It’s been a while since I last wrote something personal on the blog. I have scheduled most of the daily blog posts from the past. The news of Deep Mind piqued me that figured out the protein folding (I am not linking it here as it has little relevance to us). There were claims and counterclaims (as usual) but yes, it hasn’t replaced the hard work of identifying them by traditional methods, but it still represents a significant advance in systems biology.
Here’s a counter claim:
However, despite some of the claims being made, we are not at the point where this AI tool can be used for drug discovery. For DeepMind’s structure predictions (111 in all), the average or root-mean-squared difference (RMSD) in atomic positions between the prediction and the actual structure is 1.6 Å (0.16 nm). That’s about the size of a bond-length.That sounds pretty good but it’s not clear from DeepMind’s announcement how that number is calculated. It might be calculated only by comparing the positions of the alpha-Carbon atoms in the protein backbone – a reasonable way to estimate the accuracy of the overall fold of the protein. Or, it might be calculated over all the atomic positions, a much more rigorous test. If it is the latter, then an RMSD of 1.6 Å is an even more impressive result.
Frankly, I understand little of the counterclaim here. The author agrees that it was a significant advance.
While all of the above is supposed to sound a note of caution to counter some of the more hyperbolic claims that have been heard in the media in recent days, I still want to emphasise my admiration for the achievements of the AlphaFold team. They have clearly made a very significant advance.
The success of these claims would depend on the open sourcing of the algorithms. Unless they open these to public scrutiny, we should take it with a handful of salt.
However, this prompted a very insightful mail from Dr Bryan from 33 charts. He’s not put the weekly links on the website, but it was related to this development cited above. He had expressed concern that the current “medical innovations” are not being done by physicians but “free range thinkers with infinite resources”.
While breathless this may sound (protein folding), it represents a free advertisement for Google’s Tensor Flow- a system of their machine learning platform. A scoop from Axios that Google also followed this is keen to develop its own chips (technically SoC’s) which would integrate the same in their platforms and run the system through the data centres and mobile phones. While this opens up a lot of privacy concerns, but the unification of the biology, algorithms and the technology is inevitable.
I cannot gaze into the crystal ball and predict the need for the physicians or how the medicine would look like in the next twenty years. However, there is a fundamental shift towards the corporatisation of healthcare as it represents a very profitable monopolisation of the “industry”. The focus of healthcare is not “prevention” but assessing and dealing with the complications of the lifestyle diseases. Individuals assume that they are prone from the effects where in consumption of mainstream carcinogens like alcohol and junk food is normalised.
The associated PR machinery around the “artificial intelligence” makes us acutely aware of the “fear of missing out”. Twitter stream is full of individuals with their grants proposals or conference papers that ultimately have a poor reflection or bearing on the clinical practises and represents institutional led “advancements”. The pace of “innovation” and “research papers” is an ever confusing string of statistics that still cannot improve upon the survival of the patients as they lack both objectivity and quantification.
Dr Bryan also highlighted that physicians are constrained with the workflows that have a little bearing towards pushing for innovation. I agree completely. The innovation engine has shifted away towards the corporates who are flush with funds and a PR machinery that can outmatch any blog!
What is the way out? It is impossible to expect physicians jumping in with the idea of “protected time for research” as the responsibilities for attending to patients becomes paramount. The way out is to make the existing interaction efficient with the judicious use of technology and staying away from social media (of course!) Automation tools and pursuing the goal of “inbox zero” with provide a semblance of sanity towards the chaotic workspaces. There are numerous ways wherein workspaces can be easily adapted and adopted for the efficiency. Sometimes common sense can easily trump the artificial intelligence too.