If you could predict the future, would you?
by Jamie Meyer
Maybe you can’t predict the future yourself, but Astrologers do. You could read into Astrology and maybe learn to do it for yourself if you wanted to. If you don’t believe in Astrology and consider it to be a bit of a pseudoscience, I want you to consider, are they not just analysing years and years’ worth of data, previous events, and people and making predictions based on previous patterns? Maybe it isn’t always accurate but there have been enough people born between February and March that are considered dreamers for people to now associate that characteristic with a Pisces. However, I am not trying to convince you of the validity of astrology. I am well aware that not all Pisces are dreamers, and not all dreamers are Pisces. Instead, I’d like to tell you about a much more accurate way of predicting the future in a much more useful way.
It is widely known that cancer is the leading cause of death globally. A major contribution to this is that there is a large percentage of treatment failure due to resistance. In a study on pancreatic cancer tumour cell lines the authors separated the tumours into specific subtypes and identified some of the relevant biomarkers associated with the different subtypes. Subtype-1 was the less severe subtype while subtype-2 presented as more severe. By integrating protein expression, mRNA transcription, DNA methylation and miRNA synthesis data the authors were able to build a machine learning algorithm that predicts the response of the tumour cells to different cancer drugs. This is done by identifying specific biomarkers associated with the respective subtypes that indicate what signalling pathways are affected and therefore would be an effective target for an anticancer drug.
The results showed that mTOR and kinase signalling was affected in subtype-1 and so these tumours could be responsive to anticancer treatments that target kinases, ion channels and membrane-pump proteins. Subtype-2 showed over-transcription in genes associated with transmembrane transporter and G-protein coupled receptors, as well as an increase in AGE-RAGE signalling and so these tumours could be treated targeting the inhibition of RAGE. They were able to classify the different subtypes with a 94% accuracy.
This is the field of precision medicine where we take large datasets of patients and experiments and look at their cell lines and their drug response data, so that in the future if we look at the biomarkers of patients, we can predict what drugs they will be responsive to, based on how similar cell lines have responded in the past. This will ultimately allow us to prescribe a tailored more accurate treatment regime with a higher chance of success.
So, I may have not convinced you of the validity of astrology, but I hope that you see the immense value of using predictive models in medicine, and the impact it will have in the future by reducing treatment failure with individually tailored regimens. The goal of precision medicine is to be able to tell someone with a disease that you must take X drug, for Y amount of time and then you will be cured with more accuracy than we say all taurus’s are stubborn.
Sinkala M, Mulder N, Martin D. Machine Learning and Network Analyses Reveal Disease Subtypes of Pancreatic Cancer and their Molecular Characteristics. Scientific Reports. 2020;10(1).