Meds AI

• GIF

• 

• 

Targeted Tracks → Healthcare: Patient Safety - Sponsored by PRHI

The Problem: Adverse Drug Reactions

Let’s talk about Adverse Drug Reactions, or ADRs.

Hospitalized Americans experience over two million serious ADRs, resulting in over 106,000 deaths annually (FDA). That’s over double the amount of deaths from vehicle crashes.

As someone who was bed-bound by a medication that only worsened my condition, I understand this problem all too well. And it exists outside of the hospital, too:

1 in 5 Americans take three or more drugs (CDC). Add in the hodgepodge of unregulated over-the-counter, vitamins, and dietary supplements that are skyrocketing in popularity, and we vastly increase the risk of harmful drug interactions. Even the most acute doctors can’t keep up with the millions of supplements on the market.

What if there was a way for patients to make sure that their health products don’t conflict with one another or any pre-existing conditions, in order to prevent ADRs? And what if we could put that solution in their pocket?

The Solution: Meds AI

Meds AI is an iOS application that analyzes a new health product against the user’s current regiment, alongside pre-existing conditions. Through optical character recognition and artificial intelligence, a user can simply snap a picture of the product’s ingredients to receive a recommendation before they even enter the checkout line.

User Flow:

Sign-in with Face ID — Security is paramount for patient information, and Face ID is the gold standard. Once logged in, users will be redirected to their health products page

• Users can scroll through their current list of medications, OTCs, and supplements
• Users can navigate to their conditions page, and add any relevant conditions
• Users can add a new health product through snapping a picture of the product and its ingredients. This filters through OCR and AI to evaluate the product against the user’s current regiment and conditions, ultimately recommending for or against the proposed product with explanation. The user can ultimately choose whether they wish to add this product to their regiment. In the future, Meds AI will integrate with EHRs like MyChart to quickly upload prescription and condition information during onboarding

How We Built It ⚙️

We have built a ML-enabled full-stack application that solves a real world problem.

Design: Paper & Sharpie to Figma!

Front-end: React Native, Expo, and XCode ML Stack: Google Cloud Vision API, Tesseract, OpenAI GPT-4

Prototype our designs on paper helped us receive iterative feedback before beginning development, saving us hours of time reworking code.

Design 🎨

We were inspired by the Double-Diamond design process, which not only includes visual design, but a full-fledged research cycle in which you must discover and define your problem before tackling your solution.

1. Discover: a deep dive into the problem we are trying to solve.
2. Define: synthesizing the information from the discovery phase into a problem definition.
3. Develop: think up solutions to the problem.
4. Deliver: pick the best solution and build that.

Research

While we knew very little about patient safety at the dawn of this hackathon, we began by reaching out to field experts.

First to patient safety consultant at the PHRI, Ariana Longley, MPH, who explained the pressing matter of patient safety and what falls in and outside of the subject area. This helped us narrow our ideas significantly.

Then to NYU Professor of Medicine, Dr. Damara Gutnick, who helped us center in on managing medications and over-the-counters, as it’s a process that begins before the patient even steps into the doctor’s office.

And finally, to AVP of Operations at Montefiore Medical Center, Maura Porricolo, MPH, who emphasized that the surplus of unregulated health products falls outside the domain of any one doctor. She commented that AI is an attractive solution to this issue due to its breadth of knowledge.

Our gratitude goes out to all of them! We also did excessive research on ADRs, chiefly relying on resources by the FDA, CDC, Institute of Medicine, NIH, and WHO.

Built With

• expo.io
• faceid
• google-cloud
• openai
• react-native