Introduction
A successful colonoscopy requires ideal stool preparation, yet the percentage of insufficient bowel preparation is still very high. Frequently, the patients use website information on colonoscopy preparation. Conversely, the daily activity of the hospital staff consists of spending time with patients, communicating with them and giving answers to a high number of questions coming from patients regarding the preparation for the procedure or the exam itself. Hence, the idea to adopt a rapid and intuitive patient-oriented digital service to respond to their needs, the virtual assistant Eric.
Aims & Methods
Endo-Chat is an automated online communication platform, powered by AI (Artificial Intelligence) with the use of the Natural Language Processing and Deep Learning technologies, that answers procedure-related questions in real-time. The engine is named ERIC, stands for "Endoscopy Robot for Intelligent Chat" and is an intelligent chatbot, capable of answering autonomously to most of the doubts of a patient who is supposed to undergo an endoscopic examination.
Results
Since the introduction of the conversational artificial intelligence system at the Policlinico Gemelli starting from 24th of January 2022 to the 16th of March 2023, Eric has been able to autonomously manage 94% of the questions sent by patients, thus forwarding only the remaining 6% (of the requests) to specialized human operators.
We calculated that in the pre-pilot phase, a large part of the daily activity of nurses was dedicated to communicating with patients, around 40 hours per week, while in the post-pilot phase, the time spent for communication was significantly reduced to 15 hours per week. According to OECD Health Statistics 2022, the average annual salary of a European nurse is 35,300 euros, with an annual hypothetical saving of 22,062 euros, comparable to a part-time contract.
Also, the number of cancelled procedures for poor patient preparation has fallen from 22 to 7 per week, considering a volume of 200 procedures per week.
Regarding patient's experience, users can leave voluntary feedback on their virtual assistant experience directly on the ERIC website, evaluating the quality of the service. Patient service evaluations were enthusiastic till now, as 63% rated the service as excellent.
Conclusion
In our observational pilot study, the first in Europe to our knowledge, we assessed the benefits measurement for patients and Hospital staff. Starting from these preliminary interesting results, we can speculate that the introduction of Endo-Chat in high-volume endoscopy service is feasible and useful not only for patients but also for hospital staff.
Disclosure
The authors declare no conflict of interest