How AI is leveraging DNA to fuel modern medicine
Few areas in science benefit as much from advancements in processing power as genomics. From a computational standpoint, the amount of time and resources needed to fully sequence even a simple organism’s DNA is immense, and that’s just half the job: for that sequencing to be useful, scientists need to analyze its massive outputs of data and apply a variety of math techniques to properly map their work into practical information.
But while brute-force data processing is still of major relevance in the field, those CPU advancements also allow for a smarter approach, and one that the discipline welcomes wholeheartedly: AI-powered DNA sequencing.
With the correct AI algorithms, researchers not only enjoy faster processing times but more precise ones. Needless to say, in the last few years, machine learning has enabled this deeply interdisciplinary field to advance by leaps and bounds. And at every turn, a healthcare revolution follows in the wake.
AI has been used to accomplish a variety of clinical genomics tasks, including variant calling and annotation, variant impact prediction, phenotype-to-genotype mapping, and more. Let’s have a look at the main applications of AI in the field, how genomics analysis is impacted by it, and the ways it affects us directly.
The major uses of AI in genomics
Drug discovery & screening
A drug is, fundamentally, a chemical product that interacts with the body’s own chemistry aiming to positively alter it. For that to happen, however, pharmaceuticals need a comprehensive understanding of our biological processes.
By vastly expanding our knowledge to the inner workings of human bodies on a molecular scale, a pharmacogenomic AI makes for more efficient drug research and discovery, while also making clinical trials more predictable, and thus, safer.
Predictive genetic testing & diagnostics
While some diseases like flu or malaria are caused by external factors, many others are in some way hard-coded into our genes. Through genomics, predisposition for those hereditary conditions (for instance, some cancers) can be discovered well before symptoms appear and treated accordingly — and, in the near future, even “edited out” of our systems.
In cases where the disease isn’t preventable or treatable before symptoms arise, a clearer picture of the patient still makes for a more effective diagnosis. Rare syndromes, often genetic, that for decades have eluded doctors become more manageable when correctly identified.
While genetic testing is usually done by specialized healthcare providers, at-home testing has been picking up steam for the past few years. Ancestry tests, for example, can be safely and reliably ordered by mail and executed directly by the consumer.
The same principle applies to AI-powered genomics: instead of your genealogy tree, a small sample of your DNA can now provide you with thorough results about your genetic traits, predispositions, and general health.
Personalized treatment is perhaps where genomics shines the brightest. Since every genome is unique, mapping an individual’s genetic code allows for the ultimate customization in combating illnesses.
When paired with the aforementioned drug discovery, genomics can provide tailor-made treatment for a patient’s condition. Cancers can be sequenced, analyzed, and tackled with personalized care fueled by advanced AI technology, thereby drastically improving healing odds.
All in all, genomics’ possible applications in medicine are as vast as medicine itself. For an in-depth look, we’ve handpicked some breakthrough methods currently being pioneered by innovative companies, some of which we’ve had the pleasure to partner with.
Key players and major innovators
Recursion Pharma: Drug discovery
This American drug manufacturer, boasting over a dozen VC companies as partners, combines AI technology and chemistry research to create frameworks enabling personally crafted drugs.
By prioritizing data, their AI technology and drug development coexist in the same hierarchy as two parts of a whole, automating much of the drug screening process and enabling continuous and adaptable machine learning over many iterations.
Ezra: Cancer diagnosis
The startup, which raised $18M in its Series A, has a straightforward purpose — to detect cancer as early as possible. It is a known fact that early diagnosis is the best prevention against the disease, and Ezra’s solution (and main product) comes as a single session, full-body MRI scan able to detect up to 13 types of cancer.
Behind the screening, however, is a complex myriad of AI technologies to facilitate the radiologists’ work. Their AI technology focuses on imagery, improving the efficiency of diagnosis through machine recognition of visual patterns in tumors.
Gauss Surgical: Blood tests
Gauss Surgical specializes in the prevention of maternal mortality by monitoring blood loss during delivery and postpartum. Traditionally, hemorrhage protocols are checked and followed through human-eye estimations of how much blood has been lost, a system too prone to errors.
Gauss’s platform, the AI-powered Triton app, instead reads blood levels in real time by visually recognizing blood properties through an iPhone camera, resulting in a 300% increase in hemorrhage detection.
Pexxi: Tailored contraception
The one-size-fits-all approach of contraceptive pills can be maddening for many women, who often end up finding theirs by trial and error, all while dealing with a plethora of — and in some cases, unavoidable — side effects.
UK-based Pexxi bypasses this by individually analyzing their clients’ hormones and genetic background, which can be done via at-home test kits. After recommending the ideal match, the company keeps monitoring how the body reacts to the pill through stress, energy, body mass, skin quality, and mental health levels.
Genomic Prediction: Predictive genetic testing
The world’s leading clinical laboratory, Genomic Prediction supports parents seeking in vitro fertilization by testing the embryos’ genome to improve the procedure’s success rate and reduce the risk of potential diseases.
iTechArt’s partner for over two years, we have developed a robust network of web platforms aimed at enhancing the storage and analysis of their genetic data, alongside solutions to manage requests from and reports to their clients.
What does the future bring?
It might take some time before AI-powered genomics become as immediate and personal as insulin monitors, yet it’s easy to forget that same sequencing that took the Human Genome Project 13 years and billions of dollars to execute in the 90s can be done nowadays in less than a day and cost a fraction of a smartphone.
Nonetheless, genomics, and the power brought to it by AI development, is an outstanding example of the uncanny progress made possible by technology. From small quality-of-life improvements to treatments — and potential cures — for conditions we’ve struggled with since the dawn of time, the possibilities are as boundless as these machines’ capacity to learn.
And that is a reassuring prospect indeed.