Using GenAI and Hybrid Search to Augment Medical Data Pipelines
Renaissance
The
AI
Victoria Slocum
Machine Learning Engineer
twitter.com/victorialslocum/
/in/victorialslocum/
1
AI native tools
Technical deep-dive
Use-cases in Healthcare
2
3
different types of
in
different types of
need
different types of
data
use-cases
storage systems
Weaviate
AI-Native Vector Database
Custom HNSW implementation
Integrations into the AI ecosystem
Enterprise features for scale and security
Different applications
require
different tools
It's not too hard to build a prototype...
but much harder to move it into production
Ease of Use
Customizability
Out-of-the-box functionality that automatically works
Different parts for the user to build and adjust best for their use-case
v.s.
1
Upload your own vectors
Use an open-source model
Use a built-in integration
2
3
Vector Search
Keyword search
Hybrid Search
- search from meaning, not exact words
- handles typos out-of-the-box
- AI use-cases like chatbots
- exact matches in response
- handles short queries and technical jargon
- better response to out-of-domain data
Generative search
Generative search
The Problem
Too much data for doctor to intake
Patients still need to fill it out for records
Creates frustrating lose-lose situation
The Solution
The Solution
DEMO
AI's Dual Role
In Information Management
Victoria Slocum
Machine Learning Engineer
twitter.com/victorialslocum/
/in/victorialslocum/
