Palliative Care
Palliative care is active and ongoing care provided by an interdisciplinary team, either in an institution or at home. It aims to relieve pain, alleviate physical suffering, safeguard the dignity of the sick person and support those around them.
In France in 2013, 347 253 people used hospital palliative care (61% of the people who died in France in 2013). In the year preceding death, 84% of people were hospitalized at least once and 29% received hospital palliative care. In the presence of a treated cancer, the use of HPS was 52% (lung cancer 62%, prostate cancer 41%).
In the absence of cancer, use varied according to pathology: acute stroke (24%), heart failure (17%), multiple sclerosis (23%), dementia (17%) — data taken from the Weekly Epidemiological Bulletin.- Bulletin épidémiologique hebdomadaire data
AI to alert the physician to the need for palliative care
In 2017, a team of Stanford Artificial Intelligence researchers developed a model to identify which of the hospitalized patients needed palliative care. The model developed is capable of identifying hospitalized patients who have a high risk of mortality (death predicted within 3 to 12 months). This model was trained using Electronic Health Record. If the probability of predicted death is high, medical staff is notified by e-mail.
“Our model is an 18-layers deep neural network that takes as inputs patients data and give as output : the probability of death in the next 3 to 12 months.”
Insertion of predictions in the care protocol
This algorithm has been implemented at Stanford University’s hospital. An article published in July 2020 “Hospitals tap AI to nudge clinicians toward end-of-life conversations” shows how physicians use this model in their work with end-of-life patients.
AI does not directly influence the nature or initiation of palliative care, but it can send warning signals to physicians that they would otherwise be unable to perceive.
“A lot of times, we think about it (palliative care) too late — and we think about it when patients are really, really struggling, or they need some kind of urgent intervention to turn them around,” said Wang, medicine physician at Stanford. The algorithm has made her judgement sharper, she said, because it “makes you aware of that blind spot.”
Doctors and patients have the last word
AI is used sparingly. The patient is not warned by the caregiver that Artificial Intelligence has predicted his or her near death and the doctor may challenge the model’s prediction based on his or her own judgment. At the end, it is the patients and their families who decide to start thinking about the need for palliative care.
They (doctors) also described having to decide what to do when they disagree with the algorithm — whether that’s because they think a patient is in better health than the AI does, or because they want to initiate a conversation with a patient they’re worried about, but who hasn’t been named by the model.
On the same subject …
Compassionate intelligence : Can machine learning bring more humanity to health care?
Will Machines Be Able to Tell When Patients Are About to Die?
