Biomedical Machine Maintenance Scheduling

Tissue banks procure approximately 45,000 tissue donations per year, providing nearly 9,000,000 individuals (about half the population of New York) with life-enhancing and life-saving medical procedures. Proper biobank machine maintenance is imperative to this process. Mandatory forms of maintenance are critical to avoid unexpected malfunctions, which can halt operations and render samples unusable. Each machine has a unique reliability rate within the system; although some can quickly be repaired or replaced, many processes rely on limited machinery where even planned downtime can significantly influence the tissue processing. AlloSource, one of the largest tissue manufacturers in the United States, too often schedules these preventive events unnecessarily or inconveniently, resulting in machines breaking down at inopportune times. In response to these inefficiencies we ask, “What is the best consolidated and standardized equipment maintenance schedule that maximizes monthly maintenance events to ensure increased equipment availability while meeting the demand of the biomedical manufacturing network?” We use an optimization model to consider equipment reliability, downtime, availability, and demand to develop a preventive maintenance schedule. Our model focuses on scheduling the maximum number of events the maintenance crew can conduct each month to ensure vital equipment to the allograft process is available, which provides more opportunities for tissue therapies. In doing so, the maintenance crew is also able to complete more events, driving up annual throughput while driving down equipment downtime.

Accelerating manufacturing to start large-scale supply of a new adenovirus-vectored vaccine within 100 days

The Coalition for Epidemic Preparedness Innovations &rsquo &lsquo 100-day moonshot &rsquo aspires to launch a new vaccine within 100 days of pathogen identification. Here, we describe work to optimize adenovirus vector manufacturing for rapid response, by minimizing time to clinical trial and first large-scale supply, and maximizing the output from the available manufacturing footprint. We describe a rapid viral seed expansion workflow that allows vaccine release to clinical trials within 60 days of antigen sequence identification, followed by vaccine release from globally distributed sites within a further 40 days. We also describe a new perfusion-based upstream production process, designed to maximize output while retaining simplicity and suitability for existing manufacturing facilities. This improves upstream volumetric productivity of ChAdOx1 nCoV-19 by around four-fold and remains compatible with the existing downstream process, yielding drug substance sufficient for 10000 doses from each liter of bioreactor capacity. Transition to a new production process across a large manufacturing network is a major task. In the short term, the rapid seed generation workflow could be used with the existing production process. We also use techno-economic modelling to show that, if linear scale-up were achieved, a single cleanroom containing two 2000 L bioreactors running our new perfusion-based process could supply bulk drug substance for around 120 million doses each month, costing <0.20 EUR/dose. We estimate that a manufacturing network with 32000 L of bioreactor capacity could release around 1 billion doses of a new vaccine within 130 days of genomic sequencing of a new pathogen, in a hypothetical surge campaign with suitable prior preparation and resources, including adequate fill-and-finish capacity. This accelerated manufacturing process, along with other advantages such as thermal stability, supports the ongoing value of adenovirus-vectored vaccines as a rapidly adaptable and deployable platform for emergency response.

Multi-Agent-Based Self-Organizing Manufacturing Network Towards Mass Personalization

Mass personalization is arriving. It requires smart manufacturing capabilities to responsively produce personalized products with dynamic batch sizes in a cost-effective way. However, current manufacturing system automation technologies are rigid and inflexible in response to ever-changing production demands and unforeseen internal system status. A manufacturing system is required to address these challenges with adaptive self-organization capabilities to achieve flexible, autonomous, and error-tolerant production. Within the context, the concept of Self-Organizing Manufacturing Network has been proposed to achieve mass personalization production. In this paper, we propose a four-layer system-level control architecture for Self-Organizing Manufacturing Network. This architecture has additional two layers (namely, Semantic Layer and Decision-Making Layer) on Physical Layer and Cyber Layer to improve communication, interaction, and distributed collaborative system automation. In this architecture, manufacturing resources are encapsulated as Semantic Twins to make interoperable peer communication in the manufacturing network. The interaction of Semantic Twins consolidates system status and manufacturing environment that enables multi-agent control technologies to optimize manufacturing operations and system performance.