The future of hospital treatment is a computer algorithm telling you how long you have left to live.
Stanford University researchers have developed a revolutionary new deep learning algorithm that uses AI to predict when a hospital patient will die – with unprecedented 90 percent accuracy.
Although the idea of predicting a person’s death may not only sound a bit grim, but actually implausible, the researchers say the AI help hospitals provide better end of life care to patients.
The idea is that by being able to accurately pinpoint when a terminally ill patient will die, caregivers and family members can prioritise the patient’s wishes and ensure things like important conversations are head with loved ones before they pass on.
According to a new study, few people who are suffering from terminal or serious illness actually live out the rest of the lives as they wished. According to the data, as many as 80 percent of people wish to spend their final days at home surrounded by their loved ones, but almost 60 percent end up dying in hospital.
In order to try and close that gap, the Stanford researchers are using artificial intelligence powered by a deep learning algorithm that can accurately identify which patients are at greatest risk from dying.
The system has analysed more than 160,000 patient records, studying information such as diagnosis, medical procedures performed, scans, treatments and medications taken.
During a test, the AI was asked to predict which of the 40,000 patients at Stanford Hospital and Lucile Packard Children’s hospital would die in the next three to twelve months, the system accurately predicted the deaths in 90 percent of cases.
“We could build a predictive model using routinely collected operational data in the healthcare setting, as opposed to a carefully designed experimental study,” Anand Avati, a member of Stanford University’s AI Lab, told IEEE Spectrum.
“The scale of data available allowed us to build an all-cause mortality prediction model, instead of being disease or demographic specific.”
However, researchers also said that the tool isn’t designed to work on its own but instead to be used as part of a wider caregiving process.
“We think that keeping a doctor in the loop and thinking of this as ‘machine learning plus the doctor’ is the way to go as opposed to blindly doing medical interventions based on algorithms… that puts us on firmer ground both ethically and safety-wise”, said Kenneth Jung, a research scientist at Stanford University.