Disease Prediction

Pipeline
Output Sample 1
Output Sample 2
Output Sample 3
Output Sample 4
Comprehend Medical
SSM
Diagnosis

Inspiration

Our inspiration was based on the motto “Do not google your symptoms”.

It’s a worldwide problem that people google their symptoms searching for a quick cheap diagnosis which may lead to underestimating/overestimating their medical conditions.

These kinds of results could make them avoid the doctor because they don’t want to hear bad news or they’ll worry so much that it leads to other harmful consequences

What it does

It can help the patients to have some sort of understanding of what their medical symptoms could mean, and provide them with an initial diagnosis for their symptoms, that would be more reliable than googling them.

It is not a substitute for professional medical advice, diagnosis, or treatment.

The patient will write a text explaining his/her symptoms and the model will diagnose the patient’s medical condition.

How we built it

The project was built on the Amazon Sagemaker notebook instance on which Spacy library was installed. The classification model was first trained using a Disease Prediction dataset. Then the pipeline was built which includes using Amazon Comprehend Medical through the AWS SDK for Python (Boto).

Challenges we ran into

One of the challenges we faced was the post-processing of the output of Amazon Medical Comprehend and tweaking it to help our model. Also, it was a challenge to find an appropriate similarity method and determine its threshold.

What we learned

We learned a lot about how Machine learning can be used to draw insights from large medical data sets to enhance clinicians' decision-making, improve patient outcomes, automate healthcare professionals' daily workflows, accelerate medical research, and enhance operational efficiency.

Additionally, we learned about the strength of Amazon Comprehend Medical in extracting health data from medical texts and how it can be used in pre-processing to increase the quality and boost the performance of healthcare machine learning models.

What's next for Disease Prediction

Our solution can be extended by integrating it with chatbots that can ask the patients questions to give a more reliable result. Also, hospitals in Egypt have doctors and one of their tasks is to do an initial diagnosis on patients when they arrive to direct them to the right clinic, our solution can be extended to help/replace those doctors in this particular task.