Novel Feature: AI

[calejvaldez]

Summary

Note

This is the issue we're using to keep track of the AI novel feature discussion for Iris. This is also related to the CHI Student Design Competition.

The Developers team is going to be implementing AI to Iris as an opt-in feature. (That is, it's disabled by default.)

These are some of the ideas we have:

• Summarize all of the patient's Iris data for nurses (Daily report summary)
• AI starting a conversation about grieving with the caregivers
• Personalized care
• Generate resources (May be outdated by ~2 years)
• FAQ about Iris
• From @Electric108: Create an AI summary of the title for Care Instructions to create a detailed summary

Important

Ojos Project team: If you have another idea, please add it as a comment to this issue and I'll add it to this list!

Concept images and/or code

An idea I had for this would be to use one of Meta's Llama models since their usage would be hosted locally on-device instead of being processed in a server like OpenAI's models do.

[AyushJain2021]

• Text-To-Speech functionality (may not need AI)
• Natural Language Processing (NLP)
• Doesn't need to be Llama, but MUST be locally hosted.

[jessed7]

• Summarize the resources linked in the "Resources" page
• Generate summaries related to the past data entries of the PROs submitted by the patient, potentially with suggestions to help with some issues (if possible)
• For loved ones who may be concerned for the patient, generate summaries of how the patient is doing and/or how they are being taken care of (there are probably HIPAA concerns with this)

[calejvaldez]

https://doc.rust-lang.org/std/process/struct.Command.html

[AyushBot412]

Storage Capacity | RAM Capacity

Current:

• Storage: 32 GB
• Memory: 4-8 GB

Needed:

• Storage: 512 GB
• Memory: 12 GB

Storage Usages:

• Video Files
  • Quality
  • Length
• Patient Data
  • PROs
  • Medications
  • Care Instructions
  • Resources
  • User Information
• Settings

[calejvaldez]

TODOs:

Features

• Care Instructions: summarization from CI title

• Patient data summarization

• Sorting resources (doc incoming...)

• Model: Llama