How AI/ML Steered Major Innovations in Medicine & Healthcare In 2020

Of all the innovations in artificial intelligence and machine learning space this year, the most significant ones have turned out to be in the healthcare and medicine field, the credit of which could be given to the unprecedented pandemic situation around the world. In this article, we discover some of the major breakthroughs in 2020.

AlphaFold From DeepMind

Termed as a groundbreaking innovation, DeepMind’s AI system AlphaFold was developed to present a solution to the 50-year-old grand challenge of determining the protein structure, also referred to as ‘protein folding problem’. This system demonstrated high levels of accuracy in predicting the 3D structure of a protein. By bringing about significant progress over the core challenges in biology, this system paves the way for disease understanding and drug discovery, even in case of COVID-19, among other fields.

In theory, this problem was first posed by Nobel laureate Christian Anfinsen in 1972. He had then proposed that a protein’s amino acid sequence can be used to fully determine its structure. What followed was a 50-year quest to practically realise the same. One of the major hurdles, however, was the expensive and time-consuming nature of predicting this structure solely based on its one-dimensional amino acid sequence.

Unlike previous attempts that used methods ranging from Cryo-electron microscopy, Nuclear magnetic resonance, or X-Ray crystallography costing millions of dollars, researchers at DeepMind worked on its AI system AlphaFold to solve the challenge. This system has been trained on the sequences and structures of hundreds of thousands of proteins. This solution by DeepMind was finally recognised by CASP as a breakthrough.

In Pursuit Of COVID Vaccine Discovery

Machine learning-based techniques were massively favoured for the rapid discovery of vaccination for COVID-19. One of the most important applications was performed by the researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL). The scientists used a combinational machine learning system to introduce OptiVax, a model that introduces methods for designing new peptide vaccines and can be used for evaluating and augmenting existing vaccines.

In OptiVax, peptides are scored through machine learning by their ability to elicit an immune response which is then selected to maximise population coverage of who could benefit most from the vaccine.

Moderna, which is one of the leading institutes in the run-up to providing COVID-19 vaccine, used a combination of AWS services, analytics, machine learning techniques to accelerate the design and production cycle of the vaccine. In fact, its extensive usage of ML-based techniques is said to have given it an edge over other pharmaceutical companies and institutions.

Apart from vaccine discovery, AI and ML techniques were found to be especially effective in enforcing social distancing and other preventive measures in place to contain the spread of the deadly virus.

NVIDIA’s MONAI Framework

To accelerate artificial intelligent-based solutions in the healthcare industry, NVIDIA, earlier this month, launched Medical Open Network for AI (MONAI). A PyTorch-based framework, MONAI was first announced in April 2020 and has been in use, since then, by several leading healthcare institutions. With the new announcement, NVIDIA plans to make MONAI ready for production with the upcoming release of the company’s Clara application framework for AI-enabled healthcare and life sciences.

The MONAI framework uses AI for medical imaging through industry-specific data-handling, reproducible reference implementations of state-of-the-art approaches, and high-performance training workflows.

MONAI was released as part of the updated NVIDIA Clara offering. This new framework comes with over 20-pre trained models, including the ones developed for Covid pandemic.

AI Innovations For Mental Health

The extended period of lockdown, tectonic shift in work cultures, and uncertainty around life during the Covid-19 pandemic sure had a disastrous effect on the mental health of people around the world. Several studies and researches have also corroborated that the general mental spiralled downwards during this period, all across the world.

Taking cognisance of the same, a joint study between the researchers at MIT and Harvard University used natural language processing to monitor the impact of the pandemic on people’s mental health for better assessment, and possible remediation for such mental health scares. 

For this, the scientists collated the Reddit posts of over 800,000 users between 2018 and 2020 and applied NLP techniques such as trend analysis, supervised and unsupervised learning in order to characterise the changes in the language used by mental health support groups.

Google’s parent company Alphabet Inc.’s R&D lab X made Project Amber freely available for the mental health community to build upon. Project Amber is an early-stage mental health project compiled by a team of neuroscientists, hardware and software engineers. Its findings are based on medtech technology that combines machine learning and AI techniques with electroencephalography (EEG) to measure electrical activity in the brain for diagnosing depression.

This year also saw a rise both in invention and usage of mental health and companionship chatbots to rescue people from bouts of loneliness and seclusion.