Skip to navigationSkip to content
Close
AI predicted a patient’s odds of dying with 90% accuracy

AI predicted a patient’s odds of dying with 90% accuracy

Read more on Axios

Contributions

  • This is where the future is going. These massive data analysis algorithms, similar to IBM’s Watson, is going to be much more integrated into the professional mainstream - probably not just medicine.

    AI algorithms like these will help medical professionals focus on addressing patient’s goals of care

    This is where the future is going. These massive data analysis algorithms, similar to IBM’s Watson, is going to be much more integrated into the professional mainstream - probably not just medicine.

    AI algorithms like these will help medical professionals focus on addressing patient’s goals of care on admission to hospital (do not resuscitate, do not intubate, focusing on comfort measures), as well as their autonomy, given they provide systematically accurate and reliable information. It will also allow doctors/nurses to focus their attentions to patients who seem stable on admission, who have a higher potential to crash later, based on information that may not be seen or understood early in the patient’s admission.

    I think we’ll be seeing a lot more integration of this technology in our global healthcare systems.

  • I wonder if, even on an unconscious level, doctors will try less hard to save someone AI has deemed unsalvageable. This would mean that the 10% of patients whose deaths are inaccurately predicted would have a lower chance of being saved.

  • This kind of thing is what current “AI” is very good at—making predictions with incomplete information.

    Particularly in medicine, diagnosis will be the first thing that AI disrupts. What that leaves is treatment decisions, involving complex analysis of a given patient’s situation. This will still need

    This kind of thing is what current “AI” is very good at—making predictions with incomplete information.

    Particularly in medicine, diagnosis will be the first thing that AI disrupts. What that leaves is treatment decisions, involving complex analysis of a given patient’s situation. This will still need to be done by people in the medium to long term.

  • This is amazing and absolutely informative. Imagine if hospitals were empowered with this data, they could have a ‘red flag’ indicator to adjust treatments or be more alert and aware of patients that potentially could die. This would improve patient care, and potentially save lives. The future of medicine is here!

  • I hope Dr’s don’t see this and get less confident in saving someone. It scares me to think of my wife needing surgery and a Dr. goes into it right after a computer tells them she’s going to die.

  • Fascinating article. Wouldn’t the AI be able to ultimately share its algorithm with medical staff? Are any of the factors treatable and being missed? Is it just probability that human predictors resist? Lots of questions.

  • Okaaay. But does it then advise the doctors of the best course of action? From the article the doctors seem mystified...

  • Cool for AI. Sucks for that patient.

  • Where does one start with this, so much to understand and discuss. How would insurance companies potentially deal with this? If the computer says you are done, does the hospital stop treating you? Do I have to give my consent for such an analysis given the consequences? Will it ever be 100% and if not

    Where does one start with this, so much to understand and discuss. How would insurance companies potentially deal with this? If the computer says you are done, does the hospital stop treating you? Do I have to give my consent for such an analysis given the consequences? Will it ever be 100% and if not then is any of it useful since you never know which 10% is wrong and we are talking about life or death? And on and on.