Day 2 of AIMed Pediatric 2020 virtual conference took place early on. When it comes to the discussion of the role artificial intelligence (AI) in the future of pediatric medicine, fellow speakers began sharing examples of the impact technology has on actual clinical practice and hospital management at the moment. Dr. G. Hamilton Baker, Associate Professor of Pediatric and Congenital Cardiology at Medical University of South Carolina said his institution is now equipped with a real-time location system technology.
Digital twin, system surveillance, practice supervision and knowledge democratization
Its eventual goal is to look after 200 assets within the hospital and to be able to track all the staff as well as vital equipment like ventilators. The cumulated information will then be used to create a hospital digital twin to optimize workflow and deployment of personnel. Likewise, Dr. Alberto E. Tozzi, Chief Innovation Officer and Research Area Coordinator at Bambino Gesu Children Hospital in Rome, Italy also mentioned a European Commission funded hospital digital twin project with institutions in Barcelona, Helsinki and Madrid to simplify patients’ medical journey.
On the other hand, Dr. Kathy Jenkins, Senior Associate in Cardiology at Boston Children’s Hospital and Professor of Pediatric at Harvard Medical School talked about her interest in system surveillance. She said her team is technology to pick on and predict patient advance that they don’t wish to happen. The system being used in this case will assess changes in patients and the context of what is going on around them and in the healthcare system so they are able to mitigate and influence the trajectory of the patient.
Dr. Todd Ponsky, Professor of Surgery and Director of Clinical Growth and Transformation at Cincinnati Children’s Hospital added as a pediatric surgeon, he is into the use of image recognition for hazards prevention. He explained the technology is used live in the operating rooms, acting like “eyes in the field” to supervise surgeons and guide them in using CAT scan overlay. Eventually, this will be incorporated into an image guided surgery platform.
The other area Dr. Ponsky is venturing into right now the use of AI to democratize knowledge. With 2.5 million publication debut each year, there is just too much information for human being to pay attention to and AI can be handy to find out what are some of the real breakthroughs and what to prioritize on. With that, Dr. Ponsky and team created a model using 500 abstracts and deployed it locally within the community to to predict whether an article will be given a high rating by the editorial board.
Although the model yielded a 94% negative predictive value, Dr. Ponsky believes they will be able to hit the high 90% when it is trained with more data and with crowd sourcing. At the same time, he is also focusing on dissemination of knowledge. Dr. Ponsky cited TikTok, he thought if his children are addicted to social media channels, is it also possible to take some of these features and get healthcare providers addicted to learning? His plan is to produce short and engaging pieces of bite size content and get people to pay attention on them.
Facilitate data-sharing, cross discipline collaborations and culture change
Nevertheless, Dr. Ponsky believes it is not up to the clinicians to decide how AI may change the realm of pediatric as it is likely to be in the mercy of available data. “Even if it is not the best application but if we can get the big, powerful and cleaned data, that’s probably where we are going to see the first big application of AI,” he says. While D. Baker feels the next level is for data and knowledge getting to the right person at the right time, Dr. Tozzi said perhaps the big push will be to collaborate with other disciplines.
For example, if clinicians wish to learn how to smooth the process and simplify patients’ hospital journey, they should work with someone specializing in service design, which may not be usual but it is an important to learn. Either or, Dr. Jenkins is confident AI will help clinicians better understand patients and how to care. Particularly, when the technology can process signals that are often not taken into account today to enhance the abilities to treat patients better.
With that, Dr. Ponsky highlighted the need for hospitals to understand what part do they want to play in this AI development game. Do hospitals want to be a data provider and let external companies build the models for them or do they want to be part of the designing process too? Dr. Ponsky said so because Dr. Jenkins spoke about a dilemma she faced at her institution. They would like to use a tool that connects patients and families outside of hospitals.
However, the institution is unsure if they will choose to use a home grown tool which happens to be an extension of a clinical tool being used for care coordination or if they should use a new, machine learning driven tool that does not fit nicely into the present electronic health records (EHRs). Dr. Jenkins believes it is not about choosing one over the other but many in her institutions regard EHRs integration as an important component. She would also not buy into a model developed solely by data scientists without input from clinicians.
Dr. Baker suggests leaving some flexibility. Vendor solutions can be catered for specific challenges and home-grown solutions for other challenges. Although it is important to bring different expertise to the table but if they do not understand one another, it will not create the ideal result. At the end of the day, Dr. Tozzi believes it’s a matter of instilling a change in culture. “As doctors, we should learn different things and speak different language,” he says. Dr. Ponsky added outcomes will drive us. If we can prove that AI indeed will lead to better outcomes and it is not something hypothetical, new ventures will grow continuously.