Iris | Devpost

Our inspiration

According to the World Health Organization, an estimated 125,000 preventable deaths per year are caused by medication nonadherence. Two of the main causes of these deaths are patients lacking belief in their need for medication and having fear of a medication’s possible side effects. On top of this, forgetting to take a dose of their medication accounts for 39% of all patient nonadherence. If patients can conveniently access easy-to-digest information about their prescriptions, they will be more likely to take their medications. In addition, patients will greatly benefit from medication reminders to further increase patient adherence.

What it does

Iris is an SMS chatbot that patients can text to gain easily understandable information about the medications they are taking. With the help of a database of pharmaceutical drugs, Iris will provide information that includes a medication’s active ingredient, purpose, and potential side effects. The doctor can also provide additional patient-specific information about their prescriptions. Iris will then be able to text back information such as the total medications prescribed, proper dosage, and when each medication should be taken. If such information is provided, the patient can opt for reminder texts to be sent when it is time for the patient to take their medications.

How we built it

See our GitHub and follow our tutorial.

What we accomplished

We accomplished an SMS chatbot that serves a real clinical application - providing a portable, reliable, and accessible source of information for patients about their prescriptions. As this is our team's first hackathon, we also accomplished starting a project from scratch and finishing with a usable product.

What we learned

This process taught us skills in presenting and marketing a solution. We learned new technical and debugging skills necessary to create this project, and this project also gave us exposure of the whole process from conception to design to pitch. We also gained insight on issues surrounding patient compliance in healthcare. We learned that using an SMS-bot can make medical information much more accessible than it is now, especially for those lacking literacy in technology.

Challenges we faced

A challenge we faced was figuring out how the patient might interact with our chatbot. We had to carefully think about what questions a patient may ask before deciding how the bot would understand. See our GitHub for challenges regarding the code.

Building upon Iris

Due to time and knowledge constraints, we were unable to implement Google Cloud BigQuery in Iris. Ideally, Iris would query the RxNorm database so she can obtain more detailed information about patient’s prescriptions. Then, she can simplify the information and text it back to patients. This would improve the autonomy, versatility, and utility of Iris and the questions she can answer.

Further steps would include:

Alternative names for pharmaceutical drugs (brand names vs generic names)
Voice recognition so that people can call Iris
Forming a “bag of words” so clusters of words can be grouped together in case patients phrase questions differently
Set up a username/password system to further protect patient privacy

Built With

Submitted to

MedHacks

Created by

I coded the majority of the app and also wrote the tutorial.

Spencer Chan
I helped research the issue then create the pitch and presentation

michelleUMD Fang
Computer Science & Biology Major
I worked on the background research into medication nonadherence, the questions and answers of the chatbot, and the slides of the presentation. I also voiced over much of the presentation.

Surya Padmanabhan
Tanaya Jha