Stanford University AI Lab researchers recently announced the development of an artificial intelligence (AI) that “will vastly improve end-of-life care for patients and their families.”
Designed to help doctors screen newly-admitted patients who might benefit from learning about their palliative care options, the AI can can predict when a patient will die with up to 90 percent accuracy.
Stanford researchers developed algorithms using “a common error analysis technique” called, ablation analysis. The algorithm creates “deep learning, the popular machine learning technique that uses neural networks to filter and learn from huge amounts of data.”
The researchers then “train” the algorithm to identify certain health factors in order to predict a patient’s likelihood of mortality within three to 12 months.
The AI’s knowledge base, or “neural network,” is developed from the Stanford Hospital’s and Lucile Packard Children’s hospital’s Electronic Health Record data. It covers the medical records and health conditions of roughly 2 million adult and child patients.
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.
The AI would enable doctors to more quickly and accurately determine best practices and medical care for end-of-life patients. AI can help with pre-screening patients, and provide a roadmap for families making a range of decisions for seriously ill family members.
One of the researchers, Anand Avati, a PhD candidate in computer science at the AI Lab of Stanford University, told IEEE:
We could build a predictive model using routinely collected operational data in the healthcare setting, as opposed to a carefully designed experimental study. The scale of data available allowed us to build an all-cause mortality prediction model, instead of being disease or demographic specific.
In their paper explaining the AI, the researchers point to the connection between diagnosis, algorithms, and patient care. The criteria isn’t a one-size-fits-all.
The criteria for deciding which patients benefit from palliative care can be hard to state explicitly. Our approach uses deep learning to screen patients admitted to the hospital to identify those who are most likely to have palliative care needs.
The algorithm addresses a proxy problem – to predict the mortality of a given patient within the next 12 months – and use that prediction for making recommendations for palliative care referral.
Stanford’s researchers emphasized to IEEE Spectrum that patients don’t need to be dying to benefit from AI technology. Their pilot study revealed doctors and seriously ill patients still benefit from discussing end-of-life care even if the patients are not anticipated to die within the one year.
Jung clarified the AI’s purpose: “We want to make sure the sickest patients and their families get a chance to talk about what they want to happen before they become critically ill and they end up in the ICU.”