Scheduling our ACHT Datathon
By David H Levien
more info about hackathons and datathons
In a hackathon programmers, domain experts, project managers, and other creative team members assemble over 24-48 hours to work on a project. They work quickly, meals and other resources are provided, and they sometimes sacrifice sleep to get the job done. Hackathons are often competitions for cash prizes. There are multiple roles. In addition to participants, there are mentors, judges, and sponsors. The goal may be to create a prototype or sole a problem.
The American College of healthcare trustees is planning to hold a datathon in conjunction with our in-person conference on how to safely deploy innovations in medical devices and applications into the clinician’s armamentarium. In datathons, data scientists and clinicians get together to use real world information and applied techniques in artificial intelligence and machine learning to develop actionable intelligence. The goal is to advance along the spectrum of data, information, knowledge, and actionable intelligence. We plan to focus on the arena of critical care where rapid decisions are often required on the basis of incomplete information . Some people refer to this as type one decision making. A problem which would be worthy of study would be resolving the conflict between The fact that the ARDSNET study showed improved mortality with low title volume ventilation in a sample of patients but in an individual patient such low title volumes may cause microatelectasis and increased right to left shunting with arterial oxygen desaturation, a result which is decidedly negative for that given patient. Clearly this would be a problem worthy of deep neural networks and such artificial intelligence could possibly even be imbedded in the modern ventilator. More to come.