My wife and I are planning to have children soon, and we will be using embryo selection to improve our family’s future well-being. Basically, the idea is that we can take a small sample of cells from each embryo, analyze their DNA, and use the genetic information to help choose which embryos we want to implant.
These days, polygenic scores are good at predicting risk for a lot of diseases, like cancer, heart disease, diabetes, schizophrenia, and Alzheimer’s. And companies like Herasight or Genomic Prediction often give an “embryo health score”[1] that takes an average of the predicted risk for each disease, weighted by how bad each disease is.
But how are those weights calculated? And can we do better?
Disability-adjusted life years
Diseases are bad in two ways: they can make you die, and they can make your life worse. The “disability-adjusted life year” (DALY) metric captures both of these, by combining years of life lost and years of reduced quality of life (using a factor between 0 and 1 to account for different subjective severity of disease states).[2]
Embryo selection companies, who compute polygenic scores for many different diseases, can add up the DALY burden of all the disease predictions for a given embryo. This gives an overall “embryo health score” that is a composite of these predictions.
DALY scores for five of our embryos, shown against the possible distribution of embryos given the parents’ DNA. Note the left-tailed skew. This is because an increasing genetic risk for a condition causes a nonlinearly increasing risk of having the condition.
A note on pleiotropy
Interestingly, many negative traits are genetically correlated with each other, so selecting against one selects against many. For example, our embryos that had a high schizophrenia risk also tended to have a high risk of bipolar. Pleiotropy (where one variant affects multiple traits) is often brought up in objections to embryo selection, e.g. “if you select for IQ, you might be selecting against health”, but in practice, mutations that make embryos less healthy will also generally hurt other traits.
But if a boy is born in 2027 with a high lifetime risk, he likely won’t develop any problems until middle age. Heart disease treatment will likely look very different in 2072! Therefore, we need to take into account future medical progress when weighing the badness of different diseases:
First, some things are already easily treatable. Statins lower cholesterol, and there are plenty of good medications for blood pressure. More recently, GLP1 drugs have made type 2 diabetes much less relevant. Some of these drugs are expensive today, but they will be generic in the future, and likely cheap and accessible.
On the other limb bud, any disease that is developmental in origin will not be treatable except by replacement of the affected structure. If something doesn’t form properly in the fetus, giving a drug to the adult won’t help.
However, organ replacement will probably be an option in 10-30 years, depending on the organ. This could include xenotransplantation or lab-grown “applied developmental biology” approaches. So things like heart disease and kidney disease will be curable, though this may be very expensive.
But nobody can get brain transplants.[3] Anything that affects the brain (depression, mental illness, neurodegeneration) will still be a huge problem. Very advanced technology may be able to replace the brain neuron-by-neuron, but I expect this to have the least progress compared to other diseases. Conditions that affect photoreception (e.g. colorblindness) will also be difficult to treat.
Cancer and autoimmune conditions are also not treatable with transplants, though emerging technologies will make treatment somewhat better.
Overall, I expect rapid medical progress in most areas over the next 50 years, especially given ongoing advances in AI. And if there’s a technological singularity, and everyone uploads their minds to run on silicon, then mental illnesses and personality disorders are the only thing that will matter!
Economic discounting
Leaving aside which conditions will be treatable in the future, the fact that some conditions have later average age of onset than others means that DALYs need to be adjusted for time-preference. For example, schizophrenia usually begins in early adulthood, whereas Alzheimer’s is a disease of old age. Selection for intelligence would have benefits at all stages of life.[4]
Embryo selection companies don’t take time-discounting into account for their embryo health scores.[5] However, health economists (e.g. in a study sponsored by the Gates Foundation) typically use a 3% annual discount rate. Personally, I think a 3% rate is on the low side: money invested today (instead of spent on embryo selection) would likely grow faster than that, especially given AI progress, and could be used to treat health conditions later in life. But I’ll use 3% in this post. Applied to the question of embryo selection, economic discounting means that late-onset conditions should be weighted less.
DALY adjustments
*IQ is valued through earnings (Grosse & Zhou 2021, updated to 2026 dollars). †BMI is accounted for through the diseases it causes and has no weight on its own.
These are Herasight’s DALY numbers,[6] adjusted for 3% economic discounting, and further adjusted for future medical progress. I assume negligible progress in mental and neurological disorders, moderate progress in cancer and autoimmune, and faster progress in other categories (especially ones which are related to obesity/metabolic dysfunction).
Even after discounting, embryo selection is still worth it
For couples doing IVF, embryo selection (PGT) costs vary by provider but are often in the range of $500 per embryo.[7] IVF itself is much more expensive: we paid about $7500 per cycle at CNY, including costs of medication and travel, and other clinics are even more expensive. Side effects of hormone stimulation for egg retrieval also take a toll: based on my wife’s experience, I estimate that these side effects cause a loss of 0.025 DALY per IVF cycle.[8]
But all this is still much less expensive than the diseases that selection can prevent. As of 2026, one DALY is valued by the US government (HHS) as:
$726,000 … for analyses using a 3% discount rate, and $1,205,000 for analyses using a 7% discount rate[9]
Other US healthcare organizations place lower values on DALYs, often around $200,000. The difference is because the HHS number measures what people will pay to reduce their own risk of death, but other organizations are measuring cost-effectiveness of medical interventions relative to current treatment options.
The top embryo from our first IVF cycle had a non-discounted DALY gain of +1.35 years, and estimated IQ gain of +10 points, relative to our family average. Economic discounting and medical progress would shrink the DALY gain to +0.18 years, but this is still a gain of $133,000 for health and $151,000 for IQ, much more than we paid for the IVF and selection.
Of course we can’t be sure that our top embryo will successfully implant and develop to term, so we need to scale these numbers down to account for the chance we need to use our second-best or lower ranked embryos instead. In the end, being able to choose among the five embryos from our first cycle, rather than transferring them in random order, has an expected value of about $112,000.
Additional cycles run into diminishing returns, but I calculated that this point is reached for our family after 6 - 8 cycles.
Here, the optimal number of cycles is calculated as expected benefit (from health only, or health + IQ) with current Herasight predictors, minus the expected cost (direct financial cost and DALYs lost from IVF medication side effects).[10]
Our family is different from a typical family in several ways, including higher genetic risk of bipolar and schizophrenia, and younger than average age for IVF patients (meaning more embryos per cycle).
For typical families, embryo selection for health is worth it only at the highest estimates of DALY values. However, the picture changes when also considering IQ selection: typical families would come out ahead from doing at least 3 rounds of IVF, and interestingly, lower value placed on DALYs increases the optimal IVF cycle number, because the DALY cost of medication side effects becomes less relevant.
Conclusions
Given rate of progress in medicine, and economic growth, embryo selection to prevent early-onset disease is more important than selection to prevent late-onset disease. Prospective parents who are considering embryo selection should take this into account. Even with discounting, embryo selection for health+IQ still passes the cost-benefit analysis for most couples.
Calculation of these factors involves surveying patients and the general public about their preferences (e.g. would you rather live a year with blindness or with Crohn’s disease?), and involves a lot of statistics which I won’t get into here.
Economically, we can quantify this by the effect of IQ on future income. Grosse and Zhou estimated this as $10,600 per point (on the low end) in 2016 dollars. Adjusting for inflation, that would be about $15,000 today. As of this writing, Genomic Prediction does not offer intelligence selection but Herasight does.
I asked a representative of Herasight about this, who said that they are considering adding an optional economic discounting feature in a future update, but haven’t formalized a framework yet.
This combines years of life lost and years lived with disability. Herasight also uses different weights for male and female embryos, since women tend to live longer and some diseases affect women and men at different base rates. The table shows the average.
My wife’s experience was about as bad as a month of major depression for each cycle (30% loss of quality of life for a month). I’m very grateful for her going through this.
They get these two different numbers by back-calculating from $14.1M as the statistical value of an average adult life based on surveys; i.e., an average American would be indifferent between losing $14,100 or having an 0.1% chance of death. Also, these numbers are technically QALYs not DALYs, though the difference is minor. Finally, for effects occurring in future years, they say to raise the value by 1.117% per year due to future real income growth, which partially offsets the discounting.
These calculations depend on: the expected number of euploid embryos per cycle (3 for typical IVF patients, 5 for us), the live birth rate per euploid embryo (60% for typical IVF patients, 67% for us), and genetic risk (population average vs. our specific risks).
My wife and I are planning to have children soon, and we will be using embryo selection to improve our family’s future well-being. Basically, the idea is that we can take a small sample of cells from each embryo, analyze their DNA, and use the genetic information to help choose which embryos we want to implant.
These days, polygenic scores are good at predicting risk for a lot of diseases, like cancer, heart disease, diabetes, schizophrenia, and Alzheimer’s. And companies like Herasight or Genomic Prediction often give an “embryo health score”[1] that takes an average of the predicted risk for each disease, weighted by how bad each disease is.
But how are those weights calculated? And can we do better?
Disability-adjusted life years
Diseases are bad in two ways: they can make you die, and they can make your life worse. The “disability-adjusted life year” (DALY) metric captures both of these, by combining years of life lost and years of reduced quality of life (using a factor between 0 and 1 to account for different subjective severity of disease states).[2]
Embryo selection companies, who compute polygenic scores for many different diseases, can add up the DALY burden of all the disease predictions for a given embryo. This gives an overall “embryo health score” that is a composite of these predictions.
DALY scores for five of our embryos, shown against the possible distribution of embryos given the parents’ DNA. Note the left-tailed skew. This is because an increasing genetic risk for a condition causes a nonlinearly increasing risk of having the condition.
A note on pleiotropy
Interestingly, many negative traits are genetically correlated with each other, so selecting against one selects against many. For example, our embryos that had a high schizophrenia risk also tended to have a high risk of bipolar. Pleiotropy (where one variant affects multiple traits) is often brought up in objections to embryo selection, e.g. “if you select for IQ, you might be selecting against health”, but in practice, mutations that make embryos less healthy will also generally hurt other traits.
Which diseases will be relevant in the future?
From as far back as 1950 until today, heart disease has been the biggest killer in the USA. And the DALY scores for embryos reflect this: heart disease has a strong weight.
But if a boy is born in 2027 with a high lifetime risk, he likely won’t develop any problems until middle age. Heart disease treatment will likely look very different in 2072! Therefore, we need to take into account future medical progress when weighing the badness of different diseases:
Overall, I expect rapid medical progress in most areas over the next 50 years, especially given ongoing advances in AI. And if there’s a technological singularity, and everyone uploads their minds to run on silicon, then mental illnesses and personality disorders are the only thing that will matter!
Economic discounting
Leaving aside which conditions will be treatable in the future, the fact that some conditions have later average age of onset than others means that DALYs need to be adjusted for time-preference. For example, schizophrenia usually begins in early adulthood, whereas Alzheimer’s is a disease of old age. Selection for intelligence would have benefits at all stages of life.[4]
Embryo selection companies don’t take time-discounting into account for their embryo health scores.[5] However, health economists (e.g. in a study sponsored by the Gates Foundation) typically use a 3% annual discount rate. Personally, I think a 3% rate is on the low side: money invested today (instead of spent on embryo selection) would likely grow faster than that, especially given AI progress, and could be used to treat health conditions later in life. But I’ll use 3% in this post. Applied to the question of embryo selection, economic discounting means that late-onset conditions should be weighted less.
DALY adjustments
*IQ is valued through earnings (Grosse & Zhou 2021, updated to 2026 dollars). †BMI is accounted for through the diseases it causes and has no weight on its own.
These are Herasight’s DALY numbers,[6] adjusted for 3% economic discounting, and further adjusted for future medical progress. I assume negligible progress in mental and neurological disorders, moderate progress in cancer and autoimmune, and faster progress in other categories (especially ones which are related to obesity/metabolic dysfunction).
Even after discounting, embryo selection is still worth it
For couples doing IVF, embryo selection (PGT) costs vary by provider but are often in the range of $500 per embryo.[7] IVF itself is much more expensive: we paid about $7500 per cycle at CNY, including costs of medication and travel, and other clinics are even more expensive. Side effects of hormone stimulation for egg retrieval also take a toll: based on my wife’s experience, I estimate that these side effects cause a loss of 0.025 DALY per IVF cycle.[8]
But all this is still much less expensive than the diseases that selection can prevent. As of 2026, one DALY is valued by the US government (HHS) as:
Other US healthcare organizations place lower values on DALYs, often around $200,000. The difference is because the HHS number measures what people will pay to reduce their own risk of death, but other organizations are measuring cost-effectiveness of medical interventions relative to current treatment options.
The top embryo from our first IVF cycle had a non-discounted DALY gain of +1.35 years, and estimated IQ gain of +10 points, relative to our family average. Economic discounting and medical progress would shrink the DALY gain to +0.18 years, but this is still a gain of $133,000 for health and $151,000 for IQ, much more than we paid for the IVF and selection.
Of course we can’t be sure that our top embryo will successfully implant and develop to term, so we need to scale these numbers down to account for the chance we need to use our second-best or lower ranked embryos instead. In the end, being able to choose among the five embryos from our first cycle, rather than transferring them in random order, has an expected value of about $112,000.
Additional cycles run into diminishing returns, but I calculated that this point is reached for our family after 6 - 8 cycles.
Here, the optimal number of cycles is calculated as expected benefit (from health only, or health + IQ) with current Herasight predictors, minus the expected cost (direct financial cost and DALYs lost from IVF medication side effects).[10]
Our family is different from a typical family in several ways, including higher genetic risk of bipolar and schizophrenia, and younger than average age for IVF patients (meaning more embryos per cycle).
For typical families, embryo selection for health is worth it only at the highest estimates of DALY values. However, the picture changes when also considering IQ selection: typical families would come out ahead from doing at least 3 rounds of IVF, and interestingly, lower value placed on DALYs increases the optimal IVF cycle number, because the DALY cost of medication side effects becomes less relevant.
Conclusions
Given rate of progress in medicine, and economic growth, embryo selection to prevent early-onset disease is more important than selection to prevent late-onset disease. Prospective parents who are considering embryo selection should take this into account. Even with discounting, embryo selection for health+IQ still passes the cost-benefit analysis for most couples.
“Embryo health score” is Genomic Prediction’s term; Herasight calls this “polygenic longevity index”.
Calculation of these factors involves surveying patients and the general public about their preferences (e.g. would you rather live a year with blindness or with Crohn’s disease?), and involves a lot of statistics which I won’t get into here.
If someone got a “brain transplant” it would actually be a “body transplant”.
Economically, we can quantify this by the effect of IQ on future income. Grosse and Zhou estimated this as $10,600 per point (on the low end) in 2016 dollars. Adjusting for inflation, that would be about $15,000 today. As of this writing, Genomic Prediction does not offer intelligence selection but Herasight does.
I asked a representative of Herasight about this, who said that they are considering adding an optional economic discounting feature in a future update, but haven’t formalized a framework yet.
This combines years of life lost and years lived with disability. Herasight also uses different weights for male and female embryos, since women tend to live longer and some diseases affect women and men at different base rates. The table shows the average.
Some companies like Herasight charge a fixed fee for analysis of any number of embryos, though biopsy fees also need to be paid to the IVF clinics.
My wife’s experience was about as bad as a month of major depression for each cycle (30% loss of quality of life for a month). I’m very grateful for her going through this.
They get these two different numbers by back-calculating from $14.1M as the statistical value of an average adult life based on surveys; i.e., an average American would be indifferent between losing $14,100 or having an 0.1% chance of death. Also, these numbers are technically QALYs not DALYs, though the difference is minor. Finally, for effects occurring in future years, they say to raise the value by 1.117% per year due to future real income growth, which partially offsets the discounting.
These calculations depend on: the expected number of euploid embryos per cycle (3 for typical IVF patients, 5 for us), the live birth rate per euploid embryo (60% for typical IVF patients, 67% for us), and genetic risk (population average vs. our specific risks).