Why will the boon of Artificial Intelligence endure in Pharma/Biotech industry?

Let's go back to our days of algebra and consider there was a person named X. 

X was 82 years older and was suffering from an aggressive form of blood cancer. He already had to go through six courses of chemotherapy, but the wait to be cured was still a far cry. His saviors, the doctors, whom, in his poem, D. H. Lawrence has ornamented as “ the ones who work with science and arts, and knows the secrets of the mill”, were systematically evaluating a roster of conventional cancer medications. Their aim was to discover an effective solution, methodically eliminating options one by one." Needless to say, the drugs were doing everything but cure X.

The hopes were dying, and they were taking X with them, slowly, on a bed adorned with the persistent presence of suffering.

But then the doctors, knowing that there was nothing to lose, decided to embrace a boon that, until then, had remained just an elusive dream, hovering ever closer to the realm of reality: the boon of Artificial intelligence (AI). 

The doctors enrolled X on a trial set up by the Medical University of Vienna, where a new technology developed by a UK-based company called Exscientia was being tested. The aim of this project was to develop precise drugs for the patients. The researchers took samples from X (both normal and cancer cells), exposed them to various cocktails of drugs, and then used machine-learning-based models. The goal was to use different drugs and identify which worked, testing various treatments simultaneously. The end result was amazing. A cancer drug produced by Johnson & Johnson was discovered to be effective, a treatment option previously unexplored by doctors due to prior trials indicating its ineffectiveness in addressing his specific cancer type. 

X’s cancer was gone eventually. 

Image generated using https://wepik.com/

So basically what is AI and what does it do ?? AI is like a smart computer program that can learn and make decisions on its own. It's kind of like a robot brain. AI can do a lot of different things, examples include: recognizing your voice when you talk to your phone, suggesting what movie to watch on streaming services, playing chess or other games really well, or even helping self-driving cars make decisions to stay safe on the road. It learns from data and can do tasks that normally require human intelligence, like understanding language, solving problems, and recognizing patterns. In simple terms, AI is like a computer that can think and learn to do tasks without being explicitly programmed for each specific task.

AI is indeed helping the Biotech/Pharma industry solve human health problems by improving drug discovery and development processes, among other applications. And, its potential continues to generate a lot of excitement. Cutting through the decades, it is fair to say that AI has taken the biotech industry by storm, and the hype surrounding it has led to numerous AI startup companies coming into the pharmaceutical industry. The vision is to use AI to make drug discovery faster and cheaper. By predicting how potential drugs might behave in the body and discarding dead-end compounds before they leave the computer, machine-learning models can reduce the need for painstaking lab work. Since, for many diseases, the approved drugs are known to create severe side effects, the need for new drugs is always increasing. 

The success that AI is bringing to the drug discovery process may not be enormous, but promising.  For instance, it helped scientists discover a new antibiotic called Abaucin, which can kill Acinetobacter baumannii, a superbug that causes fever, chills, and vomiting.  AI rapidly screened 7,500 molecules that inhibit A. baumannii, and in one and a half hours, it managed to narrow down 250 potential compounds, after which Abaucin was found to be the most potent one. INS018_055, the first fully generative AI drug to treat idiopathic pulmonary fibrosis,  has recently reached Phase II clinical trials. Various other drugs, partially generated by AI, are also in different phases of clinical trials. 

On the other hand, the journey was not always a cake in the walk. A handful of molecules created by artificial intelligence have failed trials or been deprioritised. The list includes a cancer drug candidate, EXS-21546, developed by Exscientia, BEN-2293, a topical out of London-based BenevolentAI, a drug candidate developed by Benevolent AI, etc. In a time when only 5-10% of drugs that progress to human clinical trials receive approval, the challenges faced by AI-based drug development serve as a stark reminder that AI tools cannot ensure automatic success in drug development.

Will this stumble halt the AI-based approaches? The answer is a clear and resolute no. The advancement of technologies is gathering large piles of data covering various minute details about diseases, how they spread, how they tend to evade the host's defence mechanism, how they became resistant and many more. Also, the amalgamation of AI with other branches of science will help it investigate different disease aspects. For instance, in many cancer patients, signalling pathway reactivation makes them drug-resistant. Utilising mathematical modelling,  Boris and colleagues have demonstrated that achieving a full reactivation or increase in the steady-state activity of a pathway across a range of inhibitor doses necessitates the presence of two or more feedforward connection routes originating from an inhibited upstream protein to the final pathway output protein. When a drug relieves negative feedback by promoting drug-induced kinase dimerisation, this action extends the potential for paradoxical activation. In cases where the drug does not induce dimerisation, alleviating negative feedback merely results in a temporary surge in pathway activity. Without a minimum of two feedforward routes to the output protein, negative feedback regulators can only induce a transient spike in the pre-existing output signalling rather than facilitating an enhancement of steady-state output activity. These findings hold significant implications for developing strategies to overcome drug resistance. An AI-based approach, integrating such minute details, will surely help to overcome the aspects of drug resistance in many diseases, including cancer.  

Figure: AI in healthcare. Image source: Biorender. 

To summarise, AI stands as a beacon of promise in clinical research, guiding the way towards a future brimming with possibilities. The enduring boon of AI lies in its ability to continually adapt and evolve, making it a transformative force in our rapidly changing world. AI systems excel at automating tasks and providing insights and possess the remarkable capacity to learn and improve over time. This means that as AI technologies advance, they become more adept at understanding complex human language, recognising patterns in data, and even making autonomous decisions. As AI's capabilities expand and its applications diversify, its enduring significance is grounded in its potential to revolutionise how we work, learn, and interact with technology, paving the way for a future where AI remains a steadfast ally in our quest for progress and understanding and therefore, it is safe to say that, AI will endure, and a disease-free world might be on the cards in the foreseeable future. 

References: 

1. https://www.technologyreview.com/2023/02/15/1067904/ai-automation-drug-development/

2. https://endpts.com/first-ai-designed-drugs-fall-short-in-the-clinic-following-years-of-hype/

3. Kholodenko, Boris N., et al. "A systematic analysis of signaling reactivation and drug resistance." Cell reports 35.8 (2021).