During the time of Galen and his fellows, what was the primary diagnostic tool? Sense of smell. Kidneys, livers, intestines and all internal organs secrete odors during normal function. Sour, too sweet, noxious and others are all signs that things are not in balance.
If every physician would actually do a 'hands on' exam including breathing deeply of the patient, old diagnoses would be relearned. Experienced doctors with a keen social aptitude can spot a dying across a 20 acre hospital parking lot! Algorithms for the 10>10 combinations of symptoms? Madness.
1
I would never trade a death sensing cat for a death sensing computer. How horrifying.
5
I could tell my mother had begun to die by looking into my mother’s eyes because they became grey, different than her usual bright blue. She has stage 4 breast cancer and had been under hospice care for 12 months. She lasted 9 days. I stayed all day through to darkness that summer every day- surprised to see a white fluffy cat in their room. The cat did not leave her side. Later I heard the cat got evicted for biting someone.
3
My informal theme as a science teacher is What a Wonderful World.
Instead of test-prepping my students 100% of the time, I spend big chunks of every day having them explore all sorts of interesting phenomena in the natural world.
When we go back to school, I will be having the kids explore amazing animal behaviors like Oscar the cat. And because all of my students are from low-income homes, homes where animal fights are often part of the culture, I'm hoping the kids will learn to recognize and turn from animal cruelty.
Animals are amazing. Their adaptations are impressive and astounding. We might be apex predators, but we are not alone in our magnificence.
3
Everyone has heard of the "sixth sense". There are phenomena that today's science has no idea how to measure. Any scientist who admits to trying to measure them will be ostracized and lose all research funding. If that ever changes maybe we can catch up with our lowly pets.
3
It is troubling to read that 3-12 months life expectancy is the "sweet spot" of palliative care. Palliative care is not the same as hospice care. Palliative care means placing a priority on symptom management. Don't all patients deserve this, regardless of life expectancy? As for the resource utilization issue mentioned as a reason to not initiate palliative care "too early," several recent trials (including one by Temel, et al. at Massachusetts General) have shown that initiating palliative care early is associated with improved function and fewer symptoms...and, unexpectedly, improved survival. Trying to ration palliative care to only those who are dying does all of us a disservice.
4
My cat always knew when I was even a little upset, never mind really sick. Animals have amazing powers we may never come to fully appreciate. The audacity that humans would imagine creating a machine that could replicate the mysterious things that animals can do is nothing but arrogant. Can't we simply be thankful for the gifts that animals have, and leave humans out of it?
7
Hero Rats are trained to sniff TB. Perhaps they can be trained to sniff mortality.
https://www.apopo.org/en
My husband read this article and wondered who would visit Oscar and be at his side when it's his time to pass on.
11
The book, Making the Rounds with Oscar by David Dosa
is a compelling read.
2
A "dying algorithm"? I'll take the cat.
6
This 'sense' isn't limited to cats. I know a remarkable woman, born in Estonia, who emigrated to the US after WW2. In the 70s I experienced several instances of her uncanny insight - identifying the one bulimic girl in a large group she had newly met, for example - and mentioned it to her. She said that when she had first arrived in the US, she worked as an aide in a hospital ward. There she and the other staff discovered that she 'knew' well ahead of time (I don't know how long) when a patient was going to die; doctors took to asking her to come in to patients' rooms with them and tell them her perception. She finally left the hospital because she was frightening herself.
I am always skeptical of these stories - but I knew this woman, and had seen her in action. She had seen horrors as a child during the war; and later became a painter whose work was colored by these horrors. I've wondered if somehow she developed her amazingly acute vision, if that's what it was, ) during that early period.
10
Perhaps the mercy angel nurse would rub catnip on the victim patient's clothing.
Or, the cat had a strong scavenger sense of smell.
More likely it was just revenge of Schrödinger's cat.
3
Unfortunately more like Schroedinger's patient, as the patient in hospice can be considered dead and alive at the same time.
The cat's another story.
I noticed that this article came out the same day as another article praising algorithms: the one about preventing child abuse. Algorithms may be very helpful, but they should always be used by a human being who has a conscience.
14
How's the cat doing?
31
As a tolerator of four different primary cancers, including metastatic lesions to my brain, I have never asked any of my doctors to predict an expected length of survival. Some doctors may have enough experiences to make an educated prognosis, but it’s really just a burdensome guess.
At best, they might have enough statistics to tell if and when one of my cancers might return or metastasize. I’m monitored frequently for recurrence and when that happens, I hope that the recent rapid advancement of medicine will have discovered new treatment options which may keep me alive long enough for yet another advancement. But if not, eventually I’ll die. Maybe from cancer. Maybe after being hit by a truck.
No one, not our doctors nor (so far) a computer, can predict from what or when we’ll die.
(And I hope I’m never around Oscar…)
9
Something that is striking to a social worker is the incredible burden placed on doctors-often young ones- to deal with these issues without adequate support and in far, far too little time. People and families facing a life-threatening illness have times when they want to deal and times when they don't. But a hospital isn't set up to respect that. With enough time time and clinical skill it is usually possible to have the hard conversations. The poor little residents don't have what they need to achieve it. Like privacy for starters I was also shocked at the reluctance of physicians at a very famous hospital to use palliative care-NOT always the same as or hospice. "We aren't there yet" they would say of patients with highly symptomatic mets. I have also seen physicians who feel that anyone who has ever been suicidal shouldn't be allowed to use hospice. Algorithms are fine but you have to do something useful for the patient with them. Re cats of course they can sense all kinds of things about humans. But good luck training them.
Janet O'Hare, LCSW
9
I like cats, but please keep that one away from me.
8
So, what happened to the cat?
11
The algorithm sounds like a terrific idea. Too often, physicians are too "Pollyanna-ish" regarding the recovery of their patients. I worked in health care for 20 years with much of that time spent working with the elderly and very sick. I have never forgotten in my 1st year after graduating from college, the oncologist who walked in to my patient's hospital room and told his wife not to worry, that her husband would make a good recovery, despite a sudden decline in his function from the previous day. I was pretty sure that the patient was dying but could not say anything to either the wife or the doctor. After he walked out of the room, the wife started asking me what she should do, as her husband was in charge of the finances. I told her that since he was sick, it made sense that she should talk to his attorney and financial advisor to make sure that she had a better grasp of their finances. She said she would call them after she left the hospital that day. As it turned out, her husband went into a coma a few hours later and died a couple of days later.
IMO, it is good to maintain a positive outlook for both patient and family, but it is also very good to be honest about a patient's condition. How do family members, especially the spouse, prepare for their loved one's death?
22
It’s always nice to read an inspirational, heartwarming, and uplifting story, at the start of the New Year! This is not my uncle’s New York Times, anymore!!!
6
Many ancient civilizations, especially the Egyptians, knew that cats had if not special powers at least a special place. Out pets are more than loyal (or disloyal) companions. They are more than comfort for children, adults and our aged or emotionally distressed. Somehow our cats, and our dogs know something about our psychology, our emotions and our physical and mental well being. Our pets can love us or hate us or be totally indifferent. Our pets can defend us. And they can mourn our loss as well as show their joy when we return from a long separation.
When my father died in September 1971 our beagle, Amos, sensed something and comforted our youngest brother who was barely 9. He also did something. I believe that he sensed that the spirit of our father had not yet left home. I thought I saw him in the living room one night and it scared me terribly. But one late afternoon when mom was serving dinner suddenly Amos stood up, alert, and possibly excited or distressed. He began barking at something up the stairs to the bedrooms. He never did that before. There was no one in the house except mom and us 3 boys. Then we all heard a sudden slam of a door! All of us were startled. I took Amos by his collar and searched up stairs. No bedroom doors were closed. Not one. The closet doors were folding doors and could not slam shut.
I think dad was there. I also believe that Amos knew.
We don't need an algorithm. We need to understand our pets. They know something, more than us.
30
Here is my cat story: I had a cat who was very fearful of strangers, and especially of the vet. My vet came to my house to euthanize the cat whens she was old and sick. At first my cat hid under the bed, as she always did with strangers. As I stood talking with the vet, I told him I was ready to proceed. At that very moment, my cat slowly emerged, jumped up on the towel placed on the bed for her, and laid there quietly waiting for the needle. It's heart-breaking, but I have no doubt she understood what was going on. And I can't explain it.
38
The cat story was nice, but how far in advance of death did the cat start its behavior? The story sounds like it was pretty close to the time of passing, not 3 to 12 months out.
5
Cats seek warmth and dying people are often feverish as the body breaks down. That's my theory.
3
Indeed, my dying mom ran hot, but my cat stayed away from her like the plague!
There is an "old wives tale" of not letting a cat sleep in the crib with a baby because it will cause the death of the child, not necessarily from smothering.
But aside from that, cats are telepathic, mental. I call my cats by thinking about them. They soon show up at the back door. Even the neighborhood strays will answer, but only those who know me.
12
If animals know when, why do we need computers to know? Aren't we animals? Can we learn and develop this sense, or cultivate our pets to express it, with the same drive that compels our frightening faith in technology?
21
Maybe it would be possible to develop this ability if researchers knew precisely what Oscar was detecting in those terminal patients.
Mukherjee writes: "No one knows how the cat acquired his formidable death-sniffing skills. Perhaps Oscar’s nose learned to detect some unique whiff of death — chemicals released by dying cells, say. Perhaps there were other inscrutable signs."
And not all animals have this capacity -- or if others do, they haven't expressed it as Oscar did. Still, it couldn't hurt to rescue cats from shelters and place them in facilities where they can have care and company and can provide comfort to the ailing.
12
The answer is, tony, because we are stupid.
Personal electronic devices mainly serve to make us stupider and lazier.
Social media is making us tribal.
And if the computer says it's time for me to die, what if I say, "I'm not ready?"
"2001: A Space Odyssey" (1968) is still THE most brilliant explanation of the effect of technology on humans. And the Kubrick movie based on the novel is brilliant as well, far better than ANY space-themed movie that has followed it.
13
"No one knows..." Hmm. This apex predator may have been domesticated, but plenty of biologists know how and why this silent, deadly, highly sentient and perhaps even compassionate animal acquired its death-sniffing skills.
If this were a Twilight Zone episode, it would be a reverse cause and effect, the cat being the Angel of Death. But comforting, as the young Robert Redford was in that episode.
10
Call this feline superpower brainstorming: what if sickness causes distortions in our personal magnetic field that a cat can detect? After all, they can famously find their way home without a compass... like migrating birds. More interesting to me is that after detection of irregularity, why does the cat offer companionship? Some kind of empathy? Or does it trouble the cat and it is seeking comfort from the dying?
14
Cats make their way home by following the position of the sun at specific times of day, moving in the direction that matches that of their home territory.
2
With all due respect for science, why bother refining a high-tech (and high cost to create) algorithm?
Why not just train more cats and dogs like Oscar, since Oscar was apparently almost 100% right? The skills of animals at sensing things that people and the most sophisticated equipment cannot (bombs, drugs, illness) are well documented. And most of us would also prefer to be in the company of a gentle creature, like Oscar, than machines and wires when the time comes...
57
So, I will now run whenever Mr. Kitty Diggins curls up next to me. I've heard that dogs have the same senses about them, and often save people's lives.
9
Just one more reason I'm a dog person.
13
That time, the magic window, is a real gift. It's a time to make plans, to say good byes, and (most importantly) to enjoy the time you have, as much as you can. My dad used it to make photo albums with Mom - one for each of his kids. He also made a list of things to do for Mom, containing the mundane (renew life insurance in May) and the fun (get your hair done every week, travel to Arizona in winter), as well as the practical - at the bottom was the entry "Make another list".
Does it mean "give up"? - oh, heck no. Does it mean "stop taking meds"? Get real. But does it mean "take serious meds for pain"? Boy, howdy. If you are going to die in 3-12 months, why should someone care if you become addicted to a drug? Or chocolate? Or any other dang thing you need?
36
Keep this cat away from me.
15
I've been a medical oncologist for almost 30 years. I practice compassionate palliative care. Over the years , I've applied a rather simple but very effective algorithm. Families are usually comfortable with uncertainty, yet need guidance. All they need me to say is whether life expectancy can be measured in years, months, weeks, days or hours.
As humans, we need cat stories, regardless of how believable or accurate they may be.
50
Wow - Dr. Daniel - you stated everything precisely. As much as I needed to know a general ball park of how much longer I thought my mother would be around, I also needed the comfort and tenderness from our felines. Cats have always help mend my broken heart and have provided kindness and love to my hurting soul. Thank you so much for your wonderful post. Your words touched me on more levels than you can imagine. Thank you again. And thank you for the tender, gentle and kind work you do with your patients and their loved ones. I just know you are very special in your field of expertise.
5
I work (not in patient care) in a cancer center. Whether the physicians want to say it out loud in the presence of the patient or not, they are truly the cat in this story. Yes, there are outliers but for the most part deep in his heart, the physician who has seen the patient progress thru treatment, respond, not respond, and/or relapse knows when it is time to do whatever can be done for a comfortable life. My father's oncologist simply said, "You're living on the edge now and I'm not sure how long you can balance there." And we knew. Dad sat there for a few minutes and asked the physician how he felt about hospice and the doctor responded that they did great work and were very good at comfort. Dad filled out the paperwork the next day. So, this gentleman here is your cat.
19
Has anyone considered that maybe the cat was causing these patients to die?
14
Not really. Don't think the animal caused the death. Though the cat seemed to sense something peculiar about those patients. Maybe it wasn't something it smelled. It could be a sound or body temperature.
There have been other stories about domestic animals and their strong sense of smell. We routinely use dogs for sniffing out everything from people, dead and alive and substances that we with our limited, compared to dogs and cats, senses. We not only smell things less, we hear and see less than many animals.
13
Really ? What would possibly cause you to think that ?
6
I don't understand all of this paranoia from the commenters. This algorithm is another piece of information for the doctors to use when prescribing a treatment. I suppose people might have had a similar response when the thermometer was introduced into medicine: what do you need that for?! You can just press your hand to their head to see if they're hot! Why are you reducing someone's pain into a number?!
The more information, the better. If you don't want to know, tell your doctor to not tell you. I would much rather have a doctor suggest palliative care if I'm not going to make it rather than painful surgery or horrible chemo (I've had cancer, so I know.) I wouldn't want to spend the last few months of my life puking.
I don't understand why the author thinks an algorithm identifying variables is discomforting. The whole point of tools is that they give us information that is inaccessible to us by just using our senses. We can make more informed healthcare decisions as we get more information.
25
Just my anecdotal story, shared only because of my father's recent death, and which is off the main point of the article: My parent's cats (Bengals) uncharacteristically curled up with my father all day and night in the days before he passed. This was far from ordinary and one of the cats thereafter slept in the place he died. My mother had noted the cats' behavior to me the day before he died, and after his death, it struck me and I came across the story about the New England cat and then this piece today.
I am not trying to spark scientific debate (but there was no heating pad and he had been in bed for months previously). I just wanted to relate the story given the timeliness of the subject in my own life and because, in my mind, I wouldn't rule out the feline diagnosis. :)
78
Sincerest condolences on the recent passing of your father's death. I think "feline diagnoses" are similar to a human's intuition or "sixth sense" - it's something that cannot be explained or quantified, but the peculiar timing of these actions are highly unusual and often consistent. And this peculiar action by cats is not only confined to felines - I've notice dogs reacting similarly. Our next door neighbor had been recently diagnosed with pancreatic cancer. She was visiting a mutual neighbor that their dog, Jasper, came up to the ill neighbor, leaned into her leg, laid down, and stayed there until the neighbor went home. In the past, this dog would ignore the neighbor, but since her cancer was diagnosed, Jasper was always by her side. I think we can learn a great deal from animals for they have a lot to tell and teach us, we humans just need to know how to understand and interpret heir language and be turned in to their frequency . . . in a four-legged sort of way.
16
There's a key piece of information missing in this article - what fraction of patients in the sample used died within the 3 to 12 month window? If it's high enough, then a 9 out of 10 correct rate, praised as impressive by the article, could be achieved by just flipping a suitably biased coin. (In the extreme, if it's 0.9, then you can achieve that correct rate by simply declaring that every patient would die in the window). I'm sure the net did better than guessing but it would be good to know how much better. Also, this is a case where the costs of a false positive (net declares patient will die but they don't) and false negative (net declares patient will live but they don't) should be very different, and the net should be trained taking this into account. Imagine the harm done if a patient is told that The Oracle has declared that they will likely die within the year, and it's wrong. Clearly a good doctor would never put it so bluntly but people tend to place a lot of faith in technology and one can imagine it becoming a self fulfilling prophecy.
5
According to the article, the algorithm was used on 40,000 patients, and 9/10 who were predicted die within the 3 -12 month window did die within that timeframe. In addition 95% of those assigned a low probability of dying by the algorithm survived longer than 12 months.
1
Well, that fur box offered the comfort of a being with the vital spark of life and all the time in the world to devote to the dying. Anyone who's ever had a cat knows the almost hypnotic soothing of that purr--even when the purr is less a sound than a vibration.
The algorithm is for the use of the medical professionals. The cat ensured that no one within his kingdom died alone.
107
Whether it’s a cat or an algorithm, the problem Dr Mukherjee is bring to the fore is that we haven’t yet (and probably will never!) solved the mystery of life. The development of an algorithm will not make the discussions about death any easier; clinicians, patient, families and palliative care services still need to discuss whatever the algorithm reveals and engage with its implications. That’s the problem, not the fact that death is always lurking somewhere around the corner. We can take the latter for granted.
3
My mother was in a hospice hospital near Seattle where there was a resident cat named Prince. Visitor’s were told not to pick up the cat, and that when a patient was within a day of dying the cat would come to their room and lie on the bed. Why is it so hard to think some other animal might be more sentient than humans? It’s not a surprise to me and it was comforting to know the cat was ‘on the job’.
73
I've always found Dr. Mukherjee's thoughts very interesting, but this piece made me unexpectedly angry.
For two months after my mother had suffered a hemorrhagic stroke in 2015. she would cycle between 36 to 48 hour periods of lucidity and 48 to 72 hour bouts of delirium. At the beginning of each period, the doctors were fond of saying "it's like I'm seeing a completely different patient", for better or worse.
Despite the extent of her clarity in lucid periods and the high functioning she demonstrated right up until the stroke, each doctor was quick to attribute the cycles to the diagnosis of vascular dementia she had received a year earlier.
It took me two months of observation and steady pressure to get the doctors to discontinue administering hydralazine, but once done the cycles stopped and she never again entered an unresponsive state.
Her doctors were so quick to fit the patient into their preconceived expectations (especially expectations that relieved them of paying further attention to the patient when they had so many other patients to look after) that they couldn't or wouldn't consider an alternative explanation.
If those doctors could have pointed to the proceeds of an algorithm and said "your mother is going to die soon" I have little doubt that they would have.
19
The cat part sounds similar to the story of the room on an ICU ward in Tanzania where patients often seemed to die on Thursdays. A great mystery until someone solved the riddle. Turned out a cleaning crew came on Thursdays and sometimes unplugged the life support equipment to run the floor buffer.
11
Between my junior and senior year at the University of Michigan I got a job that summer at a local hospital in Ann Arbor. I worked in the salad department of the hospital kitchen. My employer began to ask me to deliver food to patients in this large hospital. It was a very interesting experience -- patients' families who tried to get me to get them drugs, a naked man shackled to a bed who seductively tried to get me to unhook the shackles, and the woman who politely asked me bring her IV hanging from its stand closer. When I did, she hurled it against the wall. But the most fascinating was when I took some uncanny notice of certain patients. I would set the food down on the tray and could not help myself from staring. Something was different about these people. Imagine my surprise when I would discover the following day that they had died. It reached a point where I began asking the nurses in these various departments what had happened to the patient in room such and such. It was a question someone from the salad department shouldn't have asked, but I did anyway. I was always told they had died overnight. It began to haunt me. I have never understood it, but I also know I saw some very, very sick people who I did not have the same reaction to. The were there day after day until they were released. But there was something "different" to me about the dying patients. I began to dread making the deliveries and was glad when I quit the job.
26
They probably should have named the cat "Schrodinger" rather than "Oscar," for he could see inside the box as to whether or not the patient was alive. The potential for such a neural network to behave similarly makes sense. The algorithm and Oscar were both in a dedicated environment where they could focus precisely on this one issue at hand. Whilst Oscar was somewhat distracted by normal cat activities as well as the daily maintenance of eating and feline hygiene, the algorithm has no such distractions and can remain singularly focused.
The caveat here is how the algorithm is used. All well and good to feed it a bunch of external data, even data of patients in one's care, perhaps making the transition toward the inevitable end an easier process. But what of the temptation to add one's own data into the mix? Would seeing what such a result would produce affect changes in behaviors to improve the odds, or would it come to be viewed as creating a probable "exit date?" Given more and better data from larger samples, the accuracy of determining said date might well improve... sounds like a potential "Black Mirror" story. Let's call it "Expiration Date."
3
Our family cats curl up with each other all the time. But we know when it's time for one of them "to go" because the others won't leave his or her side.
11
It's easy to find this creepy. Less easy to fully comprehend the suffering endured by those in reach of death and the suffering of those who care for them.
My family has experienced 4 of these scenarios, and after each we confer with one another. Would we personally want to die earlier to avoid imposing such suffering on those we love? YES! Would we have wanted the dying person to be released earlier than he or she had been? YES!
Do we believe the dying person would have wanted both of these things? YES! (confirmed by conversations with the dying person, by the way)
Now family discussions circle on how we can help one another going forward. Can we kill one another? NO. Do we live in a state that allows assisted suicide? NO.
What are humans to do?
12
"But what if an algorithm could predict death?" Really? Oh dear.
This article was difficult to read because of a strong resemblance to my mother's cancer and our family cat, Culver. He was usually as aloof as many cats can be but the summer before she was diagnosed with cancer, he never left her side. His affection, his purring, his "shadowing" her was at first sweet, tender, thoughtful. But then I had a weird hunch that perhaps something more sinister was going on with my mother's health.
Culver would always stick to any of us kids when we got sick - constantly laid on our chest or shoulder when we were bed ridden. But the moment we got better, Culver was back to being aloof and solitary.
The very last thing I want to hear from my family physician or oncology specialist is some dang algorithm formula that was devised to help prepare me when my mother was going to die. For crying out loud, a blind guy could have seen her skin tone and color change to that creepy ashen grey only days before her demise.
I'm all for high tech and scientific discoveries that could help with the prognosis and treatment of various cancers. Heck, bring 'em on! But when a person is near the very end, sometimes the mere presence and affection of an animal like a cat could be a comfort and welcomed gesture - for the patient and possibly for the family.
Since when is the success rate of estimating a patient's passing something to strive for? When did it become a contest?
50
the story is about palliative care and self and family preparation for death if it can be estimated 3-12 months before occurrence... that's a huge thing.
7
Think you let your own emotions take over and missed the point of the article. It is not about a contest nor about the last days of a dying patient, esp. one with a terminal condition. It is about how to manage the last 3-12 months of a dying person's life and likely the most important information a patient needs. Many of us have responsibilities to others such as the young plumber in the article and would want to use that remaining amount of time to take care of them as best as one can. Others may have a bucket list to complete. A more accurate prognosis can help determine what would be achievable or optimal under the circumstances.
3
Thank you for pointing out a very important aspect to my post "interested observer". I believe you are correct in that I probably did miss a key point of the article regarding the “dying algorithm” and the value of Oscar. I've experienced cancer in so many family members and close friends. Except for 2 or 3 individuals, all of the others passed away w/in a 3-24 month window. My mother and her four brothers all died of some form of cancer. Whenever I hear of another family member or close friend getting cancer, I go into autopilot and wonder how many months they will be around. The trigger for me is 99% of those I love have died from cancer, so I immediately assume that is the final outcome, whether their time on this earth are 3 months or 3 years. During much of that time, they will be suffering, in pain, hoping on some level to pass quickly. I don't personally see the benefit of an algorithm although I do now realize what you were referencing to in your post. For me it all comes down to a personal choice. I've just been "programed" to except the very worst, in a relatively short period of time, whenever a loved one is diagnosed with cancer. I saw the role of a cat with a special gift of sensing when someone is ill as a comfort. Although not sure if I would feel the same if it were I who was sick and if Oscar was paying me a visit.
Thank you for sharing your thoughts. Much appreciated. Always good to hear another point of view.
Fine, but what will you DO with the prediction??? Now that's the real question.
29
My sense from reading this article is that having a prediction that someone will (likely) die in the next 3-12 months allows that person to think about what they want from the remaining portion of their lives - conversations, bucket trips, tying up loose ends - and to plan accordingly. I wouldn't think that this would require stopping curative treatment. The days' or week's notice given by the cat isn't much time to get things done, and by then the person may be too ill to think or plan.
12
Patients are then eligible for palliative care services, which are enormously beneficial to dying patients and their families.
16
It is totally up to the patient. With that information, one can then properly plan for the remaining days.
3
In reading some of the other comments, I think this article perhaps doesn't completely illustrate the amount of unnecessary suffering that happens in the last year of life for many patients, specifically due to futile or unnecessary medical interventions. As a medical student having just completed my clinical clerkship year, I cannot begin to express how pervasive this is. Being able to appropriately harness palliative care services earlier and with more accuracy would relieve an enormous burden on patient's and their families.
108
I would rather have the cat curl up next to me when it's time, and my guess is so would the doctors. The confirmation bias of an algorithm, no matter how sophisticated, is the kiss of death to a patient that might have a fighting chance against all odds. We should be asking ourselves instead why the cat curls up next to the dying patient. The compassion of mammalian systems cannot be hard-coded into a machine. Maybe in the next millennium, but certainly not today. We need human physicians and their furry companions.
85
My sentiments exactly.
The cat may have been sensing death and giving comfort. Were the patients comforted by the cat?
In my mind, that's the most important thing.
28
Perhaps this cat also heard death as well as smelling it. The "death rattle" may begin long before human ears can hear it but at a frequency easily recognizable to the cat.
39
The physical changes that cause the death rattle could have an odor that the cat can detect.
2
I cannot help but think that the legend of the prophetic cat in that nursing home may have psychologically influenced nursing staff and patients to some very small degree so that patients saw the cat as a death curse and simply gave up and that nursing staff imperceptibly lessened their level of care. The effect may have been absolutely minuscule but it's a bit like ancient curses such as "pointing the bone" where, once the curse is cast, nothing can be done.
1
This is not the only case like this, so while I understand your thought, these people are dying after all and cats are much more acutely aware of certain things than humans are, such as smell and sound.
9
My mother was the director of an Alzheimer's care facility for many years. They had a resident cat, and she always said that the cat knew who was about to die, and would lay next to the patient on their bed. I wonder how many other people have experienced this phenomena?
55
An algorithm to determine death: a plot line for "Black Mirror" next season.
6
Fantastic article. In my lab our approach is not to implement AI through deep learning with neural networks running on computers sifting through big data - we are using chaotic circuitry that morphs at the circuit level to build the cat!
I had heard this (kind of) story before...
"When the cat would sniff the air, crane his neck and curl up next to a man or woman, it was a sure sign of impending demise."
A different result was determined. Instead of a "death sniffing cat", it was that the cat would go to the (dying) patients who had been provided heat pads since they were cold. The cat, as all cats, enjoyed they warmer bed.
Maybe this one, too, has a simpler story.
7
Oh ye of little faith.
5
Let's face it, the aged and infirm are a burden on society and take much more than they give by almost any measure, especially money. Better to get rid of them as soon as possible so we have more funds to pay benefits (and other societal costs) of illegal immigrants, and to wage war.
The nasty truth was revealed when someone examined the human decision-making process related to end of life and described those who would make such decisions as "death panels." Boy, did the politicians run for cover. No politician (or doctor) relishes making the decision to let a patient die by withholding treatment that might prolong life; so now the pols and docs want to turn decision-making over to a "dying algorithm."
Imagine this end-of-iife dialog: "Mrs. Smith, the computer tells me it's time to pull the plug on your husband." I know, the pols and docs would come up with sugar-coated language to convey this terrible message, but "pull the plug" is the essential message.
Computers do not (yet) have the true capability to make such decisions on their own. After all, it is humans who program computers and feed them the decision-making algorithms. The computers and their algorithms are merely screens for the politicians and doctors who would like to appear not to be involved in the life-and-death decisions to remove beloved, but unproductive and resource-draining, members of our society.
The "dying algorithm" is merely the latest attempt to justify eliminating the aged and infirm.
5
Does it occur to you that some people DO want to die the way they want to die and they don't want to spend every moment left trying to stay here? The whole death panel thing is trickery anyway. We have death panels we pass through throughout our lives that don't involve just docs and politicians. There is a cost to everything and if you're poor, you will likely meet your maker sooner than a rich guy. So how's that for a death panel?
18
"The latest attempt to justify eliminating the aged and infirm."
Did you even bother to READ the article? If a time frame for death can be found
the patient can receive the appropriate palliative care. I'm sure that in your bitter little world such cynical, conspiratorial analysis can be quite entertaining, but keep it out of the real world where real people are trying to solve real problems.
10
Thank you Mary Corder.
4
"Logan's Run" William F. Nolan & George Clayton Johnson 1967
Read the book or watch the movie which is different but has the same basic premise.
3
Creating an algorithm to digest medical records is like programming a computer to play chess. That's been done, of course, at enormous cost. And that's only chess. Do doctors know, or factor in, family history that makes some people resist longer than others? There are far more variables at work here.
In fact, the answer is literally under your nose. Cats are no good at math, but their noses provide them with (in this case) far better information. We know dogs can be trained to detect minute amounts of practically any chemical we wish them to, because their noses are thousands of times more discriminating than our own.
So take air samples and bring them to Oscar, or one of his feline confederates. Then analyze the protein or other compound which gets the cat to respond. Then train another cat or dog to seek it out. Frankly, I have greater faith in another living creature to make the right call, than to hope that a computer, which senses nothing, "figures it out."
19
I predict the cat will react to the air sample of a tuna fish sandwich, but that won't help much. I think you need to refine your experiment.
Beautiful.
3
Interesting article! Our bodies are complex chemical factories that for sure must distil many different odors. Diseases must produce totally different smells, which a sharp-nosed creature can detect. We know that trained dogs can smell cancer in urine with over 90% accuracy.
But why, Dr. Mukherjee, be sad that you cannot be as accurate as an algorithm? A program that has chewed varied data from 160,000 patients? Big Data is such a powerful tool. Talk about evidence-based science! The whole thing is so complex, no human could process this amount of data.
I would say that the algorithm is a fantastic tool, designed by superb professionals from many areas. We should be glad Artificial Intelligence can help human doctors. And human doctors can provide all the insights and the human compassion and communication that a computer program does not have. Seems to me a superb combination.
11
I recall a similar story, written years ago, about a cat who could sniff out death - except it turned out that the cat was attracted to the heated blankets that were placed on patients who usually were close to dying. I'd rather a cat know when my time is up than an algorithm, which sounds more like way for a hospital to save money than it does a humane approach to treatment and intervention in human end-of-life care.
21
Ha! I had posted the same story a few minutes after you! At least we will validate the stories were heard before. :)
It was an episode of "House, M.D." I'm sure most people "heard" it there.
With deep learning, the predictions will not be 100%. Will you as a doctor feel comfortable telling a patient or their loved one that they have a 94.5% probability of dying within X days? Or a 63.876% probability of dying within Y days? I can see how summoning the family to say goodbye before the end is a good thing. But to put a patient in hospice for 12 months, awaiting the z% probability, sounds cruel and inhumane.
1
I take exception to the part about 12 months of Hospice care being" cruel and inhumane". It is so entirely not. If the predictions allow patients to get into Hospice care in time to use all it has to offer to them and their families, that would be a good thing. Most people select Hospice care too late to benefit from much beyond symptom management. I was a Hospice RN for quite a few years, and have used it for family members. Of course, there are people who graduate from Hospice care (good for them!) by doing better. Certainly they were not hurt by having a multidisciplinary team to provide the best in palliative care....
18
The hospice care paled in comparison to the care in the specialty hospital.
Nothing worse than watching scores or hundreds of people waiting to die in a hospice facility. Not good for the patient. Not good for the family. And this was at a highly recommended facility.
Animals, like some algorithms can be good at spotting the obvious. I thought our dog had a strangely accurate read on our kids' illnesses. If any of our boys weren't feeling well -- complaining of a headache or sore throat -- our dog, a yellow-lab mutt, would soon start to tremble. Turns out, it had everything to do with us taking the kids' temperature. Just the sight of the oral thermometer triggered her reaction. Clearly the dog has bad memories of her own previous illness, and an uncomfortable run in with a thermometer.
3
I can hear my son screaming, "Get away from me! Everytime you touch my forehead, I get a fever!"
One thing seems pretty clear: the cat did not have access to patient records. It did not know whether a catheter was used in the spinal cord or anything else of that sort. If we are to believe that the cat could in fact sense something that was predictive of an impending demise, then someone should attempt a study based on the kind of data that the cat *could* have used, basically olfactory and tactile information about patients' bodies and their close surroundings. Why not ask patients to be allowed to place a bundle of sensors in their beds with them? The cat-sized or smaller bundle would be covered with a soft, plush material. Battery-operated, it would contain an array of temperature and touch sensors, plus either a so-called bionic nose, or at least some method of collecting air samples to be analyzed in the lab. Once a reasonable corpus of this kind of data exists, then you could do both neural network-style analysis, or traditional analysis, in order to understand the basis of the cat's predictions (and could also presumably improve accuracy by including patient record data).
8
If a cat managed to pick up on some subtle clue (and it is likely an odor) why couldn't a dog be trained to do it? They sniff out a lot of other things, and some are trained to recognize when their owner is about to have a seizure. On the other hand, if I were a potentially terminal patient and the "death dog" paid me a visit, I'm not sure how I would handle that. Anyway, I suspect the dog would outperform the computer program.
23
Had to mention dogs, didn't you?
While it is better to be more accurate in diagnosis, the biggest problems I see here have to do with the expectations of the individuals involved. Dr. Mukherjee should not have felt "shame;" regret, maybe, and certainly some sorrow at the death. Why should the doctor have been perfect? And certainly the mother's response needs to be taken as a sign of her grief.
Our quest for inhuman perfection is the problem (and way too much of this is based on money).
4
A very interesting article. Not to be negative, but surely everyone stricken with a potentially fatal condition should put their affairs in order. In fact, we are all stricken with such a condition. And the 5-year survival, even for local esophageal cancer is only 40%, no matter what the post treatment picture looks like.
7
I would have valued such an algorithm during the long declines of both of my parents, both of whom sought and endured medical interventions longer than necessary, and then were late to enter hospice. When this algorithm does become part of our lives, I hope we’ll call it “Oscar”; I’ll be ready to listen.
32
We all know death is inevitable. In many cases we know when it is getting closer. The problem is when and how do you intervene.
1
I hope someone is experimenting with cats to find which ones are more likely to be "Oscars." Probably would be as helpful as an algorithm.
5
And more entertaining.
2