Take as prescribed — Dr. AI 💊
How AI is reshaping MedTech.
3 min readMay 13, 2021
By: John Cameron
TL;DR
With COVID-19 vaccine efforts still months away from potential fruition, researchers have looked for a more immediate approach to treat the virus. One particularly effective strategy that leverages the power of advanced AI algorithms and network science is drug repositioning. By having AI systems sift through and analyze billions of data points from clinical studies, lab reports, and scientific publications, the hope is to find existing alternatives already in use today that can be tailored to curb symptoms and provide safe, reliable options for practitioners and their patients.
The Breakdown
- As of October 12, 2020, over 1.08 million people worldwide have died from COVID-19, with total cases recorded in excess of 37 million
- As societies around the world continue to grapple with containing the virus and mitigating further damage and distress, the one question on everyone’s mind is: When will there be a cure?
- The general consensus from medical experts still remains that a vaccine will be deployed at some point in 2021, but in lieu of any current option, researchers have looked towards a technique called drug repurposing, aimed at utilizing existing drugs to treat novel, challenging diseases such as COVID-19 in far less time and research expenditure
- In a recent study, such an approach was used to determine the relationship between the COVID-19 host interactome (set of all molecular interactions in a cell) and various drugs, concluding with the suggestion of 16 possible repurposed drug candidates for the treatment of the virus
The Tech
- One of the ways researchers have looked to take on the task of further analyzing drug repositioning findings is through the use and optimization of AI models to uncover hidden patterns and evidence from biomedical data
- Companies ranging from multinational Pharmaceutical companies to budding startups already use IBM Watson (IBM’s suite of enterprise-ready AI services) to sort through enormous quantities of textual data (from clinical reports and lab data to scientific publications) to aid in use cases of drugs and their continued development
- The primary subset of AI favoured by researchers in their analysis revolves around a deep learning architecture, a sub-field of machine learning that emulates a human brain through the use of neural networks (comprised of layers upon layers of transformations and weights to manipulate input data and effectively synthesize it)
- The use of these more complex AI systems has proven to be far more accurate performance-wise compared to more conventional methods of Machine Learning such as logistic regressions
The Significance
- Between April and June of 2020, it is estimated by the International Labour Organization that almost 400 million full-time jobs have been lost, translating to a 10% landslide in income, equivalent to over CAD$4.6 trillion
- Due to the nature of COVID-19 as a severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), creating a new drug from scratch (otherwise known as a de-novo drug discovery) is extremely challenging and infeasible, making it increasingly important to look towards other strategies for handling the pandemic
- While vaccine development continues, the combined efforts of advanced AI algorithms and network medicine (a branch of network science that analyzes the connections between various conditions and seeks to find potential, existing use-cases for treatment) serve as a tangible, cost-effective mechanism for driving meaningful progress in the battle against COVID-19
