>> Artificial Intelligence: A New Era of Data-Driven Medicine

Artificial Intelligence: A New Era of Data-Driven Medicine

On April 11, 2019, Tags Forecasting

Artificial Intelligence (AI) is the ability of computer systems to perform tasks that normally require human intelligence including running automated algorithms, machine-learning, deep learning and natural language processing. Over the past few years, AI has been gaining traction in the pharmaceutical and healthcare industries with many companies entering into partnerships or acquiring technologies in order bolster their AI capabilities (Fig.1). AI has a range of applications within the pharmaceutical and healthcare industries including optimising the drug development process, improving diagnostics and enhancing the likelihood of treatment success.

 Ai

Figure 1: Selected AI partnerships and acquisitions of the Top 10 pharmaceutical companies

 

The cost of developing a new drug is estimated at over $2 billion, much of which is effectively wasted on failed drug candidates, with only around 10% of investigational drug products ever making it to market. Therefore, many of the top pharmaceutical companies are now utilising AI in an effort to significantly reduce this spend. AI can be employed at various stages of the drug development process including identifying target disease mechanisms, selecting and designing potential drug candidates and optimising clinical trial parameters and logistics to maximise the chances of success.

 

Both Pfizer and Novartis have licenced the use of IBM-Watson’s “Watson for Drug Discovery” system, a cloud-based technology which utilises machine learning to analyse data from 25 million Medline abstracts, over 1 million full-text medical journal articles and 4 million patents, far exceeding the analytical capabilities of a human research scientist. It is hoped that this technology will be able to find correlations between genes, drugs and disease states so that new drug candidates and target mechanisms can be identified. Atomwise has also developed an AI approach to lead generation and optimisation and have partnered with a number of pharmaceutical companies including Pfizer, Merck and AbbVie. Their technology is based on convolutional neural networks, the same technology that enables facial recognition and self-driving cars and has been designed to extract data from millions of experimental affinity measurements and thousands of protein structures in order to predict the binding of small molecules to proteins, thereby identifying potential drug candidates. These pharma-AI partnerships have already begun to yield results, with Exscientia recently delivering the first molecule identified through its Centaur Chemist platform, a system designed to automatically design new compounds and prioritise them for synthesis based on expected potency, selectivity and pharmacokinetic properties. This molecule was delivered to GSK as part of their partnership to identify new drug candidates for COPD.

 

In addition to drug discovery and design applications, AI can also be utilised for clinical trial design and logistics. Predictive analysis and machine learning can be employed to analyse large amounts of data in order to identify the correct patients for enrolment and assess the likelihood of adherence and treatment success. Several companies have partnered with AiCure, a company who have developed a technology platform which interacts with patients, capturing visual and audio behavioural data which can be used to optimise their participation in clinical trials. AbbVie recently employed this technology as part of a Phase II trial in schizophrenia, a disease in which poor adherence to medication is a significant issue. Using smartphones, which visually confirmed the ingestion of medication, AiCure’s technology was able to significantly improve medication adherence of the trial participants. It is easy to see how this can be translated into an increased likelihood of clinical trial success, especially in diseases where adherence to medication is problematic.

 

Optimisation of clinical trials is also the objective of Novartis’ partnership with Quantum Black. As part of the collaboration, they have developed a programme called Nerve Line, which combines clinical trial data from Novartis’ internal systems, applying machine learning and advanced data analytics to monitor and predict trial enrolment, cost and quality. In doing this Novartis have indicated that they have reduced patient enrolment times in pilot studies by 10-15%.  

 

Diagnostic applications of AI are also gathering attention. Google’s DeepMind Health is developing its machine learning technology to detect differences between healthy tissues and cancer. The UK’s NHS is also piloting a number of AI-directed approaches to diagnosis including Kheiron Medical’s AI algorithm to diagnose breast cancer. The algorithm has been trained to recognise cancer using half a million scans and is reported to be more accurate than a human radiologist. The NHS is also trialling the ‘C the Signs’ app, which uses AI linked to a breadth of information including NICE guidelines and other clinical evidence enabling GPs to quickly check combinations of signs, symptoms and risk factors to identify patients at risk of cancer.

 

The use of AI in diagnosis is also being explored by Winterlight Labs who, in collaboration with Janssen, are aiming to develop technology that extracts features from speech and language collected in clinical trials, using this to develop a digital biomarker which identifies patients in the early stages of Alzheimer’s Disease. These patients can then be treated earlier with the objective of slowing the progression of the disease.

 

IBM-Watson are also using AI to improve treatment outcomes with their IBM-Watson for Oncology platform which analyses patients’ medical information alongside clinical guidelines, and information extracted from journal articles and other sources in order to recommend the treatment with the best chances of success. GNS Healthcare has also developed its Reverse Engineering and Forward Simulation (REFS) platform which, in partnership with Amgen, will be used to analyse data from clinical trials to identify factors that drive treatment response in metastatic colorectal cancer.

 

In this short review, it is easy to see the breadth of potential applications that AI has in the pharmaceutical and healthcare industries. Advocates of the technology believe that AI will revolutionise the way in which new drugs are developed and treatment decision are made. However, AI is still in its infancy, with most projects in developmental or pilot stages. Therefore, it remains to be seen how successful the use of AI will ultimately be in the pharmaceutical and healthcare industries but the activities over the coming years should be closely monitored.

 

Jo Adams

Senior Analyst

New Call-to-action

Subscribe to Email Updates

New call-to-action
RX Brochure CTA Large

Recent Posts

Free Brochure Epiomic