AI-driven quality assurance solutions can mitigate the $50 billion lost annually by pharmaceutical companies, says software development firm Modicus Prime.
It is rare throughout the course of history for an event to hold so much global significance that it comes to define the year in which it occurs. According to the Chinese zodiac, 2020 was the year of the rat. It was the year of a contentious United States election between Donald Trump and Joe Biden. And it was supposed to mark a time when the world came together to celebrate the 2020 Tokyo Olympics.
But unlike today, four years later when we can gather freely to watch the world’s best athletes compete for the gold in Paris, the Olympics didn’t happen in 2020. Instead, we worried about the health of ourselves and our loved ones. We socially distanced, wiped down our groceries, and fretted over the availability of toilet paper.
And the world of biomanufacturing never felt so critical. By the end of the year, the vaccines came, but it wasn’t without a hitch. As big pharma companies scrambled into 2021, trying to make enough doses to serve the world population, they made some mistakes. According to a US government report, nearly 400 million doses of Johnson & Johnson (J&J) and AstraZeneca vaccines were destroyed due to fears of cross contamination.
But such errors aren’t restricted to big pharma labs, nor are they confined to times of societal unrest. “Even though drug safety was a global focus at that time, drug-product quality is a systemic issue,” said Taylor Chartier, CEO and founder of Modicus Prime, an AI-software developer that specializes in intelligent image analysis of biologics. She said that such problems result “in $50 billion lost annually by the pharmaceutical industry due to product quality failures and the legal and reputational damage that ensues.”
Chartier founded Modicus Prime shortly after the onset of the COVID-19 pandemic with the hope of providing solutions to such pervasive quality lapses. She decided to leverage her career experience in machine learning (ML) and AI solutions to optimize pharmaceutical production. She told BioProcess Insider, “I was already familiar with what types of technologies could benefit big pharma and allow them to prevent the release of contaminated products to protect the public. Traditional or classic techniques are not enough to solve these drug quality issues, so turning to AI for more scalable and accurate approaches is a natural solution.”
Modicus Prime recently announced the completion of a $3.5 million in series seed funding led by Silverton Partners, with additional backing from Alumni Ventures and other firms. The funding follows the implementation of mpVision cloud software across multiple global pharmaceutical companies and contract development and manufacturing organizations (CDMOs) making biologics, cell and gene therapies (CGTs), and vaccines.
Chartier’s team developed the mpVision platform to address unmet needs in the industry such as real-time product quality assurance (QA), full agency compliance, faster go-to-market, and reduced operating costs. She explained that “it is a hardware agnostic software layer that is added to any of biopharma’s camera-equipped devices, ranging from microscopes to microflow imaging devices. To implement mpVision, our cloud ecosystem enables end users across IT, data science, automation, and/or research to simply connect their camera-equipped hardware to our cloud platform through our application programming interface (API).”
She added AI technologies are an important investment for companies, especially when companies are cost sensitive and even engaging in layoffs. Customers specifically evaluate mpVision software in terms of time and cost savings. “Companies utilizing our solution for process development quality have, for instance, reduced their experimental preparation and post-experimental image processing hours by 89% for every 100 experiments.”
According to Chartier, pharma companies experience one product quality failure per year on average, illustrating the need for better QA. “During 2021, public FDA deviations spanned across companies such as Teva Pharmaceuticals, Hospira (a Pfizer company), and Sun Pharmaceutical Industries. Examples of such contaminations reported include microbes; glass particles, silicone particles, crystallization, cotton fibers; and general lack of sterility assurance.” She emphasized that the aforementioned COVID-19 vaccine contamination incident cost the subjected company over $600 million. And she said that AI software can help.
“A common initial use case of mpVision is the generation of an AI model on small-scale process development data that is representative of large-scale manufacturing,” she explained. “Leveraging their domain expertise, end users can create their own training datasets by annotating particular drug morphologies, concentrations, or contaminants that they desire their AI to learn. Once the AI has finished learning the dataset features, the end user can leverage our scalable cloud infrastructure by distributing their AI model to different sites and departments for real-time manufacturing quality control.”
Chartier said that mpVision software offers online quality monitoring that enables users to define monitoring metrics and set alerts. “Device Access Points are initially generated for each AI model on its imaging device, and every model is tied directly to its respective alerts for monitoring. These monitoring alerts may be configured to send notifications in real-time or at specified time intervals.” The system’s monitoring metrics can identify objects or measurements that the computer vision model was trained to detect and monitor the confidence of that model. Those metrics “help ensure the identity and purity of drug products.”
“For example,” Chartier said, “an AI model may have been trained to identify a particular morphology indicating a heat spike during production or a shear stress signature.” She added that it “may also have been trained to detect microbial contamination, glass particles, silicone particles, crystallization, or cotton fibers observed in vials. Any contaminations can immediately be detected by mpVision and reported to prevent distribution of altered product.”
She added, “CDMOs have conveyed that particulates characterization and monitoring provided by mpVision is a significant differentiator, as they are able to de-risk their manufacturing by training AI models to create particulate libraries for their customers, distinguishing between high-risk and low-risk particulates.”
Chartier explained that the software generates reports that can be compared across multiple production runs and shared among colleagues. “This makes data traceability feasible, with all particles tied directly to their respective frames, experimental, and production runs.”
The mpVision software design was informed by the recommendations of a top 10 pharma company that then tested its accuracy, speed, security, and data-processing requirements on a global scale. In turn, Modicus Prime was awarded its first multi-year customer contract. She credited working with J&J JLABS for accelerating her company’s growth.
According to the J&J website, JLABS is a global life science network that provides start-up companies with resources that enable them to thrive. Said Chartier, “The J&J summits and exclusive networking events have catapulted our company into mainstream big pharma.”
“Our team has even been told by a former FDA Center for Biologics Evaluation and Research (CBER) division of manufacturing product quality inspector that he’s been waiting for mpVision, which is a significant stamp of approval and testament to our product’s value.”
You May Also Like
