Executive Summary

The Barcelona-based AIS Channel team seeked our support to tackle a seminal challenge in the field of surgical education: personalising content delivery to individual surgeons worldwide. Our collaborative effort has led to the development of a sophisticated Semantic Search Engine, boosting search relevance by over 30% from traditional methods, and unlocking fast delivery of personalised educational resources for surgeons across the globe.

Challenge

AIS Channel's mission states "democratising surgical education", ensuring every surgeon has access to educational content on state-of-the-art procedures. To materialise this mission they have created a Netflix-like streaming platform where surgeons share talks, interviews & recordings of surgeries. However, a bottleneck emerged: the existing keyword-based search engine fell short in delivering precise content to individual surgeons, often requiring exact phrases to yield relevant results, forming a limitation in a field laden with technical jargon and synonymous terms.

Goal

Our objective was to transcend this barrier by implementing a Semantic Search Engine that could interpret the essence of queries, fetching pertinent content irrespective of the phrasing nuances.

Solution

• Analysis & Choice: Initial deliberations drew us towards considering Azure Cognitive Search. However, we soon switched gears to develop a custom-built Semantic Search Engine, yielding a remarkable increase in search relevance.
• Dual-Stage Precision: Our solution employed a two-stage approach – Initial Retrieval and Result Reranking, each powered by advanced Language AI models like BERT or RoBERTa, renowned for semantic comprehension.
• Performance Leap: The custom engine outshone Azure Cognitive Search by delivering a search relevance exceeding 80%, a significant leap from the 50% relevance of traditional methods.
• In-House Implementation: Beyond the technical solution, our engagement ensured the AIS team was adept at manoeuvring the new system, ensuring data protection, and fostering a sustainable, in-house AI capability, steering clear from reliance on external AI endpoints or experts.
• Training and Knowledge Transfer: A dedicated endeavour was made to ensure AIS's team was proficient with the new system, embedding a culture of self-sufficiency.

Next Steps

• Expansion & Refinement: Harness the feedback loops from user interactions to continually refine the semantic search capabilities.

Watch the interview with AIS Channel CTO Oliver Ignasi about the collaboration.