At its peak, the UK’s National Health Service (NHS) was covid testing almost half a million people per week. When demand for these appointments began to outstrip supply, search result relevance suffered. In some extreme cases, people were recommended to cross a body of water to get a test. This was a risk to public health as the NHS wanted to avoid anyone that had covid using public transport. To solve this the NHS needed to switch the way they filtered search results. Instead of using straight line (euclidean) distance, they wanted to filter results based on travel times. They also needed a way to tailor results based on whether the searcher had access to a car and ensure public transport was avoided. There were many technical challenges to delivering this kind of search. - High user demand - needed to be able to handle 100,000 users searching concurrently - Response times - deliver test centre locations in under 50 milliseconds - User data privacy - ensuring no customer data will ever be at risk - Security - ensuring no tampering with data There was no room for months-long stress tests. It needed to deliver on performance instantly. In my presentation I’ll walk through how we built this search under a super tight deadline. I’ll also walk through many other applications of search in healthcare. Including: - Managing Europe’s nursing shortages - Improving the efficiency of emergency services - Matching mobile doctors to patients