CareCounselor: A Chatbot for Disease Management Support
Company Information
A major nonprofit organization working on some of the world’s toughest problems like extreme poverty and disease eradication around the world, which also aims to expand educational opportunities and access to technology. In need of a personalized chatbot to aid specific chronic disease patient populations, the company engaged with ProCogia to create a solution that was flexible and reliable. The goal was to create an assistant that could guide patients towards better care, encouraging them to follow up their treatment and adhere to healthy behaviors, while ensuring utmost medically accurate responses.
The Challenge
Out-of-the-box chatbot solutions currently lack the reliability to be used in medical environments. For instance, LLMs need be programmed to recognize when a query requires a healthcare professional’s expertise and direct patients accordingly. They also struggle with engaging users because they present a one-size-fits-all communication tone. Patient reactions to information, preferred communication styles, and health literacy levels differ significantly. The chatbot is intended to be used in low-resource languages and the performance of the leading LLMs in these is limited.
The Results
The development of CareCounselor resulted in a sophisticated and highly effective chatbot designed for disease management support. A custom document processing pipeline was created to extract tabular data and section metadata from medical guidance documents, enhancing the accuracy and relevance of information provided. The fine-tuned embedding model, implemented within a Retrieval-Augmented Generation (RAG) system, effectively addressed hallucination issues, ensuring precise responses. Integration with a closed-source translation application allowed seamless translation between local languages and the English-language knowledge base, overcoming language barriers.
Additionally, the personalized segmentation layer and localization tool enabled tailored interactions and provided users with essential local information. The deployment of CareCounselor as a WhatsApp chatbot maximized accessibility, making it readily available to a wide range of users. Testing by medical experts yielded highly positive feedback, underscoring the chatbot’s potential to significantly enhance patient care and support.
Portfolio Management
Analyzed the current document parsing options and improved on them, ensuring the chatbot is provided with the most accurate information.
Identified that the generative model had a significant performance decrease when dealing with low-resource languages. To improve on this, we have developed a custom solution to translate from the user input language to English, ensuring the model can perform at its best.
Patient segmentation was achieved by adding an NLP layer that would categorize the user. The segment is then used to adjust the chatbot tone, creating engaging interactions.
Developed the project in AWS Sagemaker ensuring flexibility and easing client deployment.
Added a custom localization tool, ensuring that users could ask the chatbot for directions around their neighborhood to find pharmacies, clinics and hospitals.
Fine-tuned the embedding model used to power a Retrieval-Augmented Generation system and tracked the performance results on Sagemaker Experiments.
Conclusion
ProCogia’s delivered solution represented a significant leap forward in healthcare chatbots. The developed document content extraction pipeline exceeds the performance of off-the-shelf solutions. The final product received very positive feedback from testing by medical experts. ProCogia demonstrated that by personalizing an LLM-powered chatbot, it is possible to grant patients information availability, fostering a collaborative environment and transforming patient care experience.
For a more in-depth look at CareCounselor and the integration of large language models in healthcare, read our blog: Understanding the Role of Large Language Models in Disease Management Support.
Explore more stories
Dig deeper into data development by browsing our blogs…
CareCounselor: A Chatbot for Disease Management Support
A major nonprofit organization working on some of the world’s toughest problems like extreme poverty and disease eradication around the world, which also aims to
Implemented Bookmarking Apps
A leading investment consulting firm faced challenges in managing and sharing complex data analyses through their internal tools developed with Shiny. As the firm sought
Optimizing and Scaling RESTful API for an Investment Consulting Firm
Optimizing and Scaling RESTful API for an Investment Consulting Firm Company Information A prominent investment consulting firm sought to elevate the performance and maintainability
Get in Touch
Let us leverage your data so that you can make smarter decisions. Talk to our team of data experts today or fill in this form and we’ll be in touch.
