I have a lot of thoughts, and they're really long. I find this whole case really fascinating. I currently work in health care research (not in fertility), and I regularly interact with medical...
Exemplary
I have a lot of thoughts, and they're really long.
I find this whole case really fascinating. I currently work in health care research (not in fertility), and I regularly interact with medical doctors. My experience with them is that they all have their own comfort and philosophy on how to diagnose problems and how to go about doing this, but they all have great reverence for standard of care. The internet tells me that standard of care is a legal term...but in my specific segment of health care, it's also a medical term, referring to the types of tests and specific treatments that are most appropriate for different diagnoses, based on existing research and data. If I ever suggest inviting a patient to a study that a physician does not agree with, or thinks goes against standard of care and the reams of data supporting it, God help me. The physicians I work with keep up with research (and put patients on different studies), so they are constantly evolving their concepts of what standard of care is and what is likely going to be the standard of care in the near or distant future. They are not crusty hacks, they are thoughtful and careful.
I say this because I only started working in health care relatively recently, and I didn't totally understand how doctors came up with treatment plans, diagnostic tests, or labs. In my field, the doctors are trying to hew as closely to standard of care as possible, while also trying to minimize the amount of burden on the patients. This includes the number of trips into the office, the number of blood draws, the number of and types of imaging. They're not going to send you in for an MRI if an ultrasound shows you enough to make a clear diagnosis. They're also not going to order an MRI if you have a low-risk diagnosis that they don't have good reason to have concerns about, because that's a physical and financial burden on the patient. A lot of patients also get really stressed out if you start ordering them a lot of tests, so there's also an emotional burden, too.
The author of the article is American, and he says they spent oodles of money on fertility treatment. Fertility treatment is really expensive, and a lot of aspects of treatment are not covered by insurance, or you need additional insurance specifically for fertility procedures. The comments on the article are also really fascinating - there's a few radiologists that are baffled by the results. She had multiple ultrasounds. How on earth did they miss a 6 cm fibroid?! A few of them express that if nothing was coming up on these ultrasounds, it's reasonable - perhaps even standard of care - to expect that an MRI would not show anything different. If their doctor had ordered an MRI and it showed nothing, would that be considered an unnecessary, frivolous test? Would we be grumbling about the money-grubbing state of the fertility industry, which already makes money hand over fist? Was it reasonable to interpret, based on evidence and standard of care, that an MRI was not going to give additional information that was useful and financially reasonable to the patient?
In another twist, the author also leaves a comment that their doctors knew she had fibroids, and that she had surgery to remove some during this process. Her 6 cm problem fibroid also apparently was embedded in her uterine wall, so not necessarily visible from an ultrasound. (The commenting radiologists point out that a 6 cm fibroid is basically the size of a whole-ass uterus, so its specific form of presentation is a bit baffling - how did no one know it was there?!) The author also shares that if she had had a hysteroscopy, this might have also shown the fibroid clearly. I'm not a physician. I don't know what the fertility standard of care is, or when fibroids should show up on different kinds of imaging, especially if it's in the uterine wall. I know I've had imaging done of my uterus for polyps, and that they were considered a potential hindrance to successful pregnancy very early on when I was trying to get pregnant. If this couple went through six years of infertility treatment, how reasonable is it that they didn't get an MRI or a hysteroscopy?
Given my own experience in health care, I see this in a few of ways:
The doctors were trying to follow their best standard of care, and the feedback they got did not indicate than an MRI or hysteroscopy would be useful. It was frustrating to have so many failed transfers. There are quite a lot of inexplicable cases where the doctors do not understand why pregnancy doesn't happen when it should. We do not fully understand the process of pregnancy - or why it ends unexpectedly. (Ask me how I know.) This is a case where a moonshot idea from an LLM, operating outside the standard of care, happened to have the right answer. It's not super likely that this will answer every person's problem, but it worked really well here.
The doctors missed some critical information that should have indicated an MRI or hysteroscopy. (Especially a hysteroscopy.) These are common techniques that come up in fertility treatments and diagnosis, and one of these doctors should have recognized that one of these techniques could have helped rule out potential issues, especially since the patient had a history of fibroids. The doctors let their rigid thinking and training get in the way of being flexible, and missed a diagnosis that they should have made. LLMs should be consulted regularly to test for obvious gaps in thinking on the part of busy doctors.
Women's health care, and in particular fertility treatments, are poorly researched and poorly understood. There is little effort to fund or support advances in the field, and couples who are diagnosed with infertility need to expend unreasonable amounts of time, energy, and money to attempt to treat, let alone successfully treat their infertility. Any edge to try and source solutions for these people should be supported and lauded.
This couple had seemingly endless amounts of time and money to continuously ask an LLM why they were infertile. After literal years of prompting, speculating, and asking "what else?", they finally stumbled on a random answer that happened to be right. It's entirely dumb luck, and with enough time and attempts at solving the question they figured it out. This is not expedient or sensible for the vast majority of people.
I feel like all of them are true. But to me, it's a great general question of "what is the solution here?" Is this a model of what we should do going forward, and how health care needs to change? Or is it just really lucky? The physicians I know work very hard to try and treat each patient to the best of their ability; when treatments don't work, they get frustrated. When patients decline standard of care, if the physician is concerned that the patient is potentially leading themselves to major harm, they will have conversations, sometimes over the course of years, to try and change their minds so that they can do their best to keep them healthy. It often works. The physicians I know are not money-grubbing assholes, they care very deeply about their patients and are invested in giving them the best care possible. I don't know the situation with the author's doctors, but I personally would not be comfortable at all with an LLM superseding the process that the physicians use to diagnose and treat their patients at my clinic. They will absolutely respect when a patient rejects their proposed treatment, and they will consider when patients want a specific kind of test or treatment, but they will also tell them when taking ivermectin is not standard of care and they're not going to prescribe it.
TLDR: I don't think this is a straightforward case at all. It brings up a lot of questions for me about how doctors make diagnoses, how they do right by their patients, and when it's ethically appropriate to just do what the patient wants.
As someone who has been through the IVF process (and have a healthy 3yo) as well, it's absolutely insane to me they never got a hysteroscopy, at least with us, even without a history of cysts or...
As someone who has been through the IVF process (and have a healthy 3yo) as well, it's absolutely insane to me they never got a hysteroscopy, at least with us, even without a history of cysts or fibroids or similar, our fertility doctor wanted to make sure that the uterine wall was ideal prior to transfer. I think all in all we had two hysteroscopies done, one before our first transfer, and one as well as some removal of scar tissue years after in preparation for another transfer, I cannot believe they wouldn't have done one in 6 years of fertility treatment when it was one of the first things they looked towards in idk, maybe the first 6 mo of the process?
Incidentally, it was after the second hysteroscopy, that whoops! my wife got pregnant naturally. Turns out she mostly had "sticky" fallopian tubes. So our son(3mo now) jumped queue.
All in all I'm happy this couple got the child they wanted. Fertility struggles are really, really really really stressful and costly and generally a lot of sadness and disappointment, I'd have been gutted if we'd done 5 transfers without a live birth....we didn't even have 5 viable embryos.
That's awesome that things worked out for you! I've had my own adventures trying to have a baby, it can be absolutely miserable. On one hand, I'm so happy for anyone that's able to figure out the...
That's awesome that things worked out for you! I've had my own adventures trying to have a baby, it can be absolutely miserable. On one hand, I'm so happy for anyone that's able to figure out the root cause of why their bodies don't work as intended. I have a lot of very mixed feelings on American health care, and I was very, very fortunate to have great insurance that allowed me to diagnose my specific problems.
On the other hand, I read a story like this and I feel really perplexed. Why did this outcome happen this way? I definitely had hysteroscopies (with printout pictures...), plural. I would have guessed, especially with a history of fibroids, that the patient in the story should have had a hysteroscopy. Wouldn't they have had to do one to remove the fibroids?! I feel a little like perhaps the author is not 100% accurate, either from a lack of familiarity with medical terminology or perhaps willingly to make a better story, but the logic of the procedures done over six years just sounds very odd to me. Is part of this because treatment plans six+ years ago were bad, and we've learned a lot? Was their doctor a quack? I don't have a good answer, but at least one person in this saga feels like they made some kind of monster error. It bothers me, but I'm not sure exactly whom to be most bothered with.
Definitely an interesting article! Using AI for medical questions is risky, especially since many people use it instead of a doctor. This time worked out well since they were actually smart about...
Definitely an interesting article! Using AI for medical questions is risky, especially since many people use it instead of a doctor. This time worked out well since they were actually smart about how they used it to resolve questions, and also fact-checked with actual doctors.
That said: there are so many things that can go wrong with our bodies that it's natural for human doctors to miss signs. Our memories are only so good after all, and as we refine our knowledge in any field, we can develop tunnel vision. I've heard of multiple cases of diagnoses because medical students or recent graduates recognizing symptoms of diseases which more experienced doctors subconsciously dismissed due to factors like age. (Actually, I think that exact scenario happened with my dad's heart attack at 17. Because a 17-year-old high school star athlete is NOT the expected demographic for heart attacks.)
To that end, if used correctly (and with training on the proper knowledge base), AI can be useful for overcoming that. AI won't have the natural subconscious biases people develop over time, and it can also "recall" a lot more information and potential causes. At the very least, it can be very helpful for doctors to find new directions to investigate.
Or, as in this case, it may suggest some simple test everyone overlooked that can solve everything.
Also, just putting this in spoilers because I have to comment on it:
Yeah, I think the author has a point about "fibroids" being a way to make tumors sound less scary. I've heard of fibroids in passing for years, but never actually knew what they were until now.
Which is a type of tumor. She had a tumor. A whole tumor. Non-cancerous, but still a tumor. Undetected for six years.
Yep, "tumor" definitely sounds a lot more serious than "fibroid". And obviously it is, because it taking up space was apparently the sole thing blocking her from getting pregnant. It may not have been life-or-death, but their lives would have been VERY different if they'd done an MRI and found the fibroid earlier.
I am high risk for breast cancer due to radiation treatment for lymphoma, and talked with my oncologist and gynecologist about regular breast MRI. Neither one recommended it because the contrast...
maybe we should just have MRIs as part of standard reproductive-related health exams...?
I am high risk for breast cancer due to radiation treatment for lymphoma, and talked with my oncologist and gynecologist about regular breast MRI. Neither one recommended it because the contrast agent they inject to get clear MRI pictures accumulates in fatty tissues, including the brain. Occasional use seems fine, but recurring use seems likely to be risky.
Yes. I work in health care currently and there's a lot of reasons not to get an MRI (aside from the high cost): allergies to the contrast agent, limited kidney function that can have difficulty...
Yes. I work in health care currently and there's a lot of reasons not to get an MRI (aside from the high cost): allergies to the contrast agent, limited kidney function that can have difficulty clearing the contrast agent, metallic implants or other medical devices, and claustrophobia are all reasons why an MRI are not indicated for patients that I've seen. Some patients also may have scoliosis or other physical issues that don't allow them to lie still - an MRI takes awhile, and it's not a trivial procedure, especially for older patients. Just this week, I saw a doctor decline an MRI for an older patient who was on the frail side - it's not a panacea. There's a reason why there are multiple imaging options.
That's the balance, of cost vs benefit not just financially (though unfortunately can matter too) but to your body. If it's looking at a potential spinal injury that's very different than say,...
That's the balance, of cost vs benefit not just financially (though unfortunately can matter too) but to your body. If it's looking at a potential spinal injury that's very different than say, early fertility issues which may be treatable without it.
/Obligatory distaste for being linked to Alexander's blog regardless of the author.
Fair enough, I haven't gotten an MRI so wasn't aware of that. I figured it would be fairly unreasonable to have one every time you went to a doctor, I was thinking maybe something like 5 years....
Fair enough, I haven't gotten an MRI so wasn't aware of that. I figured it would be fairly unreasonable to have one every time you went to a doctor, I was thinking maybe something like 5 years.
Still, it would be useful to have them as a standard part of the investigation process should other tests be inconclusive. It's a fairly standard and noninvasive test that can be incredibly informative. Yet the writer described facing resistance from their doctor to even getting one, which is just wild to me.
The problem isn’t only cost or risk, but also the chance of false positives. If you test everyone, then you’re very likely to get lots of false positives and submit these healthy people to...
The problem isn’t only cost or risk, but also the chance of false positives. If you test everyone, then you’re very likely to get lots of false positives and submit these healthy people to potentially dangerous procedures. If you only test those with symptoms or are otherwise high-risk, you lower the number of false positives and harm fewer people while still treating the vast majority
Although as an interesting tangent for this specific topic, a lot more negative scans, especially ones subtly close enough to positive scans that humans might misjudge them, is exactly what we...
Although as an interesting tangent for this specific topic, a lot more negative scans, especially ones subtly close enough to positive scans that humans might misjudge them, is exactly what we need to feed into an image classifier to negate that problem. It’s obviously not ethical to scan people for that reason alone if it’d require more invasive follow ups to identify positive/negative definitively, but it is worth keeping in mind as an actively changing level of risk at the individual and population level right now.
Depending on how good the stats are right now (my knowledge on this specific subtopic is a couple of years stale at this point), the classifiers may already be good enough from existing data to reduce the false positive rate to an acceptable level and justify, say, 20% more scanning of uncertain cases than would be done now because the signal to noise ratio has already shifted in the right direction and the individual risk of missing a diagnosis is now higher than the risk of a false positive. Thus giving more scans to analyse, thus giving better classifiers, thus reducing the risk of doing scans!
If you get an MRI and you have something the doctors want to monitor over time, you typically want to have them at short enough intervals that you can see how your imaging changes over time....
If you get an MRI and you have something the doctors want to monitor over time, you typically want to have them at short enough intervals that you can see how your imaging changes over time. They'll usually prescribe MRIs 6 or 12 months apart so you can monitor changes in the body. 5 years is a really long time in imaging time.
This isn't a triumph of ChatGPT, it's a flaw in the way medicine is practiced. It only makes sense to take imaging of the uterus, but doctors and insurance/healthcare systems hate performing...
This isn't a triumph of ChatGPT, it's a flaw in the way medicine is practiced. It only makes sense to take imaging of the uterus, but doctors and insurance/healthcare systems hate performing certain tests because a bunch of stupid nonsense that I'm not going to detail here.
They never should have been made to feel so desperate that they turned to a piece of software. They never should have been so ignored that the software was what prompted them to push for getting a test done :c
Why not both? If the police fail to catch a criminal and the situation gets resolved by a courageous private citizen who is there at the time, it is both a failure of the police and a triumph of...
This isn't a triumph of ChatGPT, it's a flaw in the way medicine is practiced.
Why not both? If the police fail to catch a criminal and the situation gets resolved by a courageous private citizen who is there at the time, it is both a failure of the police and a triumph of the private citizen.
Per the account without ChatGPT there was no way they’d even have an inkling to do another MRI.
Wholeheartedly agree. But. There are known and widespread flaws in how medicine is practiced relating to women’s health. There are known and widespread flaws in how medicine is practiced in...
Wholeheartedly agree. But. There are known and widespread flaws in how medicine is practiced relating to women’s health. There are known and widespread flaws in how medicine is practiced in general. And in how academic research is disseminated, and in how software is documented and tested, and in how business information is managed, and in how organisational knowledge is used, and in how…
And I really do get you, I’ve been the one hand wringing at obvious failures for most of my life because they bother me. But that’s the thing, I’ve been doing it for all those decades and these problems aren’t getting fixed because the fix is often “if people would just…”, and people generally don’t collectively “just <whatever>” even when the outcome would benefit them and everyone else.
So if we can use the machine to quickly and broadly patch human process gaps that’ve been embedding themselves for decades or centuries, I see it as a massive win. Perhaps not to the level the AI execs are promising, because I’m increasingly convinced most of them have lost their goddamn minds, but certainly to a level that’ll bring world-changing positives (will they outweigh the world-changing negatives those execs are pushing along the same path? Unclear, if I’m honest).
So yeah, it shouldn’t have happened, but if we’ve got an individual-level fix here and now and it can change things when all the plausible systematic fixes are just being ignored that’s still pretty exciting to me.
I'm not sure what particular reasons you're referring to as to why doctors don't like to perform certain tests, but a large reason why doctors don't like to do certain tests has to do with false...
I'm not sure what particular reasons you're referring to as to why doctors don't like to perform certain tests, but a large reason why doctors don't like to do certain tests has to do with false positives. If the false positive rate for a test is higher than the prevalence of that condition in the general population (which it very often is), then you will do harm more than good by doing the test. The doctor has to have reason to believe you exist within a certain subpopulation whose prevalence for the condition is higher than the false positive rate for it to be worth it. The doctor also has a responsibility to not clog up the healthcare system with unnecessary testing.
This is also compounded by a mix of narratives like cuts that are "targeting waste and these doctors who order unnecessary tests" and "patients are going to doctor Google/AI and acting like they...
This is also compounded by a mix of narratives like cuts that are "targeting waste and these doctors who order unnecessary tests" and "patients are going to doctor Google/AI and acting like they have medical degrees. So politics and the current doctor experience (and burnt out doctors post COVID too) are all factors as well.
No doctor is perfect, but neither is their education and experience worthless.
In an ideal world where the health insurance industry was ruthlessly competitive in favor of consumers, the insurance companies could actually play a substantial role in managing scarcity in the...
In an ideal world where the health insurance industry was ruthlessly competitive in favor of consumers, the insurance companies could actually play a substantial role in managing scarcity in the healthcare system. Unfortunately, we live very far from that world and there's all sorts of hogwash narratives pushed in healthcare under the guise of "reducing waste" and "cutting costs for patients" when it's really to juice profits.
They mention in the article laparascopy surgery of the abdomen visualizing the outside of the uterus, hysteroscopy, and ultrasound of the uterus. The author doesn't go into very much detail about...
They mention in the article laparascopy surgery of the abdomen visualizing the outside of the uterus, hysteroscopy, and ultrasound of the uterus. The author doesn't go into very much detail about the rest of their 6-year treatment, but it seems likely other imaging of the uterus was also done. It wasn't a failure to do any imaging, it was a failure to do yet another imaging after many imagings that hadn't found anything. Or failure to pick up on the potentially suspicious spot on the ultrasound that the AI noticed and based its recommendation of an MRI follow up on.
If you are unsuccessful for long enough on your pregnancy journey, and none of the many doctors you see can identify a problem, you get branded with a label. You now have “unexplained infertility.” This is good because there’s still hope of fixing the problem but dreaded because once you’ve reached this point the system is less interested in you. The fertility doctors don’t become unkind, but they also don’t give you the sense they are poring over medical journals trying to figure out your problem. They have looked for the keys under the streetlight and come up empty. Once you have unexplained infertility, the doctors start saying things like, “Have you considered using a surrogate?” and, “Sure, the last 5 embryo transfers failed, but what if we try a 6th and cross our fingers really hard?”
My wife and I could make healthy embryos, but none of the transfers were sticking. It was so frustrating to be able to create life only to have it trapped forever in a cooler at a hospital in Milwaukee, our babies like so many forgotten Miller Lights.
This is me right now. I had relatively major surgery last week, and the entirety of my follow up is a phone call scheduled in 6 weeks and a four page pamphlet from the hospital (with my particular...
This is me right now. I had relatively major surgery last week, and the entirety of my follow up is a phone call scheduled in 6 weeks and a four page pamphlet from the hospital (with my particular surgery circled and a note to ignore the advice for some related surgeries the pamphlet also discusses). In the meantime, I have about a hundred queries to LLMs about what’s happening and what I should be doing to recover.
If anyone is interested here author comment why fibroid was not detected using ultrasound:
If anyone is interested here author comment why fibroid was not detected using ultrasound:
Author here -- yeah, it's wild right? Here are a few more details for the interested. The problem fibroid was intramural with a submucosal component (mostly in the muscle wall), so I guess those are hard to see on ultrasound? Pathology report says 6 cm, I double checked. I'm tempted to share a pic, that thing was big and gnarly looking.
She didn’t get a hysteroscopy until after the MRI, doing that sooner would have been a good idea but it was never presented as an option.
She had other fibroids that were seen on ultrasound and removed during a surgery for endometriosis, so it's not like we didn't know fibroids could be an issue.
I'm glad that the family in the article had the great outcome they did, but I wonder what the ratio between outcomes like this, and the mirror opposite outcome - - extra unnecessary tests and...
I'm glad that the family in the article had the great outcome they did, but I wonder what the ratio between outcomes like this, and the mirror opposite outcome - - extra unnecessary tests and health anxiety over a routine illness.
Perspective: information overload has always been a problem, now made worse by speed and availability. But as always, humans have to decide for themselves when a source of too much info is harmful...
Perspective: information overload has always been a problem, now made worse by speed and availability. But as always, humans have to decide for themselves when a source of too much info is harmful to them.
When I was expecting, I got a book called What To Expect When One Is Expecting -- my midwives (human ones, excellent all) were mildly scandalized and suggested I get rid of it. The book was already considered quite outdated by then, and it contains pretty much every possible thing that could possibly go wrong no matter how tiny of a possibility. My midwives has seen many other expectant mothers worry themselves thin pouring over the pages. Meanwhile, I read the information happily, and with the knowledge of how much could go wrong and hasn't, and how minuscule the chances are, I found it comforting.
So, my suggestion is that LLM will be a godsend for those who can benefit from them, but it will be awful for those who can't handle it. Like a book that became standard and then considered out dated and ignored, I hope the nervous types will also heed the words of others and know to avoid them.
I suppose I was thinking from the perspective of something like the pan-scans that are famous for finding "incedentalomas" that turn out to be nothing, so this was thought provoking, thank you.
I suppose I was thinking from the perspective of something like the pan-scans that are famous for finding "incedentalomas" that turn out to be nothing, so this was thought provoking, thank you.
There are certainly tradeoffs, but I’m hopeful that it isn’t always a necessary tradeoff. That is, it should be possible to somehow combine AI-based research and professional advice to get the...
There are certainly tradeoffs, but I’m hopeful that it isn’t always a necessary tradeoff. That is, it should be possible to somehow combine AI-based research and professional advice to get the best of both.
I have a lot of thoughts, and they're really long.
I find this whole case really fascinating. I currently work in health care research (not in fertility), and I regularly interact with medical doctors. My experience with them is that they all have their own comfort and philosophy on how to diagnose problems and how to go about doing this, but they all have great reverence for standard of care. The internet tells me that standard of care is a legal term...but in my specific segment of health care, it's also a medical term, referring to the types of tests and specific treatments that are most appropriate for different diagnoses, based on existing research and data. If I ever suggest inviting a patient to a study that a physician does not agree with, or thinks goes against standard of care and the reams of data supporting it, God help me. The physicians I work with keep up with research (and put patients on different studies), so they are constantly evolving their concepts of what standard of care is and what is likely going to be the standard of care in the near or distant future. They are not crusty hacks, they are thoughtful and careful.
I say this because I only started working in health care relatively recently, and I didn't totally understand how doctors came up with treatment plans, diagnostic tests, or labs. In my field, the doctors are trying to hew as closely to standard of care as possible, while also trying to minimize the amount of burden on the patients. This includes the number of trips into the office, the number of blood draws, the number of and types of imaging. They're not going to send you in for an MRI if an ultrasound shows you enough to make a clear diagnosis. They're also not going to order an MRI if you have a low-risk diagnosis that they don't have good reason to have concerns about, because that's a physical and financial burden on the patient. A lot of patients also get really stressed out if you start ordering them a lot of tests, so there's also an emotional burden, too.
The author of the article is American, and he says they spent oodles of money on fertility treatment. Fertility treatment is really expensive, and a lot of aspects of treatment are not covered by insurance, or you need additional insurance specifically for fertility procedures. The comments on the article are also really fascinating - there's a few radiologists that are baffled by the results. She had multiple ultrasounds. How on earth did they miss a 6 cm fibroid?! A few of them express that if nothing was coming up on these ultrasounds, it's reasonable - perhaps even standard of care - to expect that an MRI would not show anything different. If their doctor had ordered an MRI and it showed nothing, would that be considered an unnecessary, frivolous test? Would we be grumbling about the money-grubbing state of the fertility industry, which already makes money hand over fist? Was it reasonable to interpret, based on evidence and standard of care, that an MRI was not going to give additional information that was useful and financially reasonable to the patient?
In another twist, the author also leaves a comment that their doctors knew she had fibroids, and that she had surgery to remove some during this process. Her 6 cm problem fibroid also apparently was embedded in her uterine wall, so not necessarily visible from an ultrasound. (The commenting radiologists point out that a 6 cm fibroid is basically the size of a whole-ass uterus, so its specific form of presentation is a bit baffling - how did no one know it was there?!) The author also shares that if she had had a hysteroscopy, this might have also shown the fibroid clearly. I'm not a physician. I don't know what the fertility standard of care is, or when fibroids should show up on different kinds of imaging, especially if it's in the uterine wall. I know I've had imaging done of my uterus for polyps, and that they were considered a potential hindrance to successful pregnancy very early on when I was trying to get pregnant. If this couple went through six years of infertility treatment, how reasonable is it that they didn't get an MRI or a hysteroscopy?
Given my own experience in health care, I see this in a few of ways:
The doctors were trying to follow their best standard of care, and the feedback they got did not indicate than an MRI or hysteroscopy would be useful. It was frustrating to have so many failed transfers. There are quite a lot of inexplicable cases where the doctors do not understand why pregnancy doesn't happen when it should. We do not fully understand the process of pregnancy - or why it ends unexpectedly. (Ask me how I know.) This is a case where a moonshot idea from an LLM, operating outside the standard of care, happened to have the right answer. It's not super likely that this will answer every person's problem, but it worked really well here.
The doctors missed some critical information that should have indicated an MRI or hysteroscopy. (Especially a hysteroscopy.) These are common techniques that come up in fertility treatments and diagnosis, and one of these doctors should have recognized that one of these techniques could have helped rule out potential issues, especially since the patient had a history of fibroids. The doctors let their rigid thinking and training get in the way of being flexible, and missed a diagnosis that they should have made. LLMs should be consulted regularly to test for obvious gaps in thinking on the part of busy doctors.
Women's health care, and in particular fertility treatments, are poorly researched and poorly understood. There is little effort to fund or support advances in the field, and couples who are diagnosed with infertility need to expend unreasonable amounts of time, energy, and money to attempt to treat, let alone successfully treat their infertility. Any edge to try and source solutions for these people should be supported and lauded.
This couple had seemingly endless amounts of time and money to continuously ask an LLM why they were infertile. After literal years of prompting, speculating, and asking "what else?", they finally stumbled on a random answer that happened to be right. It's entirely dumb luck, and with enough time and attempts at solving the question they figured it out. This is not expedient or sensible for the vast majority of people.
I feel like all of them are true. But to me, it's a great general question of "what is the solution here?" Is this a model of what we should do going forward, and how health care needs to change? Or is it just really lucky? The physicians I know work very hard to try and treat each patient to the best of their ability; when treatments don't work, they get frustrated. When patients decline standard of care, if the physician is concerned that the patient is potentially leading themselves to major harm, they will have conversations, sometimes over the course of years, to try and change their minds so that they can do their best to keep them healthy. It often works. The physicians I know are not money-grubbing assholes, they care very deeply about their patients and are invested in giving them the best care possible. I don't know the situation with the author's doctors, but I personally would not be comfortable at all with an LLM superseding the process that the physicians use to diagnose and treat their patients at my clinic. They will absolutely respect when a patient rejects their proposed treatment, and they will consider when patients want a specific kind of test or treatment, but they will also tell them when taking ivermectin is not standard of care and they're not going to prescribe it.
TLDR: I don't think this is a straightforward case at all. It brings up a lot of questions for me about how doctors make diagnoses, how they do right by their patients, and when it's ethically appropriate to just do what the patient wants.
As someone who has been through the IVF process (and have a healthy 3yo) as well, it's absolutely insane to me they never got a hysteroscopy, at least with us, even without a history of cysts or fibroids or similar, our fertility doctor wanted to make sure that the uterine wall was ideal prior to transfer. I think all in all we had two hysteroscopies done, one before our first transfer, and one as well as some removal of scar tissue years after in preparation for another transfer, I cannot believe they wouldn't have done one in 6 years of fertility treatment when it was one of the first things they looked towards in idk, maybe the first 6 mo of the process?
Incidentally, it was after the second hysteroscopy, that whoops! my wife got pregnant naturally. Turns out she mostly had "sticky" fallopian tubes. So our son(3mo now) jumped queue.
All in all I'm happy this couple got the child they wanted. Fertility struggles are really, really really really stressful and costly and generally a lot of sadness and disappointment, I'd have been gutted if we'd done 5 transfers without a live birth....we didn't even have 5 viable embryos.
That's awesome that things worked out for you! I've had my own adventures trying to have a baby, it can be absolutely miserable. On one hand, I'm so happy for anyone that's able to figure out the root cause of why their bodies don't work as intended. I have a lot of very mixed feelings on American health care, and I was very, very fortunate to have great insurance that allowed me to diagnose my specific problems.
On the other hand, I read a story like this and I feel really perplexed. Why did this outcome happen this way? I definitely had hysteroscopies (with printout pictures...), plural. I would have guessed, especially with a history of fibroids, that the patient in the story should have had a hysteroscopy. Wouldn't they have had to do one to remove the fibroids?! I feel a little like perhaps the author is not 100% accurate, either from a lack of familiarity with medical terminology or perhaps willingly to make a better story, but the logic of the procedures done over six years just sounds very odd to me. Is part of this because treatment plans six+ years ago were bad, and we've learned a lot? Was their doctor a quack? I don't have a good answer, but at least one person in this saga feels like they made some kind of monster error. It bothers me, but I'm not sure exactly whom to be most bothered with.
Definitely an interesting article! Using AI for medical questions is risky, especially since many people use it instead of a doctor. This time worked out well since they were actually smart about how they used it to resolve questions, and also fact-checked with actual doctors.
That said: there are so many things that can go wrong with our bodies that it's natural for human doctors to miss signs. Our memories are only so good after all, and as we refine our knowledge in any field, we can develop tunnel vision. I've heard of multiple cases of diagnoses because medical students or recent graduates recognizing symptoms of diseases which more experienced doctors subconsciously dismissed due to factors like age. (Actually, I think that exact scenario happened with my dad's heart attack at 17. Because a 17-year-old high school star athlete is NOT the expected demographic for heart attacks.)
To that end, if used correctly (and with training on the proper knowledge base), AI can be useful for overcoming that. AI won't have the natural subconscious biases people develop over time, and it can also "recall" a lot more information and potential causes. At the very least, it can be very helpful for doctors to find new directions to investigate.
Or, as in this case, it may suggest some simple test everyone overlooked that can solve everything.
Also, just putting this in spoilers because I have to comment on it:
Yeah, I think the author has a point about "fibroids" being a way to make tumors sound less scary. I've heard of fibroids in passing for years, but never actually knew what they were until now.
Which is a type of tumor. She had a tumor. A whole tumor. Non-cancerous, but still a tumor. Undetected for six years.
Yep, "tumor" definitely sounds a lot more serious than "fibroid". And obviously it is, because it taking up space was apparently the sole thing blocking her from getting pregnant. It may not have been life-or-death, but their lives would have been VERY different if they'd done an MRI and found the fibroid earlier.
Actually, maybe we should just have MRIs as part of standard reproductive-related health exams...? Because there's a lot of potential physical issues like fibroids that wouldn't necessarily have other symptoms, so could only be discovered via directly checking. Apparently MRIs and ultrasounds can also potentially be used to diagnose endometriosis, or at least the possibility of it or other issues.
Women's health is just critically under-researched and minimalized.
I am high risk for breast cancer due to radiation treatment for lymphoma, and talked with my oncologist and gynecologist about regular breast MRI. Neither one recommended it because the contrast agent they inject to get clear MRI pictures accumulates in fatty tissues, including the brain. Occasional use seems fine, but recurring use seems likely to be risky.
Yes. I work in health care currently and there's a lot of reasons not to get an MRI (aside from the high cost): allergies to the contrast agent, limited kidney function that can have difficulty clearing the contrast agent, metallic implants or other medical devices, and claustrophobia are all reasons why an MRI are not indicated for patients that I've seen. Some patients also may have scoliosis or other physical issues that don't allow them to lie still - an MRI takes awhile, and it's not a trivial procedure, especially for older patients. Just this week, I saw a doctor decline an MRI for an older patient who was on the frail side - it's not a panacea. There's a reason why there are multiple imaging options.
That's the balance, of cost vs benefit not just financially (though unfortunately can matter too) but to your body. If it's looking at a potential spinal injury that's very different than say, early fertility issues which may be treatable without it.
/Obligatory distaste for being linked to Alexander's blog regardless of the author.
Fair enough, I haven't gotten an MRI so wasn't aware of that. I figured it would be fairly unreasonable to have one every time you went to a doctor, I was thinking maybe something like 5 years.
Still, it would be useful to have them as a standard part of the investigation process should other tests be inconclusive. It's a fairly standard and noninvasive test that can be incredibly informative. Yet the writer described facing resistance from their doctor to even getting one, which is just wild to me.
The problem isn’t only cost or risk, but also the chance of false positives. If you test everyone, then you’re very likely to get lots of false positives and submit these healthy people to potentially dangerous procedures. If you only test those with symptoms or are otherwise high-risk, you lower the number of false positives and harm fewer people while still treating the vast majority
Although as an interesting tangent for this specific topic, a lot more negative scans, especially ones subtly close enough to positive scans that humans might misjudge them, is exactly what we need to feed into an image classifier to negate that problem. It’s obviously not ethical to scan people for that reason alone if it’d require more invasive follow ups to identify positive/negative definitively, but it is worth keeping in mind as an actively changing level of risk at the individual and population level right now.
Depending on how good the stats are right now (my knowledge on this specific subtopic is a couple of years stale at this point), the classifiers may already be good enough from existing data to reduce the false positive rate to an acceptable level and justify, say, 20% more scanning of uncertain cases than would be done now because the signal to noise ratio has already shifted in the right direction and the individual risk of missing a diagnosis is now higher than the risk of a false positive. Thus giving more scans to analyse, thus giving better classifiers, thus reducing the risk of doing scans!
If you get an MRI and you have something the doctors want to monitor over time, you typically want to have them at short enough intervals that you can see how your imaging changes over time. They'll usually prescribe MRIs 6 or 12 months apart so you can monitor changes in the body. 5 years is a really long time in imaging time.
This isn't a triumph of ChatGPT, it's a flaw in the way medicine is practiced. It only makes sense to take imaging of the uterus, but doctors and insurance/healthcare systems hate performing certain tests because a bunch of stupid nonsense that I'm not going to detail here.
They never should have been made to feel so desperate that they turned to a piece of software. They never should have been so ignored that the software was what prompted them to push for getting a test done :c
Why not both? If the police fail to catch a criminal and the situation gets resolved by a courageous private citizen who is there at the time, it is both a failure of the police and a triumph of the private citizen.
Per the account without ChatGPT there was no way they’d even have an inkling to do another MRI.
Wholeheartedly agree. But. There are known and widespread flaws in how medicine is practiced relating to women’s health. There are known and widespread flaws in how medicine is practiced in general. And in how academic research is disseminated, and in how software is documented and tested, and in how business information is managed, and in how organisational knowledge is used, and in how…
And I really do get you, I’ve been the one hand wringing at obvious failures for most of my life because they bother me. But that’s the thing, I’ve been doing it for all those decades and these problems aren’t getting fixed because the fix is often “if people would just…”, and people generally don’t collectively “just <whatever>” even when the outcome would benefit them and everyone else.
So if we can use the machine to quickly and broadly patch human process gaps that’ve been embedding themselves for decades or centuries, I see it as a massive win. Perhaps not to the level the AI execs are promising, because I’m increasingly convinced most of them have lost their goddamn minds, but certainly to a level that’ll bring world-changing positives (will they outweigh the world-changing negatives those execs are pushing along the same path? Unclear, if I’m honest).
So yeah, it shouldn’t have happened, but if we’ve got an individual-level fix here and now and it can change things when all the plausible systematic fixes are just being ignored that’s still pretty exciting to me.
I'm not sure what particular reasons you're referring to as to why doctors don't like to perform certain tests, but a large reason why doctors don't like to do certain tests has to do with false positives. If the false positive rate for a test is higher than the prevalence of that condition in the general population (which it very often is), then you will do harm more than good by doing the test. The doctor has to have reason to believe you exist within a certain subpopulation whose prevalence for the condition is higher than the false positive rate for it to be worth it. The doctor also has a responsibility to not clog up the healthcare system with unnecessary testing.
This is also compounded by a mix of narratives like cuts that are "targeting waste and these doctors who order unnecessary tests" and "patients are going to doctor Google/AI and acting like they have medical degrees. So politics and the current doctor experience (and burnt out doctors post COVID too) are all factors as well.
No doctor is perfect, but neither is their education and experience worthless.
In an ideal world where the health insurance industry was ruthlessly competitive in favor of consumers, the insurance companies could actually play a substantial role in managing scarcity in the healthcare system. Unfortunately, we live very far from that world and there's all sorts of hogwash narratives pushed in healthcare under the guise of "reducing waste" and "cutting costs for patients" when it's really to juice profits.
They mention in the article laparascopy surgery of the abdomen visualizing the outside of the uterus, hysteroscopy, and ultrasound of the uterus. The author doesn't go into very much detail about the rest of their 6-year treatment, but it seems likely other imaging of the uterus was also done. It wasn't a failure to do any imaging, it was a failure to do yet another imaging after many imagings that hadn't found anything. Or failure to pick up on the potentially suspicious spot on the ultrasound that the AI noticed and based its recommendation of an MRI follow up on.
From the article:
This is me right now. I had relatively major surgery last week, and the entirety of my follow up is a phone call scheduled in 6 weeks and a four page pamphlet from the hospital (with my particular surgery circled and a note to ignore the advice for some related surgeries the pamphlet also discusses). In the meantime, I have about a hundred queries to LLMs about what’s happening and what I should be doing to recover.
If anyone is interested here author comment why fibroid was not detected using ultrasound:
I'm glad that the family in the article had the great outcome they did, but I wonder what the ratio between outcomes like this, and the mirror opposite outcome - - extra unnecessary tests and health anxiety over a routine illness.
Does anyone know of any studies on this topic?
Perspective: information overload has always been a problem, now made worse by speed and availability. But as always, humans have to decide for themselves when a source of too much info is harmful to them.
When I was expecting, I got a book called What To Expect When One Is Expecting -- my midwives (human ones, excellent all) were mildly scandalized and suggested I get rid of it. The book was already considered quite outdated by then, and it contains pretty much every possible thing that could possibly go wrong no matter how tiny of a possibility. My midwives has seen many other expectant mothers worry themselves thin pouring over the pages. Meanwhile, I read the information happily, and with the knowledge of how much could go wrong and hasn't, and how minuscule the chances are, I found it comforting.
So, my suggestion is that LLM will be a godsend for those who can benefit from them, but it will be awful for those who can't handle it. Like a book that became standard and then considered out dated and ignored, I hope the nervous types will also heed the words of others and know to avoid them.
I suppose I was thinking from the perspective of something like the pan-scans that are famous for finding "incedentalomas" that turn out to be nothing, so this was thought provoking, thank you.
There are certainly tradeoffs, but I’m hopeful that it isn’t always a necessary tradeoff. That is, it should be possible to somehow combine AI-based research and professional advice to get the best of both.
Hard to do though, with some patients.