✕
☰

Enabling failure prediction in pharma as time to market demands grow

by

With AI becoming more prevalent in the pharma space, Raman Bhatnagar, Pharmaceuticals lead, Aspen Technology, explores the power of machine learning and analytics, and how this helps anticipate failure prediction in pharma manufacturing.

Key insights:

One of the biggest challenges the pharmaceuticals industry is facing is to ensure fast time to market. The speed at which companies developed and brought new Covid-19 vaccines on stream during the pandemic was widely celebrated. The fastest any vaccine had previously been developed – from viral sampling to approval – was four years – for mumps in the 1960s. Yet the Pfizer-BioNTech COVID vaccine was approved for use on 2 December 2020, just seven months after the start of clinical trials, with other vaccines following close behind.

That has set a precedent that the industry needs to live up to. Currently, half the world lacks access to essential health services, and with the global population expected to grow from 7.8 billion in 2020 to 9.9 billion in 2050, the demand for consistent access to vital drugs and vaccines will continue to increase.

Moving forwards, manufacturers will focus on developing a more efficient production cycle to support accelerating the industry’s time to market for all medicines. But achieving this will not be easy. Macro-economic disruption from China’s Zero COVID policy to the War in Ukraine continues apace. Over the past 2-3 years, pharmaceuticals businesses have become increasingly used to expecting the unexpected.

The notion of volatility, uncertainty, complexity and ambiguity (VUCA) best explains this growing trend. While all four individually are reflective of distinct elements that can cause businesses to lose control of their wider environment, they also come together to prove that with increasing volatility and complexity in an industry, the harder it will be to predict future events. This is pushing agility to the forefront for organisations.

Pinpointing the factors that matter

At a more granular level, factors impacting a pharmaceutical company’s ability to ensure supply of end products include equipment failures but also continually-changing operating conditions and the associated impact on process health. Both factors need to be addressed enterprise-wide in a timely, scalable fashion. Both can cause batch quality failures, resulting in costly production losses and disruptions to supply. With single batch values for some drugs surpassing running into the millions of pounds even one lost batch can deliver a serious blow to profits and supply. 

Finding a way forward

Digital solutions, particularly advanced analytics, can help pharma manufacturers prevent equipment breakdowns and ensure process consistency, optimising production and protecting the supply of product to customers. The good news is that pharma manufacturers increasingly understand the value of these kinds of tools.

These advanced technologies are increasingly prevalent in the pharmaceuticals space. The global AI in pharma market reached a value of nearly $699.3 million in 2020, having increased at a compound annual growth rate (CAGR) of 31.8 % since 2015. The market is expected to grow from $699.3 million in 2020 to $2,895.5 million in 2025 at a rate of 32.9%, according to  "AI In Pharma Global Market Opportunities and Strategies to 2030: COVID-19 Growth and Change." 

Predictive maintenance solutions use advanced analytics to identify the signs of pending equipment failure and warn maintenance teams in advance, allowing drug makers to plan repairs, adjust production and avoid unplanned failures that would result in lost product. 

Process multivariate analytics solutions further evaluate the complex variables in batch production and determine which are critical to quality, helping pharma producers keep batches on course and driving greater consistency and yields. With these tools in place, pharmaceutical makers position themselves to safeguard security of supply, optimise production and control costs.

Analysing precise failure patterns 

Predictive maintenance solutions analyse precise failure patterns to provide anomaly alerts and advance warnings of pending equipment failures. With weeks or even months to plan for repairs, drug manufacturers can avoid unexpected shutdowns, reducing maintenance costs and preventing production losses. 

GlaxoSmithKline (GSK) is a science-led global healthcare company that researches and develops a broad range of innovative medicines and brands. Today, it is creating a future-ready supply chain with predictive and prescriptive maintenance using Aspen Mtell. With Aspen Mtell, GSK receives up to 35 days’ advance warning of potential issues.

The manufacturer estimates that Mtell has also enabled it to save tens of millions of US dollars in lost batches avoided and achieve a 50% reduction in lifecycle maintenance costs.

As Kevin O’Keeffe, head Engineering Primary & Antibiotics Manufacturing, GSK, stated: “Early and accurate warnings, speed and scalability of deployment, and ease of use were key drivers in choosing Aspen Mtell for predictive maintenance. Aspen Mtell effectively predicts factors causing pharma process disruptions, improving production uptime and avoiding plant deviations.” 

How predictive maintenance helps prevent failures 

Predictive maintenance tools also offer faster implementation and shorter time to value than many other digitalisation investments, making them a strong choice for pilot projects in what Deloitte calls “digital incubators.” 

Easy-to-use predictive maintenance software captures and analyses data quickly, making best use of all knowledge and skillsets within an organisation. Drawing on data history, past work orders and known failure modes, plant staff can develop Agents that identify anomalies and signs of pending failure. The Agents provide alerts that allow operators to investigate anomalies to determine whether they indicate a potential problem or schedule maintenance before a breakdown occurs.

Multivariate analytics software analyses and continually monitors for how discrepancies in material properties, variations in procedures and process anomalies such as sensor drift and changing environmental conditions impact the final product. These tools can help identify and troubleshoot process and product quality issues, increase yields and reduce off-spec product. Multivariate analytics monitoring can be especially valuable when applied to the complex chemical and biological processes prevalent in many pharmaceutical manufacturing processes; not reacting to small variations can steer a high-value batch off course. 

The latest sophisticated tools support data scientists executing deep analysis while also being accessible to the operational teams most familiar with the process and helping them make sense of the available data. Without creating a need for dedicated staff, software that quickly diagnoses batch deviation to enable informed and timely action pays for itself many times over. Moreover, because these solutions can draw conclusions from sparse data, additional sensors or physical inspection rounds are not required to reap the benefits of reduced downtime.

Accelerating digitalisation with advanced analytics 

Pharmaceutical companies that embrace the power of advanced analytics solutions gain a significant competitive advantage by reducing maintenance costs and eliminating production losses to ensure security of supply. Tools that can easily scale create greater agility, and deliver greater value, while leveraging all existing process and data science talent effectively. 

Predictive maintenance and multivariate analytics tools offer fast return on investment and quick wins for pharmaceutical companies at all stages of the digitalisation journey. Deloitte estimates that these tools can improve overall equipment effectiveness (OEE) by up to 20% and deliver as much as a 5% increase in yield. These benefits add up over time, especially as organisations scale across multiple products, assets and sites. 

As the pharmaceuticals market becomes ever more complex and competitive, security of supply and fast time to market will become ever more important. Rapidly identifying ways to increase OEE, reliability and throughput, while maintaining product quality offers pharmaceutical companies a formula for success – and predictive maintenance and multivariate analytics are the tools that make that formula happen. 

Back to topbutton