Identity theft represents a major problem when talking about multiple topics such as banking, but it's also a practice that has a negative impact in the medical and pharmaceutical fields. According to the BlueCross BlueShield organization, overdoeses lead to more than a million emergency room visits each year only in the U.S., and prescription medication fraud is one of the underlying factors that increases the problem. Mixing biometric identification techniques with digital prescription generation reduces the chances of fraud, potentially saving millions of lifes across the world.
What it does
MedicBot is a virtual assistant intended to help doctors during the process of generating all kind of medical reports and prescriptions through a 100% digital experience, without using any kind of paper. Thanks to Artificial Intelligence, MedicBot is capable to guide doctors through a conversational experience to obtain all data necessary to generate and send the reports and prescriptions digitally using Anvil technologies. Also, MedicBot is able to detect, at any moment, if the person who is talking with is the doctor itself, through the bevahioral biometrics technologies of TypingDNA. In case it detects something is wrong, the user willl be asked to take additional authentication steps before receiving the report.
How we built it
For the conversational part of MedicBot, we used IBM Watson Assistant, an Artificial Intelligence engine that allowed us to translate the report fields into simple questions and answers within a conversation. The construction of the reports, in addition to the sending of emails, was developed with an API in .NET Core that connects through API calls with Anvil services. Finally, a Django app was developed to orchestrate the communication between the users, the artificial intelligence engine, the TypingDNA API and the .NET API.
Challenges we ran into
One of the challenges we faced was to include authentication with behavioral biometrics without hindering the user experience, but always making sure that it contributed to the security of the system. In the end, we managed to include part of the doctor's answers as input for validation with TypingDNA, making authentication in a certain way invisible to users.
Accomplishments that we're proud of
We are proud to have been able to build a solution developed with technology that we work with on a daily basis, such as Artificial Intelligence, to facilitate the daily life of physicians and contribute to the reduction of paper use within the healthcare industry.
What we learned
We had all used biometric authentication methods such as fingerprint or facial recognition, but with this challenge we discovered the world of behavioral biometrics and how it is used in industries that require high levels of security such as banking or medicine.
What's next for MedicBot
MedicBot could be extended to other types of medical documents, not only prescriptions, giving more flexibility to doctors. Even could help to fill documents related to health insurance services. Another thing we want to add is about security levels. Besides TypingDNA validation, we would like to include also IP and location validation, to create a more robust solution.