They’re two of the biggest buzzwords getting professionals excited across pharma – Anjali Shukla explores the ways in which AI and big data are set to galvanise the industry.
Even if you haven’t been paying attention, it is hard to miss the buzz around artificial intelligence or deep learning and how the technology is only steps away from changing lives forever. The optimism is especially significant in the pharmaceutical and healthcare industry as researchers continue to look out for cures for complex diseases.
Even though we are nowhere potentially close to understanding the cumulative length, breadth and scope of the utilisation for deep learning, that hasn’t stopped industry pundits from making predictions about the transformative nature of the technology.
Researchers and medical professionals are now able to increasingly use the technology as a supplement to add to their own training and experiences while treating patients. The technology has proven to be efficient in mining pertinent data from a vast database and pick out patterns in the data set, as well as calculate statistics at a much more rapid pace than humanly possible. That is the reason a number of tech giants are collaborating with drug makers to mine patient data for better pattern recognition in a bid to gain better understanding of diseases and hopefully find better treatment options.
As far as healthcare is concerned, artificial intelligence has the potential to generate data, make predictions, and offer detailed insights and solutions going forward. That may have been the impetus behind Pfizer’s move to partner a pharmaceutical technology firm to develop a drug discovery platform powered by artificial intelligence. The pharma giant’s collaboration with XtalPi will develop molecular modelling software that can be applied to drug-like small molecules.
“The XtalPi collaboration is an opportunity to enhance our computational modelling capabilities”, says Charlotte Allerton, Pfizer’s Head of Medicine Design. “We are looking forward to potentially utilising new tools to increase our effectiveness in small molecule drug discovery and development.”
The software platform will combine quantum mechanics and machine learning algorithms with cloud computing to enable the prediction of pharmaceutical properties relevant for drug discovery and development, the company has said.
And Pfizer is not the only one playing that game, as numbers show. Global business value derived from artificial intelligence is forecast to hit a total USD 1.2 trillion in 2018, up 70% from a year earlier, with AI-derived business value expected to more than treble to reach $3.9 trillion in 2022.
The future of drug discovery
An obvious area where AI could become invaluable is clinical trials, which are not only expensive but are also a long-enduring process seeking results. These results do not necessarily turn out to be positive. On average, it takes researchers nearly a-decade-and-a-half to develop a new therapy, with costs running into the millions. So, when the trials fail it is not only a loss in terms of money, but also of valuable time.
Swiss drug maker Roche this year announced that its combination study with immunotherapy drug Tecentriq to treat metastatic colorectal cancer failed in late-stage trials. The therapy did not demonstrate efficacy in over 95% of patients, the company said.
Roche is currently running extensive clinical trials for Tecentriq with multiple late-stage trials for various cancers including lung, kidney, skin, breast, colorectal, prostate, ovarian, bladder, blood, liver and head and neck cancers.
And Roche is not alone. Bristol-Myers Squibb has also had to end late-stage trials for Incyte Pharma’s IDO drug following failed Phase 3 studies.
Other than draining time and capital for the drug firms, disappointing clinical trials have also historically resulted in eroding the market capitalisation for the companies. With drug trials accounting for nearly 40% of the total research outlays for pharma companies, AI could become a potent tool in shielding losses from disappointing trials, going forward.
Over the past years, researchers have predicted the use of AI algorithms can potentially dramatically cut down the overall time required for a new drug development by boosting the overall process of vast data crunching.
Various research programmes in collaboration with healthcare providers and tech firms are being devised and run around the world to collect and process genetic data and find meaningful actionable patterns for medical research.
The idea is to have AI algorithms help speed up the overall workflow through the trials as well as weed out any errors in diagnosis. Going forward, the algorithms are expected to aid the precise decision-making process and lead to targeted outcomes.
Experts have forecast that AI technology, along with big data, can potentially increase efficiency in drug trials. Data-driven trials have the potential to refine the overall trial process and ensure better accuracy in data collection. In addition, data collation via various phone applications and other medical devices can help bring down the cost of running clinical trials.
Improved mechanics of data collection also means the overall quality of data will also more reliable, which in turn will allow for better diagnosis and treatments. The use of wearable technologies has made possible a round-the-clock screening of several biometric signals like blood pressure, sleep patterns and blood pressure. As wearable technologies gain popularity, collecting precise real-time patient data becomes much easier. These can help monitor the efficacy of the drug, as well as the progression of the disease.
Data: The magic pill
Collecting and analysing a large amount of data to establish actionable patterns is clearly set to pave the way for not only more efficient clinical trial protocols but are also potentially being viewed as means of finding precise treatments going forward.
A recent study conducted by pharma giant Bayer evaluated the ability of animal studies to predict human safety. It analysed toxicity of drugs for humans and animals using big data. It involved 3,290 approved drugs for which over 1.6 million adverse events were reported for both humans and animals over a period of more than 70 years. The adverse events were reported with the US Food and Drug Administration and European Medicines Agency.
The results found some animal tests to be more predictive of human response than others, depending on the species and symptoms being reported.
The study underscores the potential of big data analysis to establish efficient trials and better treatments and therapies in the end.
Efforts like this are an encouraging sign that advanced analytics can provide researchers more in-depth insights into therapies and their overall progressive impact on the patients. Additionally, deep learning can also boost the process of developing more precise endpoints for treatments.
Given the current understanding of the capabilities of the AI algorithms, the use of AI to analyse big data can potentially help drive better understanding of the diseases as well as possibly enable the development of newer and more efficient treatment options. In addition, the technology can be used to advance the development of precision or customised medical care by identifying and categorising the impact of different medication on patients.
While the debate on privacy, information control and data security will rage on as AI assimilates into day-to-day lives of people, it is hard to ignore that the technology potentially holds the key to speedier drug innovation to change healthcare systems and lives of millions of patients. Artificial intelligence and big data in healthcare then is quite literally about the greater good for the greater number.
A Product Manager with expertise in pharma marketing and sales operations