions based on medical knowledge. Machine Learning Models: We utilized a classification model trained on a medical symptom-disease dataset from Kaggle to predict diseases. Web Application: The frontend interface was built using streamlit Medical Dataset: We utilized publicly available dataset from Kaggle. Link: https://www.kaggle.com/datasets/uom190346a/disease-symptoms-and-patient-profile-dataset?resource=download

This is to train an AI model to predict the diseases based on the user input, we then integrated our AI model into the chatbot backend

Challenges we ran into

Parameter tuning of AI to ensure that the predictions are accurate.

Utilizing appropiate statistical technqiues/validation methods to validate the accuracies of user prompt input responses

Deciding on which predictor variables to use to classify the disease outcome.

Accomplishments that we're proud of

We were able to integrate a functional chatbot successfully that allows user to input a text in a natural language form and its able to output the data successfully in the ML model data format.

What we learned

Streamlit

What's next for Chatbot Disease Prediction

Figuring out how to validate the accuracy of the responses generated. Improve ML algorithms/models and prediction performances such as precision and recall.

Built With

• numpy
• open-api
• pandas
• python
• scikit-learn
• streamlit