"A correct diagnosis is three-fourths the remedy." - Mahatma Gandhi

In this fast-paced world where everything seems to be conveniently accessed in a matter of seconds at our fingertips with our smartphones and laptops, some parts of our lives can not be replaced or compromised. Let's not kid ourselves, we are all guilty of getting a scare when we see something suspicious on our skin or if we feel funny, we fall into the black hole of googling the symptoms, believing everything we read and scaring ourselves to an unnecessary extent.

Even forty-four percent of Americans prefer to self-diagnose their illness online rather than see a medical professional, according to a survey conducted by The Tinker Law Firm. That is an alarmingly large amount of people for just a country.

While it is cheaper to go to Google to self-diagnose rather than to visit a doctor, this often leads to inaccurate diagnosis and can be extremely dangerous as they might follow a wrong treatment plan or may not realize the severity of their condition.

Through our personal experiences in Asian countries, it was common to get an X-Ray scan at one place, and then another appointment with a doctor had to be booked the next day to receive an opinion. We also wanted to create a way to avoid inconvenience for some people and make it socially sustainable this way. Especially with the exponentially rising cases of damaging effects on the environment, we wanted to create a means of a sustainable health care system while reducing the negative impacts.

What it does

Doctorize.AI is an easy-to-use web application that uses Machine Learning to scan the images or audio clip uploaded, and with a simple click of a button, it is processed, and the results inform if there are any concerning medical issues recognized or if the x-ray is clear. It also lets you know if you must seek immediate medical attention and connects you to a matching specialist to help you out. Worried about something in general? Use the “Request A Doctor” to connect and talk all your worries out.

An added bonus: Patients and doctors can use Doctorize.AI as an extra tool to get an instantaneous second opinion and avoid any false negative/positive results, further reducing the load of the healthcare system, making this web application socially sustainable. It is also a safe and low-carbon health system, protecting the environment.

Our models are able to recognize and respond to cases by classifying:
- skin cancer (Malignant or Benign)
- brain tumor (Glioma_Tumor, Meningioma_Tumor, Pituitary_Tumor, No_Tumor)
- X-ray (Tuberculosis, Pneumonia, COVID-19 induced Pneumonia, or Normal)

How we built it

The frontend was built using:
- Next.js
- JavaScript

The backend was built using:
- Flask
- Python
- TensorFlow/Keras for the Deep learning models to classify images/audio
- AWS S3 for storage of large data set

Challenges we ran into

As four individuals came together, we were bursting with uncountable ideas, so it took a long discussion or two to settle and choose what we could realistically achieve in a span of 36 hours.

Here are a few challenges we ran into:
- Lack of dataset availability
- Different time-zones
- Mix between first time hacker, new hacker(s), and experienced hacker in the team
- AWS S3 - Simple Storage Service
- Storage of large data
- AWS Sagemaker
- Computational power - deep learning takes time

Accomplishments that we're proud of

- Being able to tackle and develop the Machine Learning Models with the supportive team we had.
- Creating a successful clean and polished look to the design
- Models with over 80% accuracy across the board
- Figuring out how to implement Flask
- Experimenting with AWS (S3, and Sagemaker (not as successful))

What we learned

- Together as a team, we learnt how to use and apply CSS in an efficient way and how different CSS tools helped to achieve certain looks we were aiming for.
- We also learned how to use Flask to connect ML models to our web application.
- Further, we learned how to use AWS (S3, and Sagemaker (not as successful)).

What's next for Doctorize.AI

- Allow patients and doctors to interact smoothly on the platform
- Expand our collection of medical cases that can be scanned and recognized such as more types of bacteria/viruses and rashes
- Bring in new helpful features such as advanced search of specialists and general doctors in the area of your own choice
- Record the patient’s history and information for future references
- QR codes on patient’s profile for smoother connectivity
- Voice Memo AI to summarize what the patient is talking about into targeted key topics

Built With

Share this project: