Inspiration
Over 44% of Nigerians may be at risk of dangerous health issues due to misdiagnosis. Limited resources and time constraints can make it challenging for doctors to navigate complex diagnoses and also some medical doctors' diagnosis results are limited to the tech available in the hospital, We believe having a system where doctors can make diagnosis decisions faster would improve the overall health of Nigerians.
What it does
A web app designed to assist healthcare practitioners with clinical decision-making and help them to make more informed decisions during the diagnosis of their patients. This tool allows doctors to input symptoms and relevant patient information, which the AI then analyzes to generate a list of potential illnesses or diseases. The goal of this is to help medical doctors make diagnosis decisions faster and also aid in making more accurate decisions.
Key Features Symptom Analysis: Input patient symptoms to receive accurate illness predictions. Data-Driven Insights: It does so by means a model that is trained on medical information peculiar to Nigeria, as well as other verfied medical data from the web for informed decision-making. User-Friendly Interface: Easy to navigate interface for efficient use by doctors and other relevant users. Secure and Compliant: Ensures patient data privacy and compliance with healthcare regulations.
How we built it
We used HTML and CSS to build the User Interface and Flask for the backend, We utilised Google Vertex AI (Google Cloud Platform (GCP) for building and deploy ML models faster, with pre-trained APIs within a unified AI platform) to build the models using Nigerian illness/Diseases dataset we got from open-source data website(Kaggle, NCDC), we hosted the app for live use on google cloud.
Challenges we ran into
- Building a neural net machine learning model using Google's machine learning platform (Vertex AI) that's able to give predictions based off input symptoms data.
- Learning curve of the Vertex AI platform was pretty steep.
- Technical debt
Accomplishments that we're proud of
- Being able to learn and build an artificial neural net using a sequential model of dense layers from Google VertexAI's Keras API - all within the space of 24hours!
- Coming up with a brilliant solution that aids medical professionals, in making crucial clinical decisions that ultimately leads to more live being saved, and ultimately we're proud to have helped make the world a better place.
What we learned
- Soft Skill- Teamwork
- Hard Skill - Build a model using Google Vertex AI
What's next for DiagAI
- Partner with more healthcare facilities, hospitals, govermental health agencies to get more data so we can make more accurate predictions of the diseases/illness that affect the nigerian people
- Add more functionalities like a dashboard to manage patient diagnosis made on the app, doctors can see a list of previous result generated on the app, share them and perform othe actions.
Log in or sign up for Devpost to join the conversation.