top of page

Selecting User Data for Xema

After adding Google sign-in to Xema, as mentioned in my last blog, I was posed with few more questions to make progress on the project:


What user information would be helpful for the LLM to provide relevant responses to customers’ questions, such as understanding one’s eczema symptoms when traveling to a new, unknown location?


  1. My first approach was to seek ongoing feedback from the user about their daily eczema experience. This feedback, combined with other user information available from the user’s device such as their location and weather conditions (temperature, humidity, etc.) would allow me to build a user history about their eczema experience for various geographical and weather conditions. However, building such a user history would take some time and would prevent the LLM model from providing relevant responses in the absence of such information.


  2. After brainstorming with friends and family, I thought of another approach and decided to seek one-time feedback from the user about their eczema triggers. This would allow users to share how their eczema generally symptomizes for weather conditions such as high, medium, and low, temperatures, and a few other weather parameters. Having the user’s trigger information along with forecasted weather parameter conditions for a travel location could allow the LLM model to more easily predict and respond to the user’s eczema symptoms while traveling.


How do I gather such information from the user?

For my first approach, I decided to build the following simple feedback UI, that would be prompted to the user daily:



The smiley icons are something a user can tap on to provide their response with a single action. For triggers, I decided to build a single app page that the user would fill out before initiating a chat with the model.


It has been an exciting and satisfying experience working on the project so far. My current focus is to complete an initial version of the app and make it available for Android as the first step. I hope to complete the initial version and share a few glimpses of the Xema UX in the future.

Recent Posts

See All

Data Storage in Xema

Xema's development is coming along well, with the app now working for multiple users and the LLM model able to respond to a few eczema...

Multi-User Support for Xema

I mentioned previously that I was able to make the Xema app work successfully with Google’s Gemini LLM model. My goal for the app is to...

Large Language Model for Xema

Choosing the LLM Model Today, there are many large language models to choose from, such as OpenAI’s ChatGPT, Google’s Gemini, and Meta’s...

ความคิดเห็น


bottom of page