AI, ethics and medicine

Machine learning and artificial intelligence have the potential to transform medicine in an incredibly wide range of ways. For example, doctors learn how to interpret the scans that they look at by looking at lots of them. Machines can do that with an infinite number of scans and learn much more quickly and accurately to diagnose what they’re seeing. So, one potential consequence is replacing humans with machines to do things like read and interpret scan results.

Huge data sets have the potential to enable us to do medical research much more quickly and thoroughly, because obviously the more data you’re learning from, the better. The NHS is now aware that its data of the medical treatment of everybody in the UK isn’t just an important source of information for doctors, but it is actually valuable. This idea of data sets as having value is a relatively new development in medicine.

Something that lots of people are not necessarily aware of is that the smartphones that they carry around with them in their pockets are gathering health data about them and are sources of health data. We normally think of health data as being something that’s kept in locked filing cabinets and treated as confidential, but our mobile phones contain an enormous amount of information about our fitness and health.

From the way in which you use your keyboard, big data companies can identify the early signs of Parkinson’s disease. By engaging with big data companies, you are telling them an enormous amount of information about yourself. Tracking flu epidemics, for example, can be done very easily by tracking search terms in Google; people searching for symptoms means that it’s possible for Google to tell more easily than other public health agencies where a particular flu episode is increasing. What we’re seeing is a recognition that the preoccupation of big tech companies with data also has huge health implications as well.

Genome editing is an especially exciting new development. The inventors of the CRISPR-Cas9 genome editing system, which is essentially a much simpler way to edit a genome than was previously thought possible, won the Nobel Prize. The way the law approaches something like this is to say that this might be very exciting for research purposes, but we must not do it in the clinic until we know about its long-term safety.

Worldwide, there’s a consensus among scientists that while it’s incredibly exciting to carry out research involving genome editing, perhaps even on human embryos, it should never be used in treatment. That is why the Chinese doctor who announced that he had edited the genomes of babies that have been born has gone to jail; it was regarded as beyond the pale, scientifically. So, the law says you can do this for research purposes and we need to learn about this, but we need a whole lot more evidence before you can actually do this using patients.

It’s not currently possible to edit a baby’s genome before in vitro fertilisation (IVF) treatment. That’s absolutely illegal, and a change in the law would be necessary before that became possible. But it is possible – and has been since 1989 – for couples to undergo something called preimplantation genetic diagnosis. That would be where you go through the first part of a regular IVF cycle, you create embryos in vitro and those embryos are tested to see if they suffer from a genetic condition that runs in your family. That enables you to transfer to the woman’s womb only those embryos that are unaffected by that genetic condition.

Something that has been incredibly important in recent years in the regulation of new technologies is the importance of public consultation and public education campaigns. Rather than just having an expert regulator which makes the rules and decides what’s going to happen, there’s an increasingly important emphasis on making sure that the public have been consulted. That consultation isn’t just a survey but involves providing accessible information to the public to help them understand what’s being proposed, the issues, and the advantages and disadvantages.

Regulators increasingly recognise that public buy-in is really important. If the public is content that the regulator has a firm hand on what’s going on, it will allow science to progress without facing campaigns against this progress. There’s a lot of mutually beneficial elements to public consultation; it’s really helpful if the public are on board with any changes.

Some years ago, the UK changed its law to allow something called mitochondrial replacement therapy. This is where a woman has a genetic problem in her mitochondrial DNA – this is the outside shell of the egg, which doesn’t contain that many genes – and it can cause very serious disease. The question was, should it be possible to use the outer egg of a donor and the nucleus of the woman suffering from this condition in order to create an embryo?

Of course, you can imagine the “three-parent IVF” tabloid headlines for this, because you’re using the eggs of two women to create an embryo. But the UK’s regulator engaged in an intensive public education consultation to better inform people about what these diseases are and what the process involves. Obviously, not everybody would have seen that and not everybody would have felt properly consulted. Nonetheless, it was a good example of a difficult, complicated technical process that was explained very simply through cartoons and other accessible materials that were put into the public domain to tell people what was going on. While lots of the public wouldn’t have seen that consultation, it was still a really important effort to try to explain something to the wider public.