Medify started its life as a humble idea for this year's Hack Kosice hackathon. Our team of four looked for a problem domain that would align with the hackathon's challenges. We wished to improve the day-to-day lives of people significantly. After some consideration, we decided to focus our efforts on healthcare.

What it does

Medify is a natural language interface and personal healthcare app designed to ease general practitioner visits for both the doctor and the patient. It listens to the two of them talk and takes notes of essential points in the conversation. It presents these highlights via a mobile interface accessible to both sides, raising the patient's awareness of their health state. By taking the note-taking out of the list of the practitioners' responsibilities, Medify saves their time and lets them focus on the conversational aspect of the appointment.

How we built it

We leveraged Google's DialogFlow framework for natural language understanding, then developed an Android app and integrated it with the NLU solution.

Challenges we ran into

Bridging the gap between cloud (DialogFlow) and mobile (Android) was undoubtedly the hardest part. The original DialogFlow library for Java is no longer maintained, and its code-base is ancient. There is virtually no other way to connect the two without resorting to some intermediary step (such as a web service or a cloud lambda function).

Accomplishments that we're proud of

We're glad we brought Medify to life! We weren't sure we could do it, but it turned out better than we expected it to. It's great to see something working this well after just 24 hours of work.

What we learned

Working with DialogFlow and Android was a new experience for us. It gave us insight into the inner workings of AI-powered software such as Google Assistant. It was great to spend the hackathon by the sides of so many young and talented engineers. We hope we've improved our teamwork skills too!

What's next for Medify

There's a lot of potential in a project like Medify. We think it could be polished into a very nice and useful product. For that, however, we will need feedback from real doctors.

Built With

Share this project: