Harnessing AI for Enhanced Drug Manufacturing

According to Quartic.ai, management platforms powered by AI can effectively break down data silos and provide real-time insights into drug manufacturing processes. In the realms of drug development and manufacturing, process automation excels at refining discrete subprocesses, but optimizing fully integrated processes poses greater challenges. To achieve this, comprehensive system-wide variables and interactions must be taken into consideration. When data is isolated within subsystems, advancements often remain localized, which means that while individual components may be enhanced, the overall workflow may still suffer.

The synergy of Internet of Things (IoT) technology, artificial intelligence (AI), and digital twins (which represent data mirroring) empowers pharmaceutical manufacturers to access and leverage isolated data. This innovative approach enables real-time adjustments to fluctuations in operational environments and encourages a holistic view of the manufacturing landscape, rather than merely focusing on separate instruments or processes. Furthermore, this technology can automate cognitive and analytic tasks that previously demanded human oversight, resulting in faster decision-making and improvements in quality, output, and operational effectiveness.

Three Key Challenges Addressed by One Solution

“AI addresses three significant challenges facing the pharmaceutical sector: managing complexity, generating intelligence from vast amounts of disparate data, and moving towards modern manufacturing practices,” according to Larry Taber, vice president of life sciences and consumer products at Quartic.ai. The company has introduced the Quartic Platform™, tailored specifically for smart manufacturing.

Smart manufacturing presents an opportunity for the pharmaceutical industry to navigate its increasingly complex production demands as it develops more sophisticated biologics and amps up the integration of various devices that oversee process parameters. Effectively managing this production complexity by harnessing manufacturing intelligence is at the core of Industry 4.0, marking the next phase in the evolution following the digital revolution.

Embracing Industry 4.0 uncovers previously isolated data through the application of cutting-edge tools (including AI and IoT technologies) and analyzes it in almost real-time. This not only leads to more intelligent machinery and processes but also fosters enhanced human interaction across functions. This shift becomes crucial as the industry gears up for a transition to continuous manufacturing operations.

Legacy Systems? No Troubles Here

While Quartic.ai initially centered its efforts on manufacturing systems, the Quartic Platform is applicable across the pharmaceutical value chain. It allows for the integration of data from legacy systems, providing a comprehensive overview of operations. The implementation process starts with a clear understanding of the facility’s layout.

“We begin by incorporating individual data sources—such as optimizing the primary production process,” explains Vinodh Rodrigues, manager of industry solutions at Quartic.ai. “Every adjustment necessitates an evolution in data storage, management, and analysis, and we account for that by aggregating the data.”

To create a unified data environment, the Quartic Platform employs connectors that interface with diverse data systems to extract relevant information. “For users, it’s as straightforward as linking to a data source and viewing measurements from different gauges, probes, and outputs,” Rodrigues adds. “They can then contextualize the data and identify critical variables.”

As more devices get connected, an increased volume of data feeds into models, facilitating comprehensive analyses that optimize interactions and enhance both equipment and individual processes. “When process data is continuously streamed and models are updated,” Taber states, “it effectively generates a digital twin of the manufacturing environment.”

This model bolsters predictive capabilities downstream, empowering operators to act proactively. Consequently, variability, risks, and instances of unplanned downtime can be significantly reduced.

Biotech Finally Embraces AI

Quartic.ai’s founder and CEO, Rajiv Anand, boasts an extensive background in IoT and AI, which has provided him insights into how these technologies have transformed various sectors. His cross-industry understanding led him to perceive significant changes taking place within the pharmaceutical domain.

According to Taber, Anand identified the vast collection of structured data within pharma and recognized the necessity for process control strategies for regulatory submissions. He also understood that AI and machine learning (ML) could be instrumental. “The shift towards biologics, which entails numerous variables, is a prime application for these technologies,” Taber explains. He further notes that Anand perceived the industry’s readiness when the FDA initiated its big data initiative and pilot programs to evaluate AI and ML.

In 2017, Anand established the company alongside a team of specialists in reliability, maintenance, data science, and AI. By 2018, Quartic.ai unveiled its initial platform technology, the Quartic Platform, at Interphex 2019.

Tackling the Fear Factor

“Manufacturers often believe they require massive enterprise solutions with all the trendy buzzwords—IoT, data management, analytics, etc.,” Rodrigues remarks. “In truth, they merely need to optimize the infrastructure already available to them.”

Due to perceived infrastructural hurdles, numerous biopharma companies remain reticent to adopt AI. The technology’s mystique could be a deterrent as well. “The biggest challenge Quartic faces is the lack of trust in AI,” Taber concedes. Historically, AI has functioned as a ‘black box’, consuming data and outputting conclusions without shedding light on how those conclusions were formed. In contrast, the Quartic Platform offers full transparency to users, allowing them to view the algorithms and understand model constructions, eliminating confusion.

For security, the Quartic Platform adheres to the Industrial Internet Consortium Reference Architecture, RAM14.0, IEC62443 cybersecurity standards, Open Platform Communications (OPC), and 21CFR Part 11 regulations.

“We designed our AI platform for everyday scientists,” Taber emphasizes. “Clients progress from feeling intimidated by AI to creating their models directly within the Quartic Platform. They don’t require data science or coding skills—only subject matter expertise. Witnessing their ‘Eureka!’ moments is genuinely thrilling.”

“One fermentation client had data that was scattered yet met all quality benchmarks,” Rodrigues recalls. “We executed a two-step installation, linking our AI platform to all the probes, sensors, and more.” Following this connection, operators accessed data from all measurement points instead of concentrating on isolated components.

“Subsequently, we integrated historical data to aid operators in understanding how their processes interrelated over time,” he continues. This allowed operators to apply institutional knowledge to current challenges and find swift, effective solutions. That deployment has since expanded to several significant platforms within the facility as it transitions to the entire manufacturing system.

A Compelling Use Case

As a highly regulated industry, pharma is typically averse to rapid change, yet the potential for AI can no longer be ignored. “Rectifying past issues presents one of the simplest avenues to validate the benefits of AI,” Rodrigues notes. For many organizations, previous problems have led to the collection of vast troves of data—often readily available for retrospective and prospective analyses.

“AI,” Rodrigues asserts, “acts as an accelerator, enabling manufacturers to derive additional insights and value from their existing data.”

Looking Ahead: Expanding Features

Quartic.ai is embarking on a path of broadening its insights. In collaboration with clients, the company tackles not only “what” questions—determining current happenings and deciding on appropriate responses—but also “why” questions—understanding the importance of specific interactions and leveraging leading indicators for performance predictions.

This advancement propels the AI application platform further upstream, integrating initial quality-by-design and failure mode and effects analyses into the development phase to optimize performance concerning critical process and quality attributes.

Focusing on its upcoming release, Quartic.ai is enhancing its features and capabilities tailored for more established, sophisticated companies. Key components of this enhancement include reinforcement learning and federated learning.

“In reinforcement learning,” Rodrigues clarifies, “the machine gains an understanding from the feedback provided by sensors and users, allowing it to adapt in real-time. It learns from the ongoing changes within processes.

“Federated learning, on the other hand, employs a single master model that functions across numerous assets—like motors and pumps—learning how each asset operates independently. This knowledge can subsequently aid in optimizing similar assets. For instance, insights from the battery management system of my phone could also enhance yours.”

“Our goal is to democratize technology access,” Taber adds. “We believe we accelerate learning collectively, hence our aim to make AI tools available to all. In the pharmaceutical sector, this requires a meticulous review of the entire lifecycle and the crucial parameters influencing it. Together, they weave a tapestry of vital intelligence.”

Revolutionizing Drug Manufacturing with AI Insights

According to Quartic.ai, AI-driven management platforms can dismantle data silos, offering real-time insights into drug manufacturing processes. In both drug development and manufacturing, process automation can easily enhance discrete, standalone subprocesses. However, optimizing whole, integrated processes is far more challenging. System-wide variables and their interactions must be addressed. When data from individual processes remain siloed, local enhancements generally do not create widespread improvements. Consequently, even if specific equipment and processes are optimized, the overall system may underperform.

The synergy of Internet of Things (IoT) technology, artificial intelligence (AI), and digital twins (data mirroring) enables pharmaceutical manufacturers to leverage siloed data effectively. This innovative approach allows for real-time reactions to operational variations and provides a comprehensive view of the system rather than isolated instruments or processes. More importantly, it can automate cognitive and analytical tasks that previously required human input. Such automation can reduce response times while enhancing quality, output, and operational efficiency.

Three Challenges Facing Biopharma: A Singular Solution

“AI is equipped to tackle three significant challenges within the pharmaceutical landscape: navigating complexity, deriving intelligence from vast amounts of disparate data, and shifting towards contemporary manufacturing techniques,” states Larry Taber, Vice President of Life Sciences and Consumer Goods at Quartic.ai. The company has rolled out the Quartic Platform™, engineered to facilitate smart manufacturing.

Today, smart manufacturing plays a crucial role in enabling the pharmaceutical sector to manage increasingly intricate production processes, which are now delivering more sophisticated biologics and incorporating a growing number of devices that monitor and control critical process parameters. Harnessing manufacturing intelligence to navigate production complexity embodies the essence of Industry 4.0, the fourth industrial revolution that succeeds the digital revolution.

Industry 4.0 reveals previously isolated data with the help of new tools (inclusive of AI and IoT technology) and analyzes it in near real-time. The outcome is smarter machinery and processes alongside more engaged, cross-functional interactions, which is crucial as the industry embraces continuous manufacturing operations.

Seamless Integration with Existing Systems

While Quartic.ai initially concentrated on manufacturing systems, the Quartic Platform is fully adaptable across the entire pharmaceutical value chain. This platform connects the information housed in legacy systems, providing holistic visibility into operations. Implementation begins with assessing the layout of the plant.

“We then integrate individual data sources, such as optimizing the primary production process,” explains Vinodh Rodrigues, Industry Solutions Manager at Quartic.ai. “Every adjustment brings a shift in data storage, management, and analytics. We take this into account and consolidate the data.”

To establish a unified data environment, the Quartic Platform employs connectors that interface with various data systems to extract pertinent information. “For users, it’s as straightforward as linking to a data source and accessing readings from varied gauges, probes, and outputs,” Rodrigues continues. “From there, they can enrich the data with context to identify critical variables.”

As more devices are interconnected, the influx of data feeds models, yielding a comprehensive analysis to discover optimal interaction parameters for both individual equipment and processes. “As streaming process data occurs and models undergo retraining,” Taber underscores, “a digital twin of the manufacturing environment gradually takes shape.”

This model significantly enhances predictive accuracy downstream, allowing operators to adopt a more proactive approach, thereby minimizing variability, risks, and any unexpected downtime.

The Timing is Right for AI in Biotech

With a robust background in IoT technology and AI, Quartic.ai’s founder and CEO, Rajiv Anand, has witnessed the transformative impact of these technologies across multiple sectors. This cross-industry expertise illuminated for him the significant evolution underway in the pharmaceutical realm.

Taber articulates how Anand acknowledged the extensive structured data collection in the pharmaceutical industry and the necessity for process control strategies tied to every regulatory submission. He recognized that AI and machine learning (ML) had a pivotal role to play. “The evolution towards biologics, with its vast range of variables, represents an ideal opportunity for deploying these technologies,” remarks Taber. He also notes that Anand sensed the industry’s readiness when the FDA initiated its big data program to explore AI and ML applications.

In 2017, Anand established the company with a team well-versed in reliability, maintenance, data science, and AI, unveiling the first iteration of their platform technology, the Quartic Platform, at Interphex 2019.

Overcoming Barriers to Adoption

“Manufacturers often believe they need massive enterprise solutions replete with all the jargon—IoT, data management, analytics, and so forth,” Rodrigues states. “In reality, they simply need to utilize the existing infrastructure at their disposal.”

Many biopharma companies remain hesitant to adopt AI due to perceived infrastructure hurdles, and the technology’s reputation for being enigmatic can also deter them. “Trust in AI remains one of Quartic’s most significant hurdles,” Taber acknowledges. Traditionally, AI has functioned as a black box, ingesting data and producing conclusions without disclosing how these conclusions were reached. In stark contrast, the Quartic Platform provides full transparency, allowing users to see the algorithms and the models they construct, eliminating any obscurity.

To ensure security, the Quartic Platform adheres to established standards including the Industrial Internet Consortium Reference Architecture, RAM14.0, IEC62443 cybersecurity protocols, Open Platform Communications (OPC), and 21CFR Part 11 guidelines.

“We’ve designed our AI platform with everyday scientists in mind,” Taber asserts. “Clients have transitioned from a place of trepidation regarding AI to independently crafting their own models within the Quartic Platform. They don’t require data science or coding prowess; instead, just domain expertise. Witnessing their ‘Eureka!’ moments is truly exhilarating.”

“One fermentation client faced data inconsistencies despite meeting all quality standards,” Rodrigues recalls. “We conducted a two-step installation, linking our AI platform to various probes and sensors. As a result, operators could now visualize data from all measurement points, eliminating the narrow focus on individual elements.

“Subsequently, we integrated…”

Data from earlier years serves as a valuable resource, enabling operators to grasp the historical interconnections of their processes,” he elaborates. This approach allows operators to leverage institutional knowledge to tackle current challenges and identify effective solutions swiftly. This application has since expanded to numerous significant platforms within the facility, progressively rolling out across the entire manufacturing system.

A Compelling Use Case

In the pharmaceutical sector, which is heavily regulated, adapting to change is not commonplace. Nevertheless, the case for implementing AI is compelling. “One of the quickest ways to demonstrate the value of AI is by addressing historical challenges,” Rodrigues states. For many organizations, tackling these past issues has led to the accumulation of vast amounts of data, often readily available for both retrospective and forward-looking analysis.

“AI,” Rodrigues asserts, “acts as an accelerator, enabling manufacturers to extract further benefits from their existing data.”

What Lies Ahead? Enhanced Features

Quartic.ai is poised to expand its vision. In its collaborations with clients, the company is shifting its focus beyond just answering “what” questions—identifying current events and determining responses—to include “why” questions—discern the importance of specific interactions and utilizing leading indicators to forecast performance. This approach enhances the AI application platform, pushing it deeper into development, where initial quality-by-design principles and failure mode effects analyses can be integrated to boost performance concerning essential process and quality attributes.

Regarding product development, Quartic.ai is gearing up for its next launch, which will introduce additional features and capabilities tailored for more advanced, mature companies. Two pivotal components of this evolution are reinforcement learning and federated learning.

“In reinforcement learning,” Rodrigues clarifies, “the system can adapt based on feedback from sensors and users, allowing it to remodel itself in real time. It learns from shifts in processes.”

“With federated learning, a central model operates across multiple assets—such as motors and pumps—and learns the individual operation of each asset. This understanding can then be applied to optimize similar assets. For instance, insights gained from the battery management of my phone could benefit yours.”

“We’re dedicated to making this technology widely available,” Taber remarks. “Our learning accelerates when others do, so we aim to make AI tools accessible. In the pharmaceutical realm, this entails a careful examination of the entire lifecycle and the critical parameters influencing it. All these elements interconnect to generate valuable intelligence.”