TriageRx

TriageRx is an AI program that guides users to the right medical professional, whether it’s a pharmacist or a doctor, for minor ailments.

Comment

Inspiration

As of June 1, 2023, pharmacists are now able to diagnose and prescribe medication for 21 minor ailments in British Columbia. That said, it can get confusing to differentiate when to consult a pharmacist verses a doctor. This new adjustment and confusion then leads to unnecessary visits to both doctors and pharmacists, wasting both time and inefficient use of healthcare resources. With this project, we sought to find a way to improve efficiency in healthcare and promote health literacy among patients.

What it does

TriageRx is a preliminary analysis that uses a custom AI model to assess a series of patient symptoms to better guide patients to suitable medical professionals. This is done by comparing the patients symptoms to a dataset to determine if the patient’s condition aligns with one of the 21 minor ailments pharmacists are able to prescribe (in BC). If so, patients are recommended pharmaceutical care where over-the-counter prescription can be performed. In addition, the chat bot is also able to answer questions and further describe the conditions of the patients and the reasoning behind its recommendation through the use of OpenAI and LIME. This chat bot is available on TriageRx’s website and soon to be available on the TriageRx mobile app, people are able

How we built it

To begin, our machine learning model is created using TensorFlow, a software library for ML and AI applications. We used a dataset from Kaggle, a reliable platform within the data science community.

The dataset contained 4,921 rows. We preprocessed this data by encoding non-numerical variables using scikitlearns LabelEncoder, meaning we formatted all the data be in a way that the algorithm can understand. We then trained a neural network on the model. Neural networks are computational models loosely modelled off the human brain, and they have layers of “neurons” that find patterns in data and make predictions. After training our model, we got an accuracy of 0.89, which is considered “realistic and consistent with industry standards”.

Our backend process was made with patient privacy in mind and uses binary digits to ensure that no personal information is exchanged during the process. The website uses “0s” and “1s” to represent the users’ symptoms which is then fed to the AI model where it can make a prediction of the users’ condition and whether they should see a pharmacist or a doctor. If the predicted condition is one that can be treated by pharmacists under the new Canadian regulations, the model will suggest the user see a pharmacist. Our model was developed using Python in Jupiter notebook, then exported as a Tensorflow.js model to be used in the frontend.

Accomplishments that we're proud of

We are proud of the proud of channeling so many of our technical skills into this project, from implementing API to making 3D prototypes on Blender!! We are also very proud of how we made things from scratch, from the website to creating our own AI model!!

What we learned

We learned that the key to working efficiently is to channel each team member’s skills and working simultaneously on different aspects of the same project to come together and create something bigger. We also learned valuable technological skills like how to create a custom functional chatbot.

What's next for TriageRx

TriageRx plans to pair with kiosks located outside all pharmacies in Canada that display TriageRx’s symptom-analysis test to determine whether or not pharmaceutical care is required or if redirecting to a doctor is a better alternative. If the analysis concludes pharmaceutical care, a ticket will be printed out for the patient to bring to the pharmacist to receive treatment. This program will also enhance the efficiency of pharmacists by streamlining their workflow and enabling them to allocate their time more effectively to patients in need. With the implementation of these kiosks, patients will undergo a comprehensive symptom-analysis test provided by TriageRx, effectively acting as a preliminary triage process. As a result, individuals seeking consultation with a pharmacist will already have a better understanding of their symptoms and potential treatment options.

By having this preliminary triage performed through the kiosks, pharmacists can benefit from the reduced number of unnecessary patients. This allows them to focus more on providing specialized care to those they can treat and delivering personalized prescriptions to patients.

This program will also be integrated into pharmacy systems through devices such as iPads, laptops, and computers providing pharmacists with convenient access to diagnostic support. This technology empowers pharmacists to make more accurate and informed diagnoses, ensuring the provision of optimal pharmaceutical care.

Built With

Share this project:

Updates