"Once, after a meeting with the president and vice president, I immediately showed them the minutes of the meeting, including the to-do and tracking items." Talking about the opportunity to participate in the introduction of AI, Liao Jiade, director of the Center for Empirical Medicine and Medical Policy of Chi Mei Medical Center He smiled and said, "They also asked me why I typed so fast!" The answer was obvious, "You didn't see me typing from beginning to end."
This small attempt not only demonstrated the capabilities of AI, but also encouraged Chi Mei Medical Center President Lin Hongrong to encourage Liao Jiade to start researching related applications. In the past year, he has helped develop various systems that reduce the clinical recording time of medical staff by 1/2 to 2/3, such as A+ doctors, A+ nutritionists, etc., forming an "AI application circle" within the hospital. In July 2024 alone, the number of users in the AI application circle exceeded 50,000 in a single month, setting a new high, with an annual growth of at least 4 times.
Focus on the actual pain points of users and cut into application scenarios that meet the needs
Like all innovations and changes, it is inevitable to encounter resistance. Even if a tool is launched to help everyone save time and improve efficiency, it is "reviled" in the early stages of introduction. Liao Jiade, who was also exploring generative AI from scratch, intuitively thought that doctors and nursing staff would spend a lot of time writing medical records. Why not use the speech-to-text function to first record a description of the patient's condition and then automatically convert it into text? No need to enter data manually.
Unexpectedly, medical scenes often use a mixture of Mandarin, Taiwanese, and English, and the speech recognition technology is not accurate enough to make smooth switching; moreover, nursing staff are overwhelmed with taking care of patients, and they also have to remember to record their speeches. Some people I feel like, "I can type faster than I can speak."
"What we provide is not necessarily consistent with the real needs of the front line." This prompted Liao Jiade to rethink the application scenarios of AI. "I first observed what I hate the most. Is it word processing? Or system switching?" So he narrowed the scope, conducted user interviews, and even observed on-site to focus on the real pain points and target "repetitions or overlaps" in the process of writing medical records. "Highly sensitive tasks", such as summarizing, collecting patient information, etc.
After repeated trial and error and rapid iteration, the "A+ nurse" effectively reduced the time to prepare a bed shift report from 10 to 20 minutes to less than 5 minutes; the "A+ doctor" reduced the medical report that originally took an hour to complete. Now according to my notes, it can be done in 15 minutes.
Respect users' true reactions, give feedback today and make improvements tomorrow
In the AI application scenario conceived by Liao Jiade, it is best to cover all functions and levels of the hospital organization. Among the specific results of shortening working hours, the users who surprised Liao Jiade the most were nutritionists. "They told me that they can save 225 minutes a day." When converted, it is almost equivalent to half a day of working hours.
In the past, nutritionists had to refer to and integrate cross-professional systems such as doctors, nurses, pharmacists, etc. to collect information about the same patient in order to produce a nutritional recommendation; now, all information is gathered on the same page and automatically produced. Just give the patient's dietary recommendations and then modify them.
"The time saved can allow nutritionists to communicate well with patients." Liao Jiade pointed out that there is no need to worry too much about the issue of AI replacing human labor. After paperwork is replaced, not only time is returned to medical workers and patients, but also time is returned to medical workers and patients. for better quality service and consultation.
Regarding how to make former opponents or indifferent people willing to try new tools, Liao Jiade's rule of thumb is that it is important to "make everyone feel". To do this, the problems solved must be to the point and adjustments must be made in real time. . "As long as someone gives some feedback today, I will make the changes today and let you feel it tomorrow." As long as colleagues feel that their opinions are respected, they will be more willing to put them into use and give feedback, improving the usability of the system.
From the initial "amateur investment" of only Liao Jiade and an information (IT) team leader, now there are 20 people involved in the development and application of generative AI at Chi Mei Medical Center. Liao Jiade said, "The borrowing will not affect their regular work, and in my project, they will only do things related to their own departments." For example, pharmacists are responsible for studying generative AI applications that improve the work efficiency of the pharmacy department. Assist departmental innovation.
As long as someone gives some feedback, I will complete the changes today and let you feel it tomorrow, so that the imported AI application can maintain its popularity.
