Our Technology

Kitsa Seeker: Fine-tuned LLM for precise EHR and Clinical Trial Matching

At Kitsa, we leverage cutting-edge artificial intelligence to solve one of healthcare’s most critical challenges — matching the right patients to the right clinical trials at scale. Our proprietary fine-tuned Large Language Model (LLM) – Seeker – surpasses the performance of leading open-source and proprietary models, including GPT-4, specifically for clinical trial matching applications.

Here’s how Kitsa Seeker sets a new benchmark in clinical trial recruitment:

Superior Contextual Understanding of Medical Data

Seeker is uniquely optimized to understand complex medical terminology, structured and unstructured patient records, and nuanced trial eligibility criteria.

This capability accurately identifies eligible patients, even when trial requirements are intricate or poorly defined.

High Precision and Recall in Matching

Seeker minimizes false positives and negatives in patient-trial matching by outperforming other LLMs in precision and recall.

This leads to higher enrollment rates and improved trial timelines.

Tailored for Clinical Trial Protocols

Unlike generic models, Seeker has been trained on data specific to clinical trial protocols and historical trial performance, making it contextually aware of protocol requirements and patient characteristics.

Continuous Learning and Improvement

The model continuously evolves, learning from past trial data and feedback loops, ensuring it stays at the forefront of AI advancements in healthcare.

Scalable and Efficient at the Enterprise Level

Capable of processing millions of patient records and hundreds of complex trial criteria simultaneously, Kitsa's solution offers unmatched scalability for global clinical trial operations.