[[2 Oct 2024 - HIMSS APAC Day 1]] ## Speaker: Dr Shankar Sridharan, Chief Information Officer at Great Ormond Streeet Hospital, UK ![[Pasted image 20241012162302.png]] ![[Pasted image 20241012162311.png]] ![[Pasted image 20241012162332.png]] Summary: - The speaker shared about his “three laws of healthcare AI” inspired by the Hippocratic Oath and Asimov’s Three Laws of Robotics. - He shared about the solution of generative AI being deployed in his clinic that can listen, transcript, summarise clinic consultation. - He said that it is 95% accuracy, he say that there is no perfect system, and there is no need for perfect because, even if there is 100% accurate, will the human clinician not want to check? Then who is responsible for it? - AI can be like an assistant that prompt clinicians to ask questions, or remind them to check for other symptoms based on the context. - Benefit include : 1. Reduce administrative, cognitive load on doctors to write notes, letters, (2) increase doctor-patient eye contact, - He said that not all solutions need to be highly sophisticated, to have “digital maturity”, solutions such as transcribing notes can be deployed regardless of EMR system. - In his practice, he noticed that doctors actually spent more time using EHR system now that they don’t have to spend time writing the notes. It frees up doctor time to use other features in EHR. - The audience asked about any pushbacks from clinicians/patients? - He said that trust need to be established, explain to patient that data are not used in training model. It is used in his laptop to listen and transcribe notes. Records are not saved. - What about accuracy, translation, language? - He said that need to train the model to understand different language, he believe that once it’s solved, imagine consulting with a patient across different language. - Karlina asked about the challenge of managing “hallucinations” and inaccuracy. - The speaker acknowledge that there is a need to continuously evaluate and check the system to ensure that it works. There is a need to put in place governance, processes, systems especially when the technology evolve and improves, there is a constant need to train and check. - What about other use cases, such as for nurses? - Yes, it will help nurses because there is too much administrative tasks for them. There is gap now. It shouldn’t just benefit a few people. - Can transcription auto populate into EMR, such as adding medicine list, orders? - That is next step, to reduce friction, to auto add the notes into EMR - Lastly the speaker mentioned that he find it interesting that that people expect a higher standard when it come to digital tool, people ask him do we need to start recording telemedicine (zoom sessions), he asked do we usually record clinic sessions? Do we need to tell patients that they are using AI tool? Do we tell them where they learnt their medicial knowledge, from which textbook from which page? It is a tool, and it is the human clinicians who is making the final decisions.