The COVID-19 pandemic is running its apocalyptic course around the world. One thing is certain in the midst of total uncertainty and utter chaos: SARS-CoV-2 is a very daunting enemy and we remain totally subjugated by these viral overlords. It is easy, therefore, for we humans to lose patience and become careless as well as to decrease capacity for resilience. We need to regain our composure and to be defiant in order to for us to prevail this human vs virus struggle.
A major part of this resilience and defiance is continue our work in adoption and deployment of artificial intelligence concepts and projects in clinical medicine. Here are key takeaways from today’s AIMed Paediatrics webinar in collaboration with the international society of Pediatric Innovation (iSPI).
1. Artificial intelligence in paediatrics is much more than using deep learning for imaging interpretation and can be extended into wearable technology (edge AI), administrative tasks (robotic process automation), and decision making (recurrent neural network for time series data).
2. Promising artificial intelligence methodologies for paediatric patients include: unsupervised learning and cluster analysis, reinforcement learning, and local or embedded AI for wearable devices.
3. Natural language processing, with emergence of GPT-3 and combined with machine learning can be a powerful tool in clinical insights from unstructured data.
4. The methodologies in artificial intelligence need more robust management and access to data (intelligence needs data).
5. We need the entire pipeline as computations on continuous streams of data in real-time particularly in the intensive care and operating room setting.
6. Two data sharing models include centralized analytics vs distributed format with local analysis; there are advantages/disadvantages to either strategy.
7. EHR is only one data source as other important sources include wearable devices, social determinants of health, genomic information, etc.
8. In working with EHR, more data is not always better but data quality and selection impact greatly on the performance of the model.
9. Management of data is not usually a focused effort in healthcare systems and curation of data needs to be centralized for federated analysis.
10. Data engineering is an important aspect of artificial intelligence project and needs to take into account ability to generalize to other hospitals.
11. Paediatric institutions lack strategic IT infrastructure to share data amongst hospitals for collaborative artificial intelligence projects.
12. High potential for artificial intelligence also includes automation of tasks and insights not appreciated by clinicians via unsupervised learning.
13. Hypergraph databases can be considered a future database format for more cognitive architecture in data science.
14. Public cloud is now available for storing healthcare data as the cloud facilitates collaborative work in healthcare.
15. Recurrent neural network (RNN) LSTM model can be robust to extraneous features (less need for feature selection). RNN also offer greater flexibility and improved performance over traditional models.
16. An agile mindset is good for data science projects, and coupled with a robust clinician and data scientist collaboration, can be very productive.
17. The regulatory framework from the FDA is to promote digital health while accommodating the evolving digital health device world.
18. Software as a device (SaMD) can be assisted (ultrasound guidance) vs augmented vs autonomous (IDx-DR for diabetic retinopathy) device.
19. There is an algorithm continuum of locked algorithm with discrete updates to continuously adaptive algorithm so regulatory process has to accommodate the entire continuum.
20. Perhaps one regulatory strategy is to utilize artificial intelligence to augment the regulatory process on AI (Alan Turing’s philosophy of machine vs machine).
Thank you faculty and attendees for your knowledge and expertise as we all learned a great deal today at AIMed Paediatrics!
Anthony Chang, MD, MBA, MPH, MS
Chief Intelligence and Innovation Officer
Medical Director, The Sharon Disney Lund
Medical Intelligence and Innovation Institute (mi3)
Children’s Hospital of Orange County