The surprising link between retail and healthcare. And how it’s saving doctors thousands of hours.
Leaning on AI to ease the burden when your healthcare system is at capacity may not be a natural reaction. But by turning to it, and collaborating with a Scottish NHS Regional Authority, we were able to reduce the load on NHS Lothian’s doctors, and ultimately improve patient care.
The numbers don’t lie – there’s a growing challenge in healthcare. From December 2015 to 2018, one of the most pressured specialties was Gastroenterology with the total number of patients on the waiting list for a new appointment increasing by 20%. This area saw outpatient demand double over 10 years from 2007. Moreover, in December 2017, 30% of patients were waiting longer than the standard 12 weeks before seeing a doctor. These were the significant challenges faced by NHS Lothian.
Increasing capacity by employing more doctors might seem like the only solution. However, we sought to find an answer in an unlikely spot. Our AI Studio noticed that there were parallels between the way doctors review patient letters and how retailers review customer communications. Every day, customer service staff sift through customer enquiries to decide the priority and type of query the customer has. Not too dissimilar to what specialist doctors are doing when reading through patient referral letters to decide how urgently a patient needs to be seen and for what.
For Kishan Pattni, who works within the AI Studio, this was the lightbulb moment, realising that some of the techniques used in training AI to learn the meaning of customer service queries, could be repurposed to interpret medical referral letters.
It’s something that we knew we could help the NHS automate, having previously used AI to solve the same act of triaging in the retail space, but whilst we were confident in the concept, we didn’t want to underestimate the scale of the challenge. Our AI Studio ventured on, working in collaboration with doctors, hospital process experts, and regulatory specialists to build a solution.
The Referral Intelligence and Triage Automation tool learnt from 20,000 referral letters and predicted clinical triage outcomes and urgency status with up to 80-90% accuracy for patients referred for suspected cancer. It’s a tool that on average removed 2 days from the patient pathway, enabling quicker access to appointments. It also saves around 4 minutes of the specialist clinician’s time per referral. And when you’re receiving thousands of patient referral letters per year, those minutes add up.
Drawing similarities across industries to solve problems in unique ways is something that our Ventures Studios thrive on. And it’s easy to see why. Solutions like these can have a huge impact. Not only for the businesses seeking the solutions, but for their customers, patients and society at large.