We look at three pharmaceutical and life science trends for 2021 – drug discovery, clinical trials, and personalised medicine – and what is driving their transformation.
This article examines
- Can artificial intelligence rationalise the R&D process?
- Almost all clinical trials fail. Why is this and can artificial intelligence turn it around?
- What is personalised medicine, and why do experts say it could be the future of medicine?
Drug Discovery
Despite its impressive response to the COVID-19 pandemic, the pharmaceutical industry is experiencing significant challenges. Drug discovery is becoming progressively more expensive – currently, it costs on average $2.6 billion to develop a single drug – and the process takes on average thirteen years. Very few new drugs are coming to market, and the ROI in the industry has declined (Properzi, Taylor, & Steedman, 2019).
New, effective, and less costly ways of drug development are required urgently, so it is no surprise that drug development programmes are utilising new tools based on artificial intelligence (AI). We already see significant benefits. AI is playing a critical role in discovering drugs for complex diseases such as Parkinson’s, Alzheimer’s, pancreatic cancer and is also on the cutting edge of personalised medicine. Over the coming years, the market for AI in drug discovery is poised to grow (Fior Markets, 2020).
The emerging trend is for AI-based drug discovery startups to partner with established multinational pharmaceutical companies. Examples include Cyclica and Bayer who are applying AI to peptide drug discovery and Bayer and Merck who are using AI to target chronic thromboembolic pulmonary hypertension.
Although to date AI discovered drugs have not made it through the later stages of clinical trials, it is likely only a matter of time. However, a field where AI has already made a significant impact is the repurposing of existing drugs. For instance, AI has helped identify Baricitinib, a rheumatoid arthritis drug, and Remdesivir as possible treatments for COVID-19.
The trend for pharmaceutical companies working together with innovative AI businesses to streamline the drug discovery process is set to continue. The old ways of drug discovery take too long and are too expensive. Not only will AI assist in bringing new drugs to market, drugs should also become cheaper.
Clinical trials
The pandemic has brought to the forefront the speed at which a large stage trial can be progressed. Traditionally, clinical trials are ponderous and expensive, and the failure rate is also high; up to 90% of them fail (Wong, Siah, & Lo, 2018).
AI and machine learning are beginning to make a significant impact on how researchers and health care professionals implement clinical trials, a trend that will continue to grow and yield positive results into the foreseeable future. Some critical areas are highlighted below.
Patient selection
Currently patient selection consumes around one third of the total time it takes to complete a phase 3 clinical trial and almost a third of trials fail because of patient recruitment problems. When the wrong patients are selected, the trial is doomed to fail. Trials can be onerous on the patient, which increases the risk of them dropping out before the trial is complete. By analysing vast amounts of data, AI can improve patient selection and retention. It can also make trials easier on the patient, for instance remote monitoring reduces the number of visits patients must make to the clinic. An example is IBM’s Watson for Clinical Trial Matching which has demonstrated an 80 per cent increase in enrolment to systemic therapy clinical trials for breast cancer (Fassbender 2018).
Digital biomarkers
Digital biomarkers can be used remotely to monitor patient health eliminating the need for patients to attend clinic. For instance, a smart phone tapping test app has been developed to remotely monitor the condition of patients taking part in a Parkinson’s treatment clinical trial.
Mining and analysis of unstructured data
AI empowered image analysis can rapidly detect and monitor some clinical conditions as well as and sometimes faster and more successfully than professional medical staff (Bresnick, 2018). For instance, deep learning retinal image analysis can detect chronic kidney disease, and deep learning algorithms can rapidly analyse mammograms and MRI images.
Personalised Medicine
We have briefly mentioned personalised medicine as a future trend in drug discovery, but here we look at it more deeply in terms of how it incorporates machine learning and artificial intelligence. By personalised medicine, we mean tailoring treatments to specific individuals, and there is nothing new in the concept. However, machine learning and AI in combination with massive real-world genetic and medical data repositories is now taking it to a higher level. Possibly it could be the future of medicine.
To date, progress has been impressive. Taking cancer treatment as an example, traditionally oncologists make clinical decisions using on-average data. In 2019 Turing Research created a machine learning algorithm that analyses a patient’s clinical history which it uses to predict the patient’s disease trajectory. It recommends the most effective treatments, locates patients with similar presentations, and responds to the way in which the disease progresses.
The industry trend is the forging of new partnerships between healthcare organisations and technology and AI companies. For instance, Pfizer and BMS have partnered with the AI startup Concerto HealthAI to work on precision oncology; Roche has acquired Flatiron Health to gain clinical insights by combining clinical and genomic data; Novartis and Microsoft are using AI to find new approaches to personalised medicine for macular degeneration. The big three tech companies, IBM, Google, and Microsoft, are all trailblazing in the biotechnology sector.
AI drives the future of pharmaceutical & life sciences
Across all areas of innovation from drug discovery, clinical trials and personalised medicine, the future trend is driven primarily by AI and machine learning. Additional examples include protein folding and advanced diagnostics. The COVID-19 pandemic demonstrated the urgency in responding quickly and decisively to rapidly developing healthcare emergencies, and technology is providing us with ever more sophisticated tools to help us do so. We are getting reasonably good at it too. We will need to be.
References
Fassbender, M (2018), AI drives 80% increase in clinical trial enrolment, Outsourcing Pharma, HTTPS://WWW.OUTSOURCING-PHARMA.COM/ARTICLE/2018/03/26/AI-DRIVES-80-INCREASE-IN-CLINICAL-TRIAL-ENROLLMENT
Fior Markets, (2020) Artificial Intelligence (AI) in Drug Discovery Market. (2020, June). Retrieved September 26, 2020, from https://www.fiormarkets.com/report/artificial-intelligence-a-i-in-drug-discovery-market-by-418371.html
Properzi, F., Taylor, K., & Steedman, M. (2019, November 07). Intelligent drug discovery – Powered by AI. Retrieved September 04, 2020, from https://www2.deloitte.com/us/en/insights/industry/life-sciences/artificial-intelligence-biopharma-intelligent-drug-discovery.html
Wong, C. H., Siah, K. W., & Lo, A. W. (2018), Estimation of clinical trial success rates and related parameters, Biostatistics, Volume 20, Issue 2, April 2019, Pages 273–286, https://doi.org/10.1093/biostatistics/kxx069
Bresnick, J. (2018, November 05). Top 5 Use Cases for Artificial Intelligence in Medical Imaging. Retrieved January 18, 2021, from https://healthitanalytics.com/news/top-5-use-cases-for-artificial-intelligence-in-medical-imagin
