Alpo Bots - a chatbot for patients and doctors

Isn't it so tiring and gloomy waiting for your hospital queue and don't even know yet it is the right doctor to see? Here comes Alpo Bots, a chatbot to help all the processes and get to your needs!

Comment

Alpo Bots empower the most important time and place of communication, and that is face to face communication when the doctor and patient meet. By facilitating the major administration process upfront with ease, doctors and particularly patients, can focus and get the most out of the examination session.

Inspiration

Alt text Learning from the fish bone diagram, every hospital has similar problems. Presented in the diagram is a situation where patients are waiting too long to get into ED, the most crucial rooms to treat patients immediately. Looking at this, we found just the same problem in the usual waiting room for patients who seek medical consultation. We learned that solving the ED problem needs mostly efforts from the hospitals themselves and deeper analysis on the system design, we found opportunity in the usual patient queue which has a lower time urgency but could be solved via Alpo Bots. Why do we let this long process halt us to start a more meaningful conversation with the doctor to get our treatment faster? We think, these administration processes should be able to be sorted out without going through the long and boring queue that requires a lot of time in the hospital, and this is true particularly in Indonesia.

What it does

The main capability is to enable patients to report and submit health symptoms and problems, and then register it into the hospital’s database, then the data can be accessed by the doctors. Alpo Bots is also able to remind users about their upcoming appointments and manages all the schedules and messages between the doctor and the patient. We build Alp oBots to accelerate the administration process, so that patients can access the core and more meaningful communication with the doctor. Alpo Bots is an end to end chatbot system to assist patients who are looking to make doctor appointments using Facebook Messenger as their first touch point. Patients can consult symptoms to get a better doctor recommendation and apply for a queue number before coming to the hospital to save time. Likewise, doctors, nurses and the hospital can get data entry in advance and read through reported symptoms for a clearer understanding, saving time for the actual examination of the patient. This way, we lessen the waiting time and prepare a fixed schedule for the doctors and patients so everyone knows what’s next. With this, we also finished all the administration work upfront and patients do not need to drive to the hospital or call, they just need to chat with Alpo Bots and could continue whenever they have the time.

How we built it

We build our backend services using Python and serve it on Heroku. The backend services consist of API service and Scheduler service. API service is used to handle webhook events from Facebook Messenger and give responses that are in accordance with our bot main function. Scheduler service is used to handle notification for users (patients). From the Facebook Messenger features that are required to be built in our app, we choose to use One-Time Notifications and Quick Replies.

One-Time Notifications are used for reminding patients one day before their appointment (For this competition, at 11:00am GMT+7. Of course in real life usage should be on local time).

Alt text

Quick Replies are used in cases where we can suggest choices of answers for patients, so they don’t need to type manually.

Alt text

Challenges we ran into

At first we thought to make an extensive flow, but apparently there are a lot of scenarios to consider, that is why we focus on the completion of the main consultation flow. We need to make our own word detection library for the body parts and symptoms, that is why the bot can only still detect a handful of words for now. We made some quick wins work around for the codes and it is not as clean as we wanted. The other thing is the chat recording, we recorded how it would show the usage on real users, but the video limit is too short, that is why we resort to only screen recording on the chat flow.

Accomplishments that we're proud of

Being able to detect words sparks the opportunity we see this bot can do in the long run. Alpo Bots can make multiple appointments, and the reminders capability helps users for their upcoming appointments. We are able to complete the consultation flow and enable direct back and forth from users to the doctor. This reflects the collaboration of back end engineering and design principles. We are also able to use three features that are required in this competition. Those are One-Time Notifications and Quick Replies. We also made some mock assets inspired from Facebook's product range (when the engineer is so absorbed in the coding and the designer has nothing to do :D), and of course the Alpo Bots logo.

What we learned

As a proof of concept it is quite smooth, but we tested it out, users’ still expect it to be a smoother experience. Of course we learned how to use some features from Facebook Messengers (especially One-Time Notifications and Quick Replies), learned a lot about medical concerns and the problem we are trying to solve. Although doctor referrals are common, we learned that apparently systems in hospitals are different, this gives another layer of problems and opportunity to solve.

What's next for Alpo Bots

We envision that Alpo Bots can become a chat bot framework that can be utilized by different hospitals and then grow as a centralized system to easier schedules and manages patients transfer from one hospital to another, that ultimately will lessen time for all administration work to healthcare. Users can contact different hospitals and Alpo Bots will handle all the data and schedule transfer of one hospital to another with an integrated central database so that patients can be easier to be referred to another hospital with all the administration done upfront.

Testing Flow

Below are the step by step to test the product. To test the patient flow:

  1. Search F.B. Mesnegers Hospital (or access this page https://www.facebook.com/FB-Mesnegers-Hospital-114487943639689)
  2. Tap get started
  3. Tap Menu
  4. Choose Consultation to test the consultation flow.

Alt text

To test the doctor flow, first we assume that doctors have to register in order to use Alpo Bots’ service.

  1. Search Alpo Bots (or access this page https://www.facebook.com/Alpo-Bots-101722244928623)
  2. Tap get started
  3. Tap Menu and choose Register as Doctor.

In this testing case, we simplify the registration. Testers only need to choose Register as Doctor and they will get notified every time a patient creates an appointment, for all doctors. Of course in real life usage, the registration will be more detailed and doctors will only get notification when a patient creates an appointment with them.

Alt text

After registering as a doctor and after patient creates an appointment, by accessing the Menu and choose My Appointment, testers can accesses the after-appointment flow (send medication prescription to patient).

Alt text

Built With

Share this project:

Updates