Medicine has seen many ways Artificial Intelligence (AI) disrupts patients’ outcomes. AI also transforms workflows for medical professionals and unlocks new opportunities in terms of diagnostics and treatment. Currently, pharmaceutical businesses try to leverage AI for faster drug discovery. Let’s see if we may expect a revolution in drug discovery due to the most common type of AI – machine learning.

Even though there are plenty of different drugs, they are typically specifically combined molecules that are aimed at a target disease-related protein. What is needed to discover a new drug is a screening library of molecules to find the right one that will cure a medical condition. In that vein, drug discovery is expensive and time-consuming.

Several years ago, a survey of 10 pharmaceutical manufacturers showed that their spending for testing new drugs may total up to $2.8 billion. At the same time, these enormous costs do not guarantee the launch of a medicinal product in the market. In the light of this, drug development assisted by machine learning is moving into the top gear of innovative healthcare solutions and has promising examples.

One such emerging technology is the NanoTemper Prometheus system. Hailed as the new gold standard of multi-parameter characterization, researchers are able to better optimize drug discovery with dramatically superior results when compared to more antiquated R&D tools. Offering high-quality data and boasting the ability to run several measurements – even with identical samples – user error is minimized and output is maximized. In turn, the highly sensitive and advanced system is able to allow pharmaceutical scientists to determine which candidates are ideal in any future trials. This means more efficacious treatments and fewer delays when patenting novel medications.

Absci, a U.S.-based synthetic biology company, runs an AI-assisted platform for protein discovery and the creation of new biotherapeutics. Absci’s machine learning tool allows the company to reduce the timeframe of drug discovery from years to months or even weeks. According to Absci’s CEO, Absci’s acquisition of a deep learning company “represents the perfect synergy of groundbreaking synthetic biology and cutting-edge deep learning AI to create in silico predictive protein drug design and cell line development capabilities with the potential to completely change the paradigm of biopharmaceutical discovery and development”.

Insilico Medicine, an AI-based platform for drug development from Hong Kong, announced that it had created a novel drug to cure idiopathic pulmonary fibrosis. The company’s AI-driven tools identified 20 possible targets, picked up one, and designed a treatment that proved successful in animals. At the moment, the company is going to start human trials.

Talking more generally, the AI share in the drug discovery business is reported to be $230 million in 2021 with an estimated increase of up to $ 4 billion by 2031. Here is why. Due to genomic sequencing (a technology that analyses a virus sample in comparison to other cases), the amount of biological data is unprecedentedly high while generation thereof is much more cost-efficient. Machine learning is able to identify how a disease works based on these multiple patterns and generate novel drug compounds not available before. What is also important is that machine learning is able to predict the properties of a drug, i.e. medicinal compounds with ineffective properties are not chosen. It all goes without saying that lab professionals can save their time and avoid repetitive analysis of samples and images.

“Predictive models are central to our work. These are statistical models that predict whether a compound idea – a not-yet-synthesized molecule – will produce the desired activity. The technologies we’re using mostly relate to machine learning,” says Friedrich Rippmann, Director, Computational Chemistry & Biology at Merck.

In other words, machine learning and AI tools could be incorporated into all stages of drug manufacturing: prediction of drug-protein interactions, synthesis, target discovery, determining drug efficacy, optimization of bioactivities of molecules, clinical trials, etc. What is highly appreciated about machine learning is that it recognizes that individual variations are also considered in the analysis of possible drug compounds to target relevant pathogens.

Conclusion

Machine learning unleashes the potential of enormous amounts of data in many spheres and is being deployed for smarter and faster drug discovery. As a result, pharmaceutical manufacturers get a better understanding of the code of life and disease and save time and costs. A thing to be flagged is that advancements in drug discovery should be a cooperative process with the combination of biology and machine learning.