Am I OK?

Screenshot of the website.

Inspiration

We all know that the local healthcare system has many issues, one of them being the inaccessibility of care in the public sector. People often leave the doctor's office with unanswered questions after being rushed out to keep the flow of patients going. Others have questions about their medical conditions and cannot find any appointments to speak to a professional. Our team wants to provide an easy way for people to learn and ask questions about the healthcare provided to them, their diagnoses, and prescribed medications and treatments wherever and whenever they want.

The first goal is to save healthcare professionals’ time by automating the dialog between a patient and professionals about their prescriptions. The second goal is to reassure and inform patients in an easy and accessible way, any time they want.

Cutting the time for such interactions makes time for professionals to focus on finding optimal prescriptions and care for more patients. This could help the healthcare system be more efficient and generally keep patients more informed on their medical situation.

What it does

Our overall idea is to connect the electronic patient files with a website accessible to the patients. After a visit to a clinician or other healthcare professional where they received a diagnosis or got prescribed new medication, our service would contact them through text messages to bring them to our website if they have unanswered questions about their condition and prescriptions. Possibly, a person could access this chatbot at any time by giving it their current medical information or by retrieving information from their medical files from their last visits to the doctor.

How we built it

We used the Cohere.ai API to generate the answers to the patient's question with different levels of complexity and depth depending on the patient's preference. We used Twilio to send the message that brings the patient to our website. Ideally, the message would contain a personal secure link so that the patient's information is already on the web page (extracted from the patient's medical file). We used Python with Flask to create a backend and frontend that connects with the Cohere API and retrieves Cohere's output and displays it to the user.

Challenges we ran into

We had many struggles while building the web app since we had little to no experience in web development. Unfortunately, a lot of time was spent figuring out the workings of dynamic web pages and the connection between frontend and backend.

Accomplishments and what we learned

We are proud to have a working locally hosted webpage and managed to have a working prototype despite our limited experience in web development. We learned a lot in this process and we could definitely produce a polished product in the future.

What's next for Am I OK?

There are many features we would like to add to Am I OK? to make it a more complete service. We want our product to be able to inform on everything related to their current health condition. Below are some of the features we would like to include in a final version of the product:

Information on normal symptoms that could occur
Contraindications
What symptoms to look out for in case of complications
Alternative symptoms relief methods
Local resources for additional treatment
Gathering feedback on the condition’s improvement and medication's efficacy
Security features to protect confidentiality

Additionally, we would like to add a separate function that is invisible to the patient but acts on the side of the medical professionals. A current problem in healthcare is the occurrence of erroneous prescriptions. Sometimes, the choice of medication or the dosage prescribed by the GP is inappropriate considering the patient's history and the medications they are currently taking. This gives rise to either complication and adverse effects manifesting in the patient, or delays in the pharmacy when the pharmacist needs to contact the doctor and correct the prescription. To avoid these kinds of problems, we would like to include another AI component that would evaluate the appropriateness of the prescription according to the patient's medical history. The goal is to save time and save lives!