AI in Pharma. What does Artificial Intelligence bring to the pharmaceutical industry?

Radek Kamiński
March 2, 2021

Over the past few years, the term AI has been heavily abused by marketing teams worldwide. Across many different industries, the term has become a convenient shorthand for any consumer-facing technology which shows signs of automation – but not much intelligence. This nonchalant use of the term AI introduced certain confusion and effectively diminished the term’s impact, and made people ignorant of AI’s actually important developments today.

The innovation brought about by the new developments in the area of artificial intelligence and machine learning is nothing short of spectacular. Solutions based on artificial intelligence algorithms are finding applications in many areas of our reality. The pharmaceutical industry is a recent beneficiary of the trend.

Artificial intelligence in pharma is only gaining momentum. There are a growing number of pharmaceutical companies considering – or already using – AI-based solutions in their research, discovery, and manufacturing processes. However, because the knowledge about the state of knowledge and benefits AI brings to the pharmaceutical world is still relatively limited, we thought it would be a great idea to offer an overview and assessment of the pharmaceutical industry’s current AI solutions.

Artificial intelligence and machine learning in pharma #

Slowly but surely, AI is making its way into the pharmaceutical sector, opening new possibilities and creating new competitive ground for innovative pharmaceutical companies ready to leverage bleeding-edge technologies.

AI is a broad term that covers areas including replicating human cognitive processes (i.e., symbolic logic) and machine learning (ML). Using machine learning allows for accurate predictions, classifications, and identification of patterns – in the same way as the human neural network, but much more efficiently and on a much bigger scale.

The pharmaceutical industry lands really well for the application of AI and machine learning. Due to the complexity of manufacturing and regulatory processes, the pharmaceutical industry readily embraces AI offerings in the hope of solving its most nagging problems.

Who’s leveraging artificial intelligence in pharma today? #

It is probably not a big surprise that the leaders in the technological arms race are the biggest players in the pharmaceutical industry who can afford to invest a lot of money in artificial intelligence and machine learning solutions.

Almost all leading pharma companies use some incarnation of artificial intelligence technology or big data solutions to prompt research and development in the area. Artificial intelligence in the pharmaceutical industry can be seen among the following companies:

  • Pfizer uses IBM Watson, a system that uses AI and big data analysis, to power its search for immuno-oncology new drugs with a drug discovery platform.
  • GlaxoSmithKline is a British pharma giant investing in machine learning and AI to automate drug discovery.
  • Sanofi is a French multinational pharmaceutical company headquartered in Paris which leverages AI to accelerate their research into metabolic-disease therapies.
  • Genentech (a Roche subsidiary) is leveraging an AI system provided by the data analytics company GNS Healthcare for researching and creating new cancer treatments.
  • BenevolentBio is a London-based start-up that uses data from sources such as research papers, patents, clinical trials, and patient data into its AI big data platform to gain actionable insights for the pharma industry. BenevolentBio builds artificial intelligence tools to pinpoint relationships between genes, symptoms, diseases, proteins, tissues, species, and drugs.
  • F. Hoffmann-La Roche AG developed a data-driven medical research platform leveraging deep learning.
  • Roche has acquired Flatiron Health, a startup using AI for cancer research and patient care improvement.

But AI is not a domain of the world’s leading pharmaceutical corporations – the technology can actually help smaller pharmaceutical companies level the playing field and possibly gain an edge in this race. Novartis is a Swiss multinational pharma company that has also partnered with IBM Watson, Massachusetts Institute of Technology, Intel, and Quantumblack to bring AI to various healthcare and pharmacy industries – drug trials discovery, patient analytics.

In the same way, the application of AI in pharma is not limited to research and discovery – it finds application in the various stages of development – a complex process that starts from discovering the needs and ends with patient support, dosage control, and ongoing post-market research and analysis of treatment results.

Based on an illustration from Deloitte Analysis

Major areas where AI brings value to the pharma industry #

AI in drug discovery and development #

Benefits of AI in pharma for drug discovery and development

Drug discovery and drug development are the core area of operation of the pharma industry. It abounds in the greatest number of more or less mature solutions. The most promising results of using AI are achieved in the following areas:

  • Data-driven target discovery (e.g., cancer drug targets)
  • Next-generation sequencing
  • Pre-clinical and early-stage drugs discovery
  • Late-stage drug candidates
  • Small molecule therapeutics
  • Novel drug design
  • Novel biological targets

While it is true that drug research is a huge business with eye-watering sums of money at stake, it also costs enormous money to develop a working medicine. One can spend big money on many candidate therapies that ultimately fail along the way and get stuck somewhere in trials or regulatory approval. AI is the answer to how to develop new drugs cost-effectively. It further remedies the problem and optimizes the process of new drug development.

Areas where using AI can help in new drugs discovery and development

AI solves many of today’s big pharma problems – offering quicker, cheaper, and more effective drug discovery.

AI in pharmaceutical manufacturing #

Use of artificial intelligence for pharmaceutical manufacturing

AI allows to streamline production processes in pharma companies – the improvements can span several areas such as:

  • More consistent quality control, helping consistently meet Critical Quality Attributes (CQAs),
  • Shorted design phase,
  • Improved waste management,
  • Supply chain management,
  • Inventory management,
  • Improve production reuse,
  • Predictive maintenance.

Because AI can make production more efficient, resulting in faster output and less waste. This is mainly possible due to reduced human intervention and data processing. Machine learning algorithms ensure that certain tasks are performed more precisely and help identify the areas that could be further streamlined, significantly boosting production processes.

Processing biomedical and clinical data #

AI in pharma to support clinical data processing

Human beings are particularly inefficient at reading, grouping, and interpreting large volumes of data, but this is exactly where AI shines and makes a difference. Researchers in the pharma industry can save vast amounts of time they would otherwise need to spend examining the enormous amounts of data – e.g., research publications – to validate or discard hypotheses.

The application of AI is a boon for the pharma industry. The benefits can span beyond research and manufacturing – AI can also help gather and cross-reference valuable visual, qualitative, and quantitative data collected in clinical studies. The aggregated information on when patients take a drug, what other medications they take, and what reactions they experience can be collected and interpreted with data science. Machine learning can also tap into the vast trove of anonymized information from millions of healthcare providers worldwide to process, analyze and identify important patterns in possible side effects, symptoms, and health improvements.

Personalized medicine and rare diseases #

AI powered software for personalized treatment and new therapeutic approach for rare diseases

Medical information collected by artificial intelligence can be used to produce so-called “knowledge graphs” for various medical conditions, linking genes that are associated with it and compounds affecting it. Essentially, it gives the manufacturers a helicopter view helping to understand the myriad of complex relationships – a daunting task for a human.

What’s important, with natural-language processing and voice recognition, AI-powered platforms can make use of not annotated data.

So, what’s the benefit of personalized medicine? Better diagnosis and treatment for the patients. This is not a pipe dream – this technology is already in use. By combining various medical imagery and patient information gathered from biology and analytics, AI systems like IBM Watson for Oncology can help doctors detect cancer and predict health issues based on genetics. Watson recommends a personalized treatment plan based on each patient’s medical information and history.

AI also offers tremendous support in developing personalized drug treatments based on an individual’s test results, genetics, allergies, and historical data on the patient’s reactions to previous treatments.

Identifying clinical trial candidates #

AI in pharma to support clinical trials and select clinical trial candidates

Medical research’s high cost can be substantially reduced by improving clinical studies' success rate and decreasing the pharma R&D burden. AI could mitigate the current trend of many drugs not reaching the market despite increasing pharma R&D investment.

AI promises to transform how clinical trials are performed from study preparation to execution towards improving trial success rates. For example, a drug trial can be optimized using artificial intelligence by better identifying candidates and streamlined monitoring and coaching of these patients during the trials.

AI makes it possible to apply advanced predictive analytics to analyze patients’ genetic information and help researchers to identify the appropriate patient group for a trial. This can help determine the optimal sample size – a task that would take weeks or months if performed manually by people.

Again, convenience comes as a secondary gain. AI technology such as speech and free-form writing recognition can significantly decrease the processing time of doctor’s notes and intake documents.

Predicting treatment results #

AI in pharma for clinical trials and predicting patient treatment results

Treating cancer is a complex task involving different radiation and medication therapies on top of surgical interventions. To increase the success rates of cancer treatment therapies, results have to be accurately predicted. This is possible today with advanced artificial intelligence models.

AI can support cancer therapies and help to predict a patient’s response to possible drug treatments by identifying relationships between factors potentially affecting the results,

  • The body’s ability to absorb the compounds,
  • The distribution of those compounds around the body,
  • A person’s metabolism.

The medication must often be combined to improve the treatment’s effectiveness and reduce the side-effects to treat cancer effectively. Because experimental screening of drug combinations has historically been very slow and expensive, the benefits of combinational therapy have not been fully discovered yet. Artificial intelligence models are a boon, helping to identify the best combinations to kill cancer cells with specific genetic or functional makeup selectively.

Researchers at three universities in Finland (Aalto University, University of Helsinki, and the University of Turku) joined forces to create an artificial intelligence model that analyzes different cancer drug combinations to kill specific cancer cells. The model was trained with datasets obtained from previous studies on the relationships between drugs and cancer cells. The results have been promising so far: the model found associations between drugs and cancer cells that were not observed previously.

Drugs supply chain #

Use of AI in drug supply chain maintanance

The biopharma supply chain is based on many complex processes and relationships which could be vastly improved with AI to process. This includes decision making, orchestrating operational efficiency, and, ultimately, creating a cost-effective, near-autonomous, and thriving supply chain. According to Deloitte, there are five critical areas and processes of the supply chain where AI is likely to have the highest impact.

End-to-end visibility #

End-to-end visibility means processing data on drug purchases and identifying demand triggers across the whole drug supply chain.

Demand forecasting #

Forecasting demand and precise adjustment of supply and inventory levels are needed to ensure that patients can always obtain timely, reliable, and uninterrupted access to their therapies. This is especially important for drugs with short expiry dates. By leveraging predictive analytics, AI-based technologies can derive insights from the supply chain and better sync market demand and drug supply.

Intelligent process automation #

With digitalization and intelligent process automation (IPA), companies can establish cost-effective, reliable, and robust processes coordinated across the supply chain. This helps make advanced decisions based on robotic inputs' outputs, minimizing errors, improving performance metrics, and generating strategic insights.

Predictive maintenance #

Like any other manufacturing company, biopharma firms constantly deal with many compliance, quality, and safety-related challenges. They need to monitor equipment performance, forecast potential faults and maintenance actions to improve their operational effectiveness, and ensure machine uptime. Predictive maintenance is an AI technology area that makes it a little easier and removes human error from the equation.

Protecting the supply chain #

To tackle the problems associated with counterfeit or substandard drugs, pharma companies are investing in AI technologies. With improved security, transparency, and traceability, they can protect their supply chain integrity and improve trust in the products.

Pharmacovigilance #

Use of artificial intelligence for pharmacovigilance

Pharmacovigilance is the science and activities behind drug safety monitoring – detecting, assessing, understanding, and preventing adverse effects of drugs or other possible drug-related problems.

Pharmacovigilance involves collecting huge amounts of data and then processing it. Recently, the program has broadened its concerns by including herbals, traditional and complementary medicines, blood products, medical devices, herb vigilance, hemovigilance, and materiovigilance. The sole amount of the data points to observe and make conclusions from makes it a great place to apply deep learning algorithms and use AI for advanced analytics. Artificial intelligence in this area opens opportunities to address classification and prediction problems. This drives effectiveness and the generation of new insights.

AI-powered applications can automate the manual and mundane tasks associated with clinical case processing, therefore decrease the time it takes to process and cut overall costs for conducting pharmacovigilance. Another value-adding case for using AI would be applying natural language processing (NLP) to a broad set of data, such as white papers, articles, literature, or electronic medical records, to detect unexpected effects of a new therapeutic product.

Drug adherence and dosage #

AI in pharma supporting drug adherence and dosage

No medical therapy is effective without drug adherence. Only following medical advice on medication, diet, exercise, and mental health can improve the patient’s chances for success.

It is estimated that up to 60% of patients don’t adhere to medical advice – effectively reducing the success rates and ramping up treatment costs. Nonadherence costs the U.S. hundreds of billions of dollars and thousands of deaths annually – something that could be easily prevented with proper technology in place.

Because of the lack of resources to offer remote care services at the patients’ homes, the adherence can only be controlled with regular in-office visits. AI technology is gaining traction in adherence monitoring. This can be done in many ways by leveraging various internet-of-things devices and centralized data collection.

  • Ingestible sensors with RFID tags can transmit a unique signal to a relay device and, once the pill is swallowed, send a signal to a cloud-based server.
  • Smart pill dispensers and bottles: although it never guarantees the pill was actually ingested, a smart dispenser can measure the remaining pill count and send reminders if the pill has not been taken. There are many connected medication platforms on the market, including Pillsy, HERO, PRIA, TinyLogics, or CYCO.
  • Point-of-care drug assays involve the use of smart “bedside” or in-clinic testing devices to assess drug adherence, e.g., by testing urine or serum samples.

Regulatory affairs #

AI in pharma to support regulatory affairs

Regulatory affairs are another important area of the pharmaceutical industry that can benefit from embracing AI technology. On top of the complex discovery and research processes, pharma companies must also remain compliant with changing guidelines and date with the latest international and local industry standards and regulations. Managing this vast knowledge and ensuring compliance would typically involve a regulatory team working alongside the pharmaceutical staff. But even the most diligent team cannot guarantee the drug will make it to market.

AI can be used as a tool to centralize the information on important updates from regulatory bodies:

  • FDA
  • EMA
  • TGA
  • Health Canada
  • Medsafe
  • CHMP
  • PRAC

Pharmaceutical companies need precise interpretation, application, and communication within and outside the company to ensure drug approval. In addition to staying on top of international regulatory requirements, the responsibilities of regulatory affairs professionals would include:

  • Preparing drugs for regulatory submission and managing the approval process with regulatory agencies,
  • Negotiations to ensure authorization of drugs,
  • Finding workarounds for scientific and legal constraints,
  • Collecting and evaluating scientific data,
  • Building strategies related to the commercialization of medications and the success of the company,
  • Making sure the drug’s packaging and advertising complies with local and international regulations and guidelines.

Advanced artificial intelligence systems can provide timely, insightful data to optimize these workflows for regulatory affairs professionals, analyzing new and existing federal laws concerning drug development processes. Once the drug is in production, the AI platform can provide ongoing support, ensuring compliance with marketing, legal, and technical documentation.

With AI, it is possible to provide pharmaceutical regulatory affairs professionals with alerts regarding the latest global, federal, and state legislation. With this dashboard, the regulatory team can assess the medications in the market and create well-informed strategies.

Conclusion #

AI-based solutions in pharma are gaining momentum, becoming the new competitive battleground for many manufacturers. The pharmaceutical industry desperately needs digital transformation and new technologies to process vast amounts of health data efficiently. It identifies significant relationships between them, effectively decreasing time-to-market in drug manufacturing.

Pharmaceutical companies may soon not remain competitive without solid investment in cutting-edge AI and machine learning technologies.

References #

Artificial Intelligence Predicts Drug Combinations That Kill Cancer Cells More Effectively
Intelligent drug supply chain - Creating value from AI

Now, let's talk about your project!

We don't have one standard offer.
Each project is unique, rest assured that we will approach the next one full of energy and engagement.