🏆 Winner of the HealthCare Track at the MINeD Hackathon
Instant MD is a Investigation, Medication and Chief Complaint recognition application using NLP
Windows:
$ python -m venv venv
$ venv\Scripts\activate.bat
Unix/MacOS:
$ python -m venv venv
$ source venv/bin/activate
$ pip install -r requirements.txt
$ python app.py
- Prakhar Jain
- Nikhil Rajput
- Parth Desai
- Parth Panchal
- Nora Surani
All libraries versions used are listed in requirements.txt
Our NLP approach uses both text and sentence tokenization.
No third party NLP libraries were used for the project and all pattern matching was done through Regex
This helps us serve results instantly as opposed to using popular libraries like nltk or spacy, giving us the name InstantMD. We preserving sentence structure while also searching the tokenized words. We can then extract symptoms using a list of symptom keywords and the chief complaint by matching with the part of the anatomy that the patient is experiencing symptoms in. This is done through the anatomy keywords. Other factors are similarly extracted using keywords and rules designed by us and implemented through Regex.
Being able to have machines understand unstructured textual content already plays a big part nowadays in our life. NLP can contribute largely to the advancement of medical science. NLP is used to extract information from free text narratives written by a variety of healthcare providers. Here we approach natural language processing algorithm where we
- Evaluate the free text and compare it with the dataset of NLP Dictionary.
- Annotated the important textual content from the text like Medication, Investigation (Pathology, Radiology), Chief Complaint.
- After Segregation the highlighted data should automatically enter the flow of EMR.
To develop a solution, the first step is to understand the problem. Develop an NLP module to identify the keywords related to a patient's investigation, medication and chief complaint from a free text in the text box. Highlight the extracted content and feed them as input in EMR’s Chief complaint, Investigation and Medication module.