DIY CRISPR Kits – Gene Editing for the Rest of Us

One might think with the amazing advance of technology and easy access to nearly infinite data via the Internet that we, as a society, would see a reduction in false claims of benefit for novel medical procedures and untested medications. Sadly, it seems to be just the opposite. I seem to be spending gradually more time with my patients reviewing the results of their internet research for new solutions for their chronic back pain. Their efforts are laudable even though the “hoped for” benefits claimed in their researched solutions are woefully lacking. Unfortunately, often this exercise in reviewing the outside data takes valuable time away from the remainder of the office visit.

Reviewing false or confusing information is one thing but preventing patients from self-experimentation with untested medications or unproven treatments is another. Enter the biohacker and companies offering do-it-yourself (DIY) kits claiming to allow anyone to experiment with CRISPR (a method of genetic editing) for self-administration. Emily Mullin covers biohacking and DIY CRISPR very nicely in her recent article in the December Technology Review. To me, this has the feel of the 1980s when a curious kid with some basic programming knowledge, an inexpensive computer and a modem can access previously forbidden government systems, potentially unleashing havoc on the rest of us (WarGames, anyone?) After all, now that we know the human genetic code, all we need is for someone to just provide the instructions and tools for editing that code, then anyone could tweak their own DNA. Easy peasy lemon squeezy, right?

Recently, the FDA has been busy trying to prevent medical clinics from administering untested stem cell treatments (see Neil Skjoldal’s recent November blog entry on (Stem Cell Clinics & the FDA). Imagine the significant increase in the scope of the regulatory problem if individuals can order a DIY CRISPR kit off the Internet!

While we might chagrin at the naiveté required to believe the street-side pitch of the Old West Carter’s Little Liver Pill salesman, that same pitch via a modern tech savvy YouTube video (complete with separate internet links) somehow offers a new level of legitimacy. The Technology Review article speculated that one of the featured companies was preparing not a vaccine but a treatment for herpes. In less than 8 weeks from the article’s publication, Aaron Traywick, CEO of Ascendance Biomedical, publically self-injected himself with his firm’s untested and non-FDA approved “treatment” for herpes. The linked article by Reegan Von Wildenradt in the popular magazine Men’sHealth offered an excellent counter as to why this type of “science” might be suspect, including quotes from ethicist Arthur L. Caplan at NYU in support of the standard FDA process for screening medical treatments.

We often lament in this blog that technology is advancing so rapidly that we fail to have a fair public hearing and discussion of the ethics involved in a particular biomedical advance. Now it seems our time may be better spent speaking out first about the basic risks of the new technology and doing our best to support the FDA in their massive task of policing the Internet to prevent a DIY CRISPR kit from falling into the wrong hands – ours.

P.S. – I’m accepting names for the title of the future Hollywood blockbuster where the son of Matthew Broderick and Ally Sheedy injects himself with his own DIY CRISPR-modified DNA and …

Citizenship, Surrogacy and the Power of ART

A recent LA Times article by Alene Tchekmedyian explores a complicated case involving birthright citizenship, surrogacy and same-sex marriage. Briefly, a California man, Andrew Banks, married an Israeli man, Elad Dvash, in 2010. At the time, same-sex marriage was not legal in the US leaving Elad unable to acquire a green card for residency (via the marriage) so the couple moved to Canada where Andrew has dual citizenship. While in Canada, the couple conceived twin boys, Aiden and Ethan, using assisted reproduction technology (ART) whereby eggs from an anonymous donor were fertilized by sperm from Elad and Andrew and then implanted within the womb of a female surrogate and carried to term. When the US Supreme Court struck down the federal law that denied benefits to legally married gay couples in 2013, Elad applied for and was granted his greed card. The present controversy occurred when Andrew and Elad applied for US passports for the twins. US State Department officials required detailed explanation of the boys’ conception, eventually requiring DNA tests which confirmed Aiden to be the biological son of Andrew and Ethan to be the biological son of Elad. Aiden was granted a US passport while Ethan was denied. The family has since traveled to the US (Elad with his green card and Ethan with his Canadian passport and temporary 6 month visa) where they are now suing the State Department for Ethan’s US birthright citizenship. They are arguing that the current applicable statute places them wrongly in the category of children born out of wedlock rather than recognizing their marriage, thus discriminating against them as a binational LGBTQ couple.

Birthright citizenship is a complicated legal arena and I am no lawyer. The US is even more complicated because we allow birthright citizenship to be conferred jus soli (right of the soil) in addition to jus sanguinis (right of blood). The twins were not born in the US so establishing “bloodline” is needed. The law specifies conditions where one parent is a US citizen and one is not a US citizen, and there is further differentiation depending on whether the children of the US citizen were born in or out of wedlock. They also vary depending on whether the US citizen is male or female, with the law more lenient (easier to acquire citizenship) for the child of a woman than of a man.

While the legal challenge here will almost certainly involve potential issues of discrimination of LGBTQ binational couples, the problem is really with the current legal definitions of parent as it relates to surrogacy in general. The State Department actually has a website dedicated answering questions related to foreign surrogacy and citizenship. The real issue is that the State Department relies upon genetic proof of parentage for foreign surrogacy births. In the present case, the surrogacy occurred outside the US, Elad is the genetic father of Ethan and Elad is not a US citizen; therefore Ethan is not a US citizen. While I’m deep in the weeds here, technically, Aiden and Ethan are not fraternal twins in the usual sense but rather half siblings (and this assumes that the donor eggs are from the same woman; otherwise the boys would be unrelated despite sharing the same pregnant womb through the magic of ART). Had Ethan been physically born via surrogacy in the US, he would have acquired his citizenship via jus soli (see US map for surrogacy friendly states near you).

This problem is just as confounding for heterosexual couples using foreign surrogates, and the problem is global. A more detailed technical legal discussion may be found here. A heterosexual couple using donor eggs and donor sperm and using a foreign third party surrogate would have exactly the same problem establishing US citizenship for “their” child. A similar problem would exist for an adopted embryo gestated in a foreign country by a foreign surrogate. If either the egg or the sperm of the US citizen is used for the surrogate birth, the child would be granted birthright citizenship.

The main difference for homosexual couples is that only one spouse can presently be the biological parent. I say “presently” because with ART it is theoretically possible (and may become actually possible in the future) to convert a human somatic cell into either a male sperm or a female egg. At that point, both spouses within a same-sex marriage could be the biological parents of their child. The present legal issue is not the result of a cultural prejudice against anyone’s sexuality but with the biological prejudice of sex itself. ART has the potential ability to blur the categories of sex as culture is now blurring the categories of gender. Should we consider this a good thing?

Given the present technological limits of ART, the simple issue of US citizenship could be resolved in all these cases if the US citizen parent simply adopted the child. Elad correctly points out that while adoption of Ethan by Andrew would grant Ethan US citizenship, it would not grant Ethan birthright citizenship, a necessary requirement for Ethan to someday run for US president. ART may be forcing us to look at changing our definition of parent but should it change our definition of biology? Ethan is the biological son of Elad. He is able to be the legally adopted son of Andrew and enjoy the benefits of US citizenship as currently does his half brother Aiden. He is not able to become the biological son of Andrew and enjoy the additional benefit of birthright citizenship via jus sanguinis.

Should we change the definition of birthright citizenship because ART is changing our definition of parent?

Fertility with frozen eggs: not a sure thing

In case you didn’t see it, the Washington Post has this story about how more women are trying to improve their overall chances of having a baby—particularly in the later reproductive years of their 30’s and 40’s—but success is far from certain.  Human oocytes (eggs) are fragile things, and it was not until recent years that freezing techniques developed to a point that would allow the eggs to survive being frozen and, some time later, thawed (the “freeze-thaw” cycle).  Then, they would be fertilized in the lab, by in vitro fertilization, and implanted into the womb of the would-be mother.

As the article points out, women are born with their entire endowment of eggs, which become less likely to be successfully fertilized and develop into a healthy baby as they, and the woman, age.  Hence a woman’s inexorably declining fertility, particularly from their mid-30’s on.  Freezing eggs for later use is increasingly popular, if one can afford it, or if employers offer it as a perk, as some do, to their female employees.

It’s still expensive, and success appears to depend on the age of the woman (and eggs) at their harvest, and the number harvested and kept in frozen storage.  One must use the qualifier “appears,” because, as the article also points out, reliable statistics are not being kept.  The not-so-subtle implication is that the fertility “industry” wants to sell the process but would rather not know that the ultimate success rate could be as low as, or lower than, the 50-60% rate quoted by New York University.

Clear implications: better data and more transparency are to be desired, and there appear to be at least some remaining biologic limits, strong if not absolute, to reproductive freedom.  Beyond that, as I opined in May of 2013 (fairly bluntly, I do confess) are the radical implications for our concepts of parenthood and begetting children, and for turning said procreation into just plain old, quality-controlled, fully artificial creation.  Things haven’t gotten quite so absolute, yet.  But better quality control of egg freezing and the outcomes, if possible, would be a move in the direction of more artificial reproduction.

It’s a good article from the Post.  Too much to try to do justice to here.  Read the whole thing.

What’s really happening with doctor-assisted suicide?

Recently, Wesley Smith posted on the National Review’s “Corner” blog new concerns that Oregon’s “Death With Dignity” law may not be as tightly regulated as advertised.  Specifically, a Swedish fellow named Fabian Stahle, who evidently is troubled by the prospect that his country might embrace doctor-assisted suicide, claims to have carried out an e-mail exchange with someone in the Oregon Health Authority to ask how the law is interpreted in that state.  The responses included a statement that, to qualify for assisted suicide, a patient must have a “terminal illness” but said illness could include a potentially treatable condition which, if allowed to take its course without treatment, would be expected to cause death within 6 months.  That suggests that assisted suicide might be legally employed in Oregon in cases in which the patient refused treatment or the patient’s insurance company refused to pay for effective treatment.  The Oregon official cited by Mr. Stahle is quoted by him as having written that “the law is best seen as a permissive law…[that] does not compel patients to have exhausted all treatment options first, or to continue current treatment.”

You can read Mr. Stahle’s entire account here.

I must say that, while I mention this for this blog, I have not attempted to confirm these assertions by contacting the Oregon officials myself, and my first reaction is skepticism that what Mr. Stahle reports is in fact the correct interpretation of the Oregon law.  Even I, a staunch opponent of assisted suicide, must allow that the intent of such laws seems to be that assisted suicide is intended for cases for which potentially effective treatment options HAVE been exhausted.  But I suppose that further investigation is in order.

While at it, Wesley Smith also cited a 2005 British House of Lords inquiry into the Oregon law, from which a group apparently opposed to assisted suicide posted some comments here.  The entire House of Lords transcript, BTW—all 744 pages of it!—is available here for interested parties.  I must confess I have not had time to read the whole thing.

The most recent data summary from the state of Oregon that I am aware of is for 2016.  It reports 133 deaths from taking drugs from 204 lethal prescriptions filled in the state in 2016.  These numbers were slightly down from 135 and 214 in 2015.  The 2016 “death with dignity” rate is cited as 37.2 per 10,000 deaths in Oregon.  Of the 133 people who died with medical assistance in Oregon in 2016, 96% were white, about 80% had cancer, nearly all had some form of insurance, and about 85% were age 65 or older.  The two most commonly-stated reasons for seeking assisted suicide were loss of autonomy and loss of enjoyment of life—about 90% in each case.  Inadequate pain control was listed for about one-third of the cases.  The median time the patient had been seeing his or her prescribing doctor was 18 weeks, and the prescribing doctor was present for 13 of the 133 deaths.

Of course, all of this assumes the reporting is complete and accurate.  I have no information that would lead me to believe otherwise.  I state the facts in the preceding paragraph without commentary or, in some cases, without the irony I feel in reading them.

I can’t conclude from the Oregon report that patients who availed themselves of assisted suicide there were foregoing potentially effective treatment for their disease, much less that an insurance company refused to pay for it.  Of course, there is the one famous case of a person there getting a “suggestive” note about PAS from his insurer, some years ago—I can’t locate it at the moment.

In general, I’d say that the concerns raised through the above-mentioned posts are ones we must keep in mind, but that the slope may not yet have gotten that slippery.

Finally: the National Academies of Science, Engineering, and Medicine will hold a conference in Washington DC on Monday and Tuesday, February 12 and 13.  Looks like a webcast is available.  You can find information about it, and sign up to attend in person or by the web, here.  I’m going to try to at least watch some of it. From the information at that site:

This workshop will include discussions, and background materials, that address:

  • What is known empirically about the access to and practice of physician-assisted death in the U.S. and in other countries?
  • What are potential approaches for physicians, including those practicing in states where it is legal, those who receive a request for access when the practice is legal in nearby states but not in the state of practice, and those who practice in a state where it is legal but are personally opposed to physician-assisted death.
  • What is known about how palliative care and hospice services have incorporated the practice of physician-assisted death in states where it is legal?

Will Medical Compliance Ever Become Non-Voluntary?

A recent article by Dr. Lisa Rosenbaum in the New England Journal of Medicine explored both the benefits and drawbacks of Digital Adherence Monitoring. The focus was on the FDA’s recent approval of Abilify MyCite, a medicine technology that combines the medication aripiprazole, used to treat various psychiatric diseases such as schizophrenia, certain features of bipolar disorder and depression, with a digital ingestion tracking system. This voluntary digital health feedback system (DHFS) works by having the patient wear a skin patch that is triggered when the pill contacts the acid in the stomach. This event is then recorded and tracked on the patient’s smartphone. The patient can then permit their caregivers and/or physicians to access the data via a web portal. The company responsible for the DHFS, Proteus, has shown improvement in patient’s systolic blood pressure using DHFS compared with standard care. The article primarily focuses on using the technology to help doctors work with their patients to determine the reasons for non-compliance.

While this presently voluntary technology obviously can track pill ingestion and this data can certainly help doctors and patients improve medication treatment adherence, I wondered about non-voluntary uses of the technology. This particular DHFS confirms that the prescribed pill was actually ingested regardless of what the patient or their caregiver may claim. Would an insurance company be permitted to have access to this data in exchange for payment for a particularly expensive medication? Could a government agency require such a system in exchange for providing coverage for a patient for a procedure whose subsequent outcome is improved with the use a given medication?

Dr. Rosenbaum offered in her article that she thought it unethical to withhold coronary artery bypass from one of her patients with whom she was fairly certain would not subsequently take the dual antiplatelet therapy post revascularization. Using a DHFS eliminates mere suspicion. Prematurely discontinuing of thienopyridine therapy (antiplatelet drugs such as Effient, Ticlid, or Plavix) after a similar cardiac stent placement has been shown to increase the risk of both re-hospitalization and death within the subsequent 12-month period. Given the success of the Proteus DHFS in reducing systemic high blood pressure, mandating this DHFS to monitor antiplatelet therapy immediately post cardiac stent placement should reduce both patient morbidity and mortality during the following 12-month period.

A consequentialist in charge of public health care expenditures might disagree with Dr. Rosenbaum regarding the ethics of providing a revascularization procedure in an individual who is poorly compliant with beneficial post-procedure medication compliance. Bluntly, why spend the money if the patient (for whatever reason) is going to act in a manner to reduce the benefit of her procedure? Thankfully, money is not the only healthcare utility worth measuring and economists are not yet fully in charge of healthcare delivery, though they appear to have an ever increasingly important seat at the table.

So, I think DHFS technologies such as Abilify MyCite will slowly become non-voluntary.

Racial inequalities in cancer survival

Three studies published in a supplemental issue of the journal Cancer this month come to disturbing conclusions: in the United States, the survival rates for colon, breast, and ovarian cancer are lower for black people than for white people.

The news isn’t all bad: overall cancer survival rates are going up. The three studies mentioned here draw from two larger studies of worldwide cancer survival, the CONCORD study, published in 2008, and the CONCORD-2 study, published in 2015. Between CONCORD and CONCORD-2, cancer survival increased across all groups. But in both studies, the survival of black people in the United States lagged behind that for white people by about 10%.

Now there are certainly many reasons for the difference. Black people may be getting diagnosed with cancer at later stages, when survival is lower. There may be differences between the two groups, like genetic factors or the presence of other illnesses, that cause the cancers to be more aggressive in blacks. Mistrust of the health care system is more common among black patients, so they may be less likely to access care or access it early enough.

But other reasons include socio-economic status and access to health care; those who can’t access medical care because they can’t afford it or because it is not available nearby are less likely to receive necessary screening and treatment. And most troubling is the “consistent finding that black women do not receive guidelines-based treatment compared with white women, even when treated within the same hospital.”

The situation is too complex for simplistic answers; one can’t say from these studies that doctors are individually practicing blatant racism, and there is after all such a thing as systemic racism. But the medical profession should take such findings seriously, and seek every available avenue of education and self-monitoring to ensure that of all the places people encounter racism, the health care system is not one of them. And Christians especially, who believe that God created all people with equal dignity because all of us are created in His image, should work towards a health care system where the value — or length — of one’s life does not vary based solely on the color of one’s skin.

Uterine Transplantation – for Men?

Susan Haack began exploring the topic of uterine transplantation in women on this blog back in February 2014. In just under 4 short years, the technology has not only successfully resulted in live births in several women who received the uterine transplants, but outgoing president of the American Society of Reproductive Medicine, Dr. Richard Paulson, is suggesting we consider exploring the technique in men. While there are certainly hurdles to overcome (need for cesarean section for the actual birth, hormone supplementation, complicated nature of the transplant even for cisgender women), Dr. Paulson does not consider these barriers insurmountable for transgender women.

Dr. Julian Savulescu, professor of practical ethics at Oxford, has cautioned that initiating a pregnancy in a transgender woman may be unethical if it poses significant risk to the fetus. The above-linked article misquotes his concern as a concern over “any psychological harm to the child born in this atypical way”. The following is his actual quote from his own blog:

Therefore, although technically possible to perform the procedure, you would need to be very confident the uterus would function normally during pregnancy. The first US transplant had to be removed because of infection. There are concerns about insufficient blood flow in pregnancy and pre-eclampsia. A lot of research would need to be done not just on the transplant procedure but on the effect in pregnancy in non-human animals before it was trialled in humans. Immunosuppressives would be necessary which are risky. A surrogate uterus would be preferable from the future child’s perspective to a transplanted uterus. Uterine transplantation represents a real risk to the fetus, and therefore the future child. We ought to (other things being equal) avoid exposing future children to unnecessary significant risks of harm.

One putative benefit might be the psychological benefit to the future mother of carrying her own pregnancy. This would have to be weighed against any harm to the child of being born in this atypical way.

His concerns are the baseline medical risks involved in using a transplanted uterus to conceive a child regardless of the sex of the recipient. None of his concerns relate to the psychological harm to the child potentially caused by a uterine transplantation in a transgender woman as opposed to a cisgender woman. Savulescu is explicit in the beginning of his blog that “[t]he ethical issues of performing a womb transplant for a [sic] transgender women are substantially the same as the issues facing ciswomen.” Is the only risk to the child “born this atypical way” just the additional need for hormone supplementation in the transgender woman compared to the cisgender woman? Can we really know, a priori, what all of the attendant risks to the child really are with uterine transplantation in a transgender woman?

Regardless, let’s assume Savulescu is correct, that there is indeed no ethical difference between carrying a child to term via uterine transplantation between a cisgender woman and a transgender woman. There certainly can be no ethical difference between carrying a child to term via uterine transplantation between a transgender woman and a cisgender man. If the foregoing is true, can there be any ethical barrier preventing a man via uterine transplantation to use his sperm to fertilize a donor egg and carry his baby to term? After all, per Savulescu, all we need be concerned about from a bioethical standpoint are the technical issues/risks of uterine transplantation regardless of the recipient’s biological sex or self-identified gender.

In Genesis, God created two complimentary sexes and stated this difference was good. We are moving toward eliminating differences between the sexes and arguing that this is good. Both of us cannot be correct.

I wonder if Dr Haack thought that we would get this far down this particular bioethical slippery slope in four short years?

Being thankful

My wife is a writer. She writes everything from murder ministries, for which I am occasionally the in-house consultant, to a humor column and, of course, a blog. For years she has written a column at this time of year about the weird things that she is thankful for. You can read this year’s list here.

I am not the accomplished writer that my wife is. What I do write is usually about ethics and medicine. I also like to reflect at this time of year about what I am thankful for. The other day I received an invitation to the 40th year reunion of my medical school class which will be held next May. That got me thinking about what I have experienced over nearly 40 years of being a physician.

When I look back, the thing that I am most thankful for is the opportunity to be involved in the lives of those who have been my patients. Medicine is a relational profession. The people that we care for allow us to become intimately involved in their lives. No matter how much technology we employ it is the relationship between a person in need in a person with the training and skills to help with that need that is the foundation of the medical profession. My patients have allowed me to share with them their joys and their sorrows. I have shared the joy of a new life coming into the world and the joy of recovery from a serious illness. But the most meaningful times have been those in which there was very little that I could do beyond being there with those for whom nothing more could be done and letting them know that I cared about them.

I thank God for the opportunity to share in the lives of those who have chosen to call me their family doctor.

Is Your Polygenic Risk Score a Good Thing?

Back in October, Jon Holmlund wrote a blog entry regarding the popular company 23andMe and their collection of your health-related information along with your genetic material. I missed the significance of that relationship at the time. It took a recent article in Technology Review by my favorite technology writer Antonio Regalado to raise my ethical antennae. In his article, he explains the nexus of big data mining of genetic data and health information (such as is collected by 23andMe) and its future potential use to select embryos for IVF, selecting not only against polygenic diseases such as type 1 diabetes but potentially for non-diseases such as height, weight or even IQ.

Yikes.

Pre-implantation genetic diagnosis (PGD) already is used to select for particular embryos for IVF implantation that do not have genetic patterns such as cystic fibrosis or Down syndrome. Diseases that result from multiple genes (polygenic disorders) presently defy current PGD methods used to detect future diseases. Using Big Data analysis of health information compared against linked genetic data, scientists are getting better at accurate polygenic risk scores, statistical models which may more accurately ‘guess’ at an embryo’s future risk for not only juvenile diabetes but also later-in-life diseases (such as heart disease, ALS or glaucoma) or other less threatening inheritable traits (such as eye color, height or IQ) that result from multiple genes (and perhaps even environmental factors). There is confidence (hubris?) that with enough data and enough computing power, we can indeed accurately predict an embryo’s future health status and all of his or her inheritable traits. Combine that further with all of the marketing data available from Madison Avenue, and we can predict what type and color of car that embryo will buy when he or she is 35.

Ok, maybe not the color…

Seriously, companies such as Genomic Prediction would like to see IVF clinics eventually use their expanded statistical models to assist in PGD, using a proprietary technique they are calling Expanded Pre-implantation Genomic Testing (EPGT). Consider the following two quotes from Regalado’s article:

I remind my partners, “You know, if my parents had this test, I wouldn’t be here,” says [founding Genomic Prediction partner and type 1 diabetic Nathan] Treff, a prize-winning expert on diagnostic technology who is the author of more than 90 scientific papers.

For adults, risk scores [such as calculated by 23andMe] are little more than a novelty or a source of health advice they can ignore. But if the same information is generated about an embryo, it could lead to existential consequences: who will be born, and who stays in a laboratory freezer.

Regalado’s last comment is dead-on – literally. Who will be born and who stays in the freezer is another way of saying “who lives and who dies”.

Technologies such as EPGT are poised to take us further down the bioethical slope of choosing which of our children we want to live and which we choose to die. For the sake of driving this point home, let’s assume that the technology becomes essentially 100% accurate with regard to polygenic risk scoring and we can indeed determine which embryo will have any disease or trait. Since we already permit the use of single gene PGD to prevent certain genetic outcomes, should there be any limit to polygenic PGD? For instance:

(A) Should this technology be used to select against immediate life threatening illnesses only or also against immediate mentally or physically permanently crippling diseases that don’t cause death directly?

(B) Should this technology be used to select against later-in-life diseases that are life threatening at the time or also against mentally or physically crippling diseases that don’t cause death directly? (Would it make a difference if the disease occurred as a child, teenager or adult?)

(C) Should this technology be used to select against non-disease inheritable traits that society finds disadvantageous (use your imagination here)?

(D) Should this technology be used to select for inheritable traits that society finds advantageous (a slightly different question)?

Depending upon your worldview, until recently, answering Questions A through D used to be the purview of God or the random result of chance. Are we ready (and capable) to assume that responsibility? Make your decision as to where you would draw the line then review this short list of famous scientists and see how many on that short list your criteria would permit to be born.

Are you happy with that result? Would you call it good?

It would be nice to get this right since it now appears to be our call to make…

Is Medical Artificial Intelligence Ethically Neutral?

Will Knight has written several articles over this past year in MIT’s flagship journal Technology Review that have discussed growing concerns in the field of Artificial Intelligence (AI) that may be of concern for bioethicists. The first concern is in the area of bias. In an article entitled “Forget Killer Robots – Bias is the Real AI Danger”, Knight provides real world examples of this hidden bias affecting people negatively. One example is an AI system called COMPASS, which is used by judges to determine the likelihood of reoffending by inmates who are up for parole. An independent review claims that algorithm may be biased against minorities. In a separate article, Knight identified additional examples in other AI algorithms that introduced gender or minority bias in software used to rank teachers, approve bank loans and interpret natural language processing. None of these examples argued that this bias was introduced intentionally or maliciously (though that certainly could happen).

This is where Knight’s the second concern becomes apparent. The problem may be that the algorithms are too complex for even their programmers to retroactively examine for bias. To understand the complexity issue, one must have an introductory idea of how the current AI programs work. Previously, computer programs had their algorithms “hard-wired” so to speak. The programs were essentially complex “if this, then do this” sequences. A programmer could look at the code and generally understand how the program would react to a given input. Beginning in the 1980’s, programmers started experimenting with code written to behave like a brain neuron might behave. The goal of the program was to model a human neuron, including the ability of the neuron to change its output behavior in real time. A neurobiologist would recognize the programming pattern as modeling the many layers of neurons in the human brain. A biofeedback expert would recognize the programming pattern as including feedback to change the input sensitivities based upon certain output goals – “teaching” the program to recognize a face or image in a larger picture is one such example. If you want to dive deep here, begin with this link.

This type of programming had limited use in the 1980s because the computers were too simple and could only model simple neurons and only a limited number at one time. Fast forward to the 21st century and 30 years of Moore’s Law of exponential growth in computing power and complexity, and suddenly, these neural networks are modeling multiple layers with millions of neurons. The programs are starting to be useful in analyzing complex big data and finding patterns (literally, a needle in a haystack) and this is becoming useful in many fields, including medical diagnosis and patient management. The problem is that even the programmers cannot simply look at these programs and explain how the programs came to their conclusions.

Why is this important to consider from a bioethics standpoint? Historically, arguments in bioethics could generally be categorized as consequentialist, deontological, virtue, hedonistic, divine command, etc… One’s stated position was open to debate and analysis, and the ethical worldview was apparent. A proprietary, cloud-based, black-box, big data neural network system making a medical decision obscures, perhaps unintentionally, the ethics behind the decision. The “WHY” of a medical decision is as important as the “HOW”. What goes in to a medical decision often includes ethical weighting that ought to be as transparent as possible. These issues are presently not easily examined in AI decisions. The bioethics community therefore needs to be vigilant as more medical decisions begin to rely on AI. We should welcome AI as another tool in helping us provide good healthcare. Given the above concerns regarding AI bias and complexity, we should not however simply accept AI decisions as ethically neutral.