Inspiration
The project is driven by the idea that the past history of a patient is a critical source of actionable information. However, due to ethical reasons, full access to electronic health records (EHR) is laden with road bumps. Still, we believe valuable health information abounds if one simply decides to look for it. Explanation of benefit reports (EOB), together with medical billing, and pharmacy receipts are a goldmine of information with minimal barrier to access. In an era of big data, these cannot be ignored.
What it does
We created a platform to simplify the collection and storage of patient data. A user can electronically scan the EOB, medical billing, pharmacy receipts or any related documentation he or she received. The app then extracts the useful information from these documents and stores it in a database for future reference. Additionally, it currently supports bar-code reading functionality targeted towards over the counter medication purchase. Leveraging on its stored data and known associations within this data, it can provide useful purchase recommendations to the user with the push of a button.
How we built it
For BenefitBase app, we used angular.js as the front end, with a node.js as the back end. All the database and underlying logic were built using python.
Challenges we ran into
The technical challenges include ways of extracting meaningful information from the EOB. The inconsistency of formats and notations between different insurance companies makes the problem even more confounding, forcing to settle upon reasonable middle-ground. Other challenges we met are the defining a logic to decide which drug is not recommendable. Building an app also caused several troubles.
Accomplishments that we're proud of
We filled the gap of personal medical history that is valuable for the patient to make decisions when purchasing drugs. Furthermore, we also thought of several ways to utilize the data we collected only by cross-referencing the information with the interaction/side effect/precaution of the certain drug. For instance,
- If the records show that one pays a visit to the hospital during the flu season (simple information such as office visits can help), we might assume the patient has a higher likelihood of being sick with the flu. In this case, some drugs such as certain antihistamines, are no longer suitable.
- If one has been on a certain medication recently, the detection of interaction between medications is going to prevent some serious side effects.
- If one receives X-ray exam of spine/kidney/lung, the drug with the precautions on these organs might be crossed off the list.
What we learned
A better understanding of the source of healthcare data and the ethical predicament proposed by it. We also learned that the establishment of a database involves many details, how to integrate information from different sources into clean and consistent data.
What's next for BenefitBase
Try to find a smarter way of processing the files using NLP methods. Collect different templates of EOB from major companies to achieve the automatic cleaning of the data. Try to make more accurate predictions using the built dataset by data mining and machine learning methods.
Log in or sign up for Devpost to join the conversation.