Health Grade Search

Clinia's Health grade search harnesses the power of AI and vector search to deliver highly relevant, accurate, and context-aware search results across various healthcare data sources.
This enables building tailored and efficient search solutions on provider information, medical records, or clinical data.

What is Health Grade Search?

Health Grade Search (HGS) is an AI-driven search platform designed for healthcare applications, enabling healthcare providers and organizations to efficiently search structured and unstructured data. Leveraging vector embeddings, HGS provides scalable and precise search capabilities that can be integrated into existing healthcare systems.

Key Use Cases

• Provider Search: Locate healthcare providers using customizable filters such as specialties and availability.
• Information Search: Retrieve clinical documents, research articles, or structured healthcare information from large datasets.
• Electronic Medical Record (EMR) Search: Efficiently access patient-specific information within EMR systems for enhanced healthcare delivery.

How It Works

HGS operates through a combination of Ingestion Pipelines, Embedding Models, and an API Endpoint to provide an efficient and scalable search solution. Here's a technical overview of each component:

1. Ingestion Pipeline

The ingestion pipeline is responsible for processing incoming data into a format compatible with HGS’s vector search algorithms. Key features include:

• Segmenters: Break down large text into smaller segments for optimized indexing. Use the default segmenter or integrate your custom solution.
• Vectorizers: Convert text into vector embeddings using Clinia’s models. Model selection is configurable, allowing for specialized use-case tuning. Current models include embedder_medical_journals_qa (more to come, such as embedder_providers and embedder_emr).

For detailed pipeline configurations and setup, see the Ingestion Pipeline Documentation.

2. Embedding Models

HGS uses healthcare-specific embedding models to convert ingested data into vectorized representations, enabling efficient search and retrieval.

• Current Models:
  • embedder_medical_journals_qa: Optimized for medical literature and clinical documents.
• Upcoming Models:
  • embedder_providers: Optimized for provider search.
  • embedder_emr: Focused on querying electronic medical records.
    Embedding models are continuously updated. More models will be released and improvements in the upcoming months.

3. API Endpoint

HGS exposes a REST API that supports vector and hybrid search methods. With the API, you can:

• Execute searches using client-configurable models.
• Customize response formats and highlighting rules.
• Integrate seamlessly into your existing systems.

4. Ranking and Future Enhancements

Search results are ranked based on vector similarity. Updates will be rolled out as new features become available.

Getting Started

To get started on configuring the HGS in your environments, the following steps are

1. Configure the Ingestion Pipeline: Set up segmenters and vectorizers according to your data and use-case requirements.
2. Select Your Embedding Model: Choose the appropriate model for your search context.
3. Query the API: Perform search requests using the API and refine results using available customization options.