Virtual Cabinet

Inspiration

Weak health literacy regarding medications leads to adverse drug events (ADEs) in every age range through drug-drug interaction and drug incompatibility. Adverse drug events account for 700,000 emergency department visits and 100,000 hospitalizations each year.

Although both prescribed and non-prescribed (over-the-counter, OTC) medicine cause ADEs, prescribed medication tends to be regulated systematically, whereas non-prescribed medicine is less overseen at the point of use. So, we have focused on developing an app that could help decide which OTC medicine is the best choice for certain situations.

What it does

The target users are the general public, as most households have several OTC drugs. Also, specific users, such as caregivers, can use Virtual Cabinet for deciding the best medication in an immediate or emergency situations.

From the list of OTC drugs that people already have, it recommends the best-fit medications. It filters out the types of medication by considering the patient's current symptoms, the compatibility with the prescribed medicine that the patient already takes, and the specific age range for certain drugs. On the page where we show the information of the drug that is screened, the essential information is included, such as dosage calculation and method of administering. This will reduce the error caused by drug administration, and ultimately reduce the personal and national healthcare costs.

We are aiming for social benefits by saving up to 70% of the costs used in adverse drug events.

How we built it

We used an android studio platform to develop our "Virtual Cabinet" using Java. Its user data are stored in Firebase Realtime Database, and they are used for authentication, verifying and comparing data to set up the appropriate medicines. We have applied a way to add different kinds of medicines that users already have by taking a picture of medicine and recognizing texts using Cloud Vision API.

Challenges we ran into

Firebase was used to deploy the authentication process and real-time database was applied to allow low-latency access to the inserted user data. The first design of the application used Firebase Machine Learning API for text recognition, but the library was deprecated. So, in the final design, we used Google Cloud Vision API instead of the Firebase Machine Learning API to automatically recognize and store the name of the drug into the firebase cloud.

Accomplishments that we're proud of

Our group successfully implemented the function for Google Cloud Vision API using Camera and was able to recognize accurate text and save it to the users’ data through Firebase.