This year’s AIMed North America took place between 13 and 15 December in Dana Point, California. The event encompassed a pre-conference workshop day on the 12th and a 3-day panel discussions thereafter. There was an estimated of 600 delegates from medical, research, technology and industrial sectors believed to be present. Over 70 speakers openly explored issues concerning the use of new technology in various facets of medicine. All sessions were also live-streamed for viewers around the globe.
In general, there was a healthy diversity and active discussions going on during the conference. Certain topics such as the pressing need for more open sources, integrating artificial intelligence (AI) assisted tools into physicians’ workflow and medical education reform were constantly brought up by speakers. While definite resolutions are awaiting to be uncovered, here are some of the highlights from the programs.
How do we make small data work?
Dr. Jack Li, professor of college of medical science and technology, Taipei Medical University, Taiwan, said data are “low-hanging fruits” for AI in medicine. If data is effectively employed, algorithms will not only overcome medical errors but also increase the quality of care and to execute exercisable disease prevention.
The problem is, both machine learning (ML) and deep learning (DL) require an enormous amount of data. As a result, it’s rather common to face an inadequacy. Dr. Pei-Ni Jone, associate professor of pediatrics-cardiology, school of medicine, University of Colorado, cited her own domain of specialty as an example. She threw an important question to the ground: “How do we make small data work?”
Most speakers agreed collaborations play a part in the solution. As Dr. Randall Wetzel, director of virtual pediatrics care unit, Children’s Hospital Los Angeles said, it’s absurd to see data as a commodity. “Competition or profit making is something we need to resist and data yearn to be free” Dr. Wetzel said.
Others like Dr. Peter Chang, director of center for AI in diagnostic medicine, University of California Irvine and chief executive officer of Avicenna.ai suggested partial trained models. This means the same AI will be passed around and taught by different groups of clinicians and scientists, who all have limited data at hands.
From analytics to something useful
On top of developing reliable models, how to ensure they fit into the workflow of clinicians was also widely debated. Dr. Kevin Seals, resident physician of diagnostic radiology, University of California Los Angeles said, most hospitals do not design their software infrastructure with “ease of integration” in mind. AI needs to be convenient and non-threatening to increase its adoption in a clinical setting.
Likewise, Dr. Tad Funahashi, chief innovation & transformation officer of Kaiser Permanente added building AI support is all about “transferring analytics to something useful” and he acknowledged there is still a gap out there.
Some enterprises had chosen to address this by showcasing what they are capable of achieving with the use of AI. Systems Oncology, an AI driven pharmaceutical company leverages on cognitive computing to produce 12 pipeline of oncology programs. Acceptto, in partnership with Intel, has come up with a smart patient room monitoring system to keep track of high-risk patients while not invading their privacy.
An effort to include patients and young audience
As mentioned, AIMed North America 2018 has an assorted blend of audience. This includes patients, their families as well as students who aspired to be in the field one day. Tom Murickan, a present high school junior, was invited to talk his founding of the AIMed club in his school. 16-year-old Caprice Bussell presented her idea of designing an augmented game for children to ease their anxiety in hospital. One of this year’s top abstract winners went to a group of four students aged between 15 and 17, as they presented a ML approach to diagnost motor vehicle trauma patients.
Dr. Anthony Chang, AIMed Founder and chief intelligence and innovation officer of Children’s Hospital at Orange Country (CHOC) invited his patients to speak about their notions of AIMed at the beginning of each day’s session.
The Lietzau Family talked about their near a decade journey with Dr. Chang, as their adopted child – Hannah suffers from severe heart condition. Jason, father of the family, expressed his view towards artificial intelligence as something “formed from those of movies”.
“Perhaps there maybe a little apprehension but patients begin to understand the benefit of diagnosis even though there are still misconceptions out there,” Jason said.
Melody, Jason’s wife continued, “As patients… I am amaze with all the technology, I want to learn and know what are the benefits for me… I think we need to take the time to educate the patients. It takes time to trust human, so we may need longer to trust non-human”.
A science writer with data background and an interest in current affair, culture and arts; a no-med from an (almost) all-med family. Follow on Twitter.