Hello health people! here is our repository for Bangkit 2021 Capstone Project. we are making a medical chatbot (Medichat) at this capstone. Medichat is data-based health chatbot model from the alodokter platform
| Nama | Bangkit-ID | Path |
|---|---|---|
| Ibrahim | M111D4KY3323 | Machine Learning |
| Ryo Richie | M013D4KY2024 | Machine Learning |
| Muhamad Fauqi Al Azzami | M299D4KY3008 | Machine Learning |
| Syahrizal Raksa Negara | C393D4KY0794 | Cloud Computing |
| Aditya Suryandaru | C200D4KY0358 | Cloud Computing |
| Muhamad Hasbi As'siddiq | A393D4KY3766 | Mobile Development |
| Muhammad Fiqri Febriansyah | A393D4KY3761 | Mobile Development |
In today's digitally saturated era, abundant medical information often leaves people perplexed with technical jargon, conflicting advice, and even outdated information on the internet. This frustration often leads many individuals to resort to inaccurate self-diagnoses, potentially resulting in delayed necessary treatments or unnecessary interventions. However, there is hope on the digital horizon. An innovative solution has emerged: health chatbots. With user-friendly interfaces, these chatbots can help tackle these complex issues. Imagine a simple conversation: you describe your symptoms in easy-to-understand language, and the chatbot responds with explanations, potential diagnoses based on extensive medical knowledge, and recommendations for next steps. While health chatbots are not meant to replace doctors, they can serve as valuable virtual assistants. For medical professionals, these chatbots can assist in handling basic inquiries and offer initial triage, allowing them to focus on more complex cases. On the other hand, for users, these chatbots provide 24/7 access to healthcare assistance, offering initial guidance and reducing anxiety while waiting for appointments with doctors. By combining medical expertise with advanced technology, health chatbots can provide an enjoyable and trustworthy experience for anyone in need. With their accessibility and ease of use, these chatbots become valuable first points of contact for those requiring assistance with their health issues. This is our capstone project.
- Questions and answers related to health via chat
- Question History
- Doctor Speciality Recommendation
-
SCREENSHOTS/DEMO VIDEO: https://www.youtube.com/watch?v=r4o6negbJQ0
-
DATASET LINK: https://huggingface.co/datasets/ryorichie/medicalquestion
-
DEPLOYED LINK: Our app : https://github.com/Bangkit-Teams/MediChat/tree/main/Android Our trained model : https://huggingface.co/ryorichie/MediChat_Medium API request URL : https://generate-response1-t7yc42rinq-uc.a.run.app/llama API recommendation URL : https://recommendation-t7yc42rinq-uc.a.run.app/recommendation App release : https://github.com/Bangkit-Teams/MediChat/releases/tag/release
-
10-MIN PRODUCT PRESENTATION LINK: https://youtu.be/j-lmfiEuWyA
Strengths: Our product can provide first treatment in an emergency and find the right doctor so that the doctor's time is not wasted if he sees an unsuitable doctor. Our LLM is trained with Indonesian language and gives a related and fairly accurate response. Weaknesses: Because of limited resources, our LLM have high response times and can’t give follow-up answers because of resource limitation. Opportunities: Indonesia is a country with one of the fewest doctors per capita in the world, we need to use these limited resources as efficiently as possible. So often we meet the wrong doctor that does not relate to our symptoms and that is a waste of time, resource. Threats, LLM can’t replace doctors, using it is an act of self-diagnose. We can be wrong with our symptoms and LLM can give false diagnostics, we need to use this model with caution.