There are many conferences related to medical informatics. However, if you had to pick the flagship one, it would be the American Medical Informatics Association (AMIA). AMIA hosts three major events annually, with the Annual Symposium being the largest among them. It takes place from Saturday to Wednesday over a weekend, and due to my schedule, I attended starting from the third day.


This is a photo of me. I may look awkward, but pictures are always right.

Day One

It was my first day, but the third day of the conference. As soon as I landed at San Francisco Airport, I took a taxi straight to the site. The conference was held at the Hilton Hotel, and there was a protest nearby, advocating for the rights of hotel workers.


Handout related to the protest

The first session I attended was the AMIA Clinical Informatics Fellowship (ACIF) group meeting. Although it was intended for Clinical Informatics (CI) fellows, I attended because I am interested in the fellowship. There were other residents interested in the CI fellowship as well. It was an opportunity to learn about the current state of the CI fellowship, employment status of recent graduates, and other various aspects. One thing I noticed while researching the CI fellowship was the lack of a clear introductory page about it, and it was impressive to see ACIF trying to resolve such issues.

The most impressive part was the new ACIF website being developed.


https://www.acifellows.org/ is being created by ACIF, providing various fellowship-related information such as the process of applying to CI fellowships and the list of currently open CI programs. They say all the information will be complete by early next year, so those interested in the CI fellowship should take a look (it’s not yet mobile-friendly or completed at the moment).

After spending time socializing with the fellows, I attended a lecture on social networks. In this era where personal branding is important, it was an enjoyably delivered lecture. The highlight was the emphasis on consistency, which is easy to talk about but difficult to practice.

Then, I briefly attended an NLP lecture before checking into the hotel. The Hilton Hotel is quite expensive, so I stayed at a relatively cheaper hotel nearby.

In the evening, I joined a CI fellows’ after-party gathering to network with people. It was held at a bar called Local Edition near the hotel, with a nice ambiance and tasty cocktails. It was great to hear about various programs. The most surprising part was meeting the best friend of a colleague from my residency program; the world is indeed small.


Local Editions inside view. Source: futurebars.com

I intended to rest afterward, but through an introduction from an acquaintance, I met a teacher with similar interests attending AMIA. It was quite a spontaneous meeting, and I don’t think I would have had the opportunity without the intervention of a college colleague recently back from training in Korea. The person was interested in geriatrics and digital healthcare and had already two children despite being similar in age to me (truly impressive!). We found a lot of common ground through our conversation and engaged in a deep and meaningful discussion. The meeting took place at a restaurant named Kokkari, which served delicious food.


Food consumed atKokkari

After such a meaningful meeting, I wrapped up the first day.

Day Two

On the second morning, I attended various lectures. I was particularly interested in those related to NLP and attended a presentation by the company suki.ai and a session on the LLM model Me-LLaMA.

Suki AI focused on how AI-based medical scribes can assist with additional administrative tasks. The main topics were reducing patient burden and charting work for medical staff. Particularly in the U.S., the most time-consuming tasks for doctors include charting, sorting emails, preparing charts, writing clinical notes, preparing orders, and making referrals, and the session highlighted how AI could contribute to reducing burnout, optimizing workflow, and alleviating workforce shortages. Although these are urgent and significant issues, I felt disappointed as the presentation did not demonstrate actual solutions.

I found the Me-LLaMA lecture more interesting since I had experimented with LLaMA before. The LLaMA 13B/70B model was pre-trained and tuned to be optimized for the medical domain, using MIMIC data for training. There was such a large crowd that I had to listen from the hallway and couldn’t delve deeply into the details. You can check the PhysioNet link or Arxiv for more information.


Me-LLaMA’sPhysioNet page

During lunch, thanks to an introduction from the teacher I met the previous day, I had the opportunity to meet a professor of medical informatics from Seoul National University. Meeting with an authority in the field provided deep insights and helped sketch a path for developing my career. Although I realized I still have a long way to go, I gained motivation to further my growth. I was genuinely grateful to the teacher who arranged this opportunity.

Afterward, I took the self-driving taxi service Waymo in San Francisco to grab a meal at In-N-Out Burger. (The autonomous driving was more stable than I had feared!)

Video of riding in a Waymo taxi.

Regarding In-N-Out Burger, I thought it offered good taste for its price, a huge merit given the sky-high prices in San Francisco. The burgers didn’t feel particularly outstanding (which is hard for a fast-food chain), but considering the price, overall satisfaction was excellent. The French fries were good too.

Later, I attended a session called Wearable and Go, which spoke about the clinical use of wearable devices. It mentioned several technical aspects and advancements. There was a presentation on patient token issuance using blockchain technology and decentralized medical record management, which reminded me of discussions in Korea about Medibloc. This didn’t captivate me, and the idea of using NFTs in the medical field made me somewhat skeptical despite the strengths of the technology itself. Other lectures discussed how to enhance wearable data classification technically, how voice recognition-based self-monitoring applications (Alexa-based) performed and impacted patients, and how the prediction algorithm worked using EHR and patient monitoring data. As it is one of my main interest areas, I found the discussions highly engaging.

Later, I walked among the exhibition booths of various companies and schools at the conference. EMR-related technologies caught my eye, with companies like Cerner and Epic actively utilizing large language models (LLMs). It was notable how these technologies are being used to alleviate the hassle of sifting through numerous complex charts by improving summaries of patient records, diagnosis timings, exam histories, and periods of disease exacerbation and recurrence. Such technology seemed positively impactful since it made medical professionals’ work more convenient. Comfort is always beneficial!

After that, I traveled to Palo Alto to dine with a college senior I hadn’t seen in almost 8 years. Seeing the Palo Alto nightscape made me wish I had visited during the day too (it rained the following day, which is a secret).


Street in Palo Alto

Last Day

On the last day, I attended the session The Future of Clinical AI, featuring OpenAI and Ambience Healthcare. OpenAI explained how the learning process of the GPT model and reinforcement learning from human feedback (RLHF) improved the model. OpenAI also mentioned the need for better clinical reasoning and security for clinical applications of GPT.

Ambience was similar to the previously mentioned Suki in being an AI medical scribe, creating outpatient records from clinical conversations, summarizing necessary information before outpatient visits (pre-charting), entering disease codes during consultations, and handling prior authorizations required for prescriptions. It was a fascinating session.

The closing remarks included various awardees and Distinguished Paper presentations. However, I couldn’t cover everything due to the length of time. A link was provided to the PPT presentation, but it hasn’t been posted yet. Once the materials are available, reviewing those papers may be a good idea.


The next symposium will be in Atlanta. See you in Atlanta!

This concludes the AMIA Annual Symposium held in San Francisco. While I didn’t present personally, I was able to network with a variety of people and discover the latest trends in medical informatics. The meetings with experts from different backgrounds were incredibly valuable experiences. The fact that approximately 2,500 people attended this conference emphasized to me that my field is gaining a lot of interest and potential, which was encouraging. I will continue to strive diligently. There’s a long way to go 😂😂