The Problem

In today’s digital era, patients increasingly prefer using instant messaging platforms like WhatsApp to communicate with their healthcare providers. While this offers convenience for patients, it poses significant challenges for doctors. The medical professionals were inundated with an overwhelming number of patient messages on WhatsApp, making it difficult to manage communications effectively. The sheer volume of messages led to delays in responses, increased stress, and potential oversights in patient care. Without an organized system to prioritize and summarize these messages, doctors struggled to provide timely and personalized care, ultimately affecting patient satisfaction and health outcomes.

The Solution

To address this challenge, we developed an AI-enhanced application specifically for doctors. The application mirrors patient messages from WhatsApp and adds functionalities that streamline the doctors' workflow.

1. Development of the Doctor's Application

We built a separate web application using Next.js for server-rendered React applications, ensuring fast load times. The interface was styled with Tailwind CSS to provide a responsive and modern design. This application offers doctors an organized platform to manage patient communications efficiently. It allows doctors to send and receive messages to WhatsApp and have everything they need within the app.

2. AI-Powered Symptom Summarization and Recommendation

Integrating a Large Language Model (LLM)—specifically the chatgph/medix-ph model—we enabled the application to process and analyze patient messages. Doctors can press a button within the web app to generate concise summaries of patient symptoms. Additionally, the AI suggests potential recommendations that doctors can review, edit, and personalize before sending, ensuring that each patient receives accurate and tailored advice.

3. Integration with WhatsApp

The backend, developed using Express.js and Node.js, is registered as a webhook for WhatsApp messages. It receives patient messages and allows doctors to send responses directly through the application. This integration ensures seamless communication, with patients receiving messages within their existing WhatsApp conversations.

Technical Overview

Below you can see the architectural diagram of the solution.

System Architecture

The flow is the following:

  1. Patients use WhatsApp to send and receive information about their symptoms
  2. Doctors use the AI-powered application to manage communication with patients
  3. Doctors can request the AI to summarize the symptoms of a patients
  4. Doctors can request the AI to provide a draft of recommendations to be sent to patients. Two separate generate requests are used for this functionality:
    1. One request for recommendations
    2. One request to summarize the recommendations - as we are sending messages to WhatsApp we want to keep them as short as possible
  5. Doctors can modify this draft and send it directly from the application. Users receive the recommendations in WhatsApp.

Results and Impact

The implementation of this AI-powered system significantly reduced doctors' workload. By automating the summarization of patient symptoms and generating initial recommendations, doctors save a significant amount of time. The streamlined process improved response times, increased efficiency, and enhanced patient satisfaction by providing timely and accurate medical advice.

Conclusion

Leveraging AI and modern web technologies, we provided an effective solution to manage the overwhelming volume of patient messages. The application empowered doctors to deliver high-quality care efficiently, balancing the demands of a busy practice with the personalized attention each patient deserves. This project demonstrates the potential of AI in transforming healthcare communication and improving outcomes for both doctors and patients.

Read more about our experience with AI.

Do you need an AI expertise at your company?

Check out the AI services we offer and don't hesitate to contact us for a free consultation.