Smart Assistants for High-Risk Areas
Making pharmaceutical processes more efficient and compliant with effective man-machine interaction
By Manfred Hörter, Senior Manager, msg industry advisors and Philipp Csernalabics, Head of Digital Assistance Systems, Chat and Voicebots, msg systems
Change at warp speed, but don’t lose a millimeter of precision in the process: this particularly challenging combination of objectives characterizes the digital transformation in many pharmaceutical companies. The market-specific challenge is that the integration of new technologies in the regulated environment has to meet particularly strict requirements and is correspondingly complex. After all, patient safety, product quality and data integrity are always a top priority.
Speed and precision are perfectly compatible — if you understand the framework conditions and continuously improve man-machine interactions. In this article, we explain how this works using the following aspects and practical examples:
· Paradigm shift in the pharmaceutical industry: Two transformation paths to value creation
· The Neo assistance system: AI support at all levels of automation
· Bringing AI technologies to the company: Five success factors
· Best practice example: Smart glasses with Neo assistance in the clean room
Paradigm shift in the pharmaceutical industry: Two transformation paths to value creation.
Short trips to virtual worlds or exchanges with a digital assistant are already part of everyday life for some employees in pharmaceutical companies. Using VR glasses or even spatial projections in the style of a holo-deck, they design new generations of molecules on the “virtual drawing board” in the R&D lab. For example, each individual molecule can be rotated and turned freely in space, zoomed in to explore the enzymatic areas and manipulated to simulate how it can be bent and modified.
But there is not a single chemical application for this; the digital molecule twin is based exclusively on software applications and data analysis. The same applies to the software bot, which is voice-controlled and answers questions from colleagues in production and quality control when they document activities, deviations or need instructions on a work step.
Both examples are not science fiction, but already a reality — but these are only two out of a multitude of promising technologies. Their use should not result from actionism, but should follow a strategic perspective, a “big picture” for the digital transformation of the company. But what special framework conditions must be taken into account in the case of the pharmaceutical industry? In our projects, two transformation paths repeatedly prove to be decisive: the further development of the value creation process and the implementation of the paradigm of consistency in man-machine interaction.
If a digital tool replaces a manual process step, this must not increase the risk in the overall process.
Transformation of the value creation process
Whether it’s a headache pain killer or a flu vaccine: for every drug, there are fixed rules governing the process the product goes through, from preclinical research, through clinical development, production and all the way to dispensing to the user (see chart 1). In the course of digitization and automation, however, the technological means that can be used in the value creation process have changed dramatically in recent years.
This applies in particular to the starting point, i.e. the research and development (R&D) and preclinical research phase: before digitization, for example, the pharmacological effect of chemical substances was determined using screenings and the trial and error principle. Depending on the project, this could involve hundreds, thousands or millions of individual substances. These and many other time-consuming research methods have now been replaced by software-based computational models, data evaluation, and pattern recognition, increasingly supported by artificial intelligence (AI). This opens up completely new, data-based perspectives on the application options and the development process of the products.
However, one constant remains. As soon as a drug has been successfully designed in this first phase, the company must now provide proof of its efficacy and safety. To this end, clinical trials are conducted, and the drug is tested in clinical development for the first time for use in humans. This is done according to the guidelines for good working practices, the so-called “GxPs”, where the “x” can stand, for example, for “M”= Manufacturing or “D”= Distribution. All manufacturing and production processes — including the interactions of employees within their environment — run according to precisely defined, predefined processes and within precise, predefined limits. And any change to these processes or to the framework conditions requires a strict control process.
It is precisely these requirements that make the phases after preclinical research and drug development critical hot spots for digitization, where opportunities and risks for value creation are not always apparent at first glance. When we introduce new digital tools or further develop existing ones in pharma transformation projects, we, therefore, do so under a clear premise: Wherever a computerized system replaces or supplements a manual process step, this must not increase the level of risk in the overall process — in terms of a risk to patient safety, product quality or data integrity.
Transformation of man-machine interaction
At the same time, however, it is also true that anyone who implements a fully automated and validated process and designs it in such a way that no or only minor human variations are possible makes this process and the data particularly secure. And so this is the second essential transformation path we are pursuing in corresponding projects: optimized man-machine interaction in the smart factory or in the context of Pharma 4.0. The focus here is on establishing the best possible process control and quality assurance beyond research, from production to distribution, so that we minimize errors and, consequently, the process risk.
When designing technological infrastructures, the pharmaceutical world must be guided by the integrity of the data and process flows.
To this end, we also use state-of-the-art technologies such as digital twins, augmented, extended & virtual reality, or digital assistants such as our AI-based, voice-controlled software bot, Neo (see chapter 2) — anything that supports employees in their workflows in some way, frees up their hands, relieves them or prevents distractions is valuable. However, this task goes beyond technological issues. Because in the conservative pharmaceutical world, this is often associated with changes in corporate culture or the need to readjust the objectives of IT or digitization strategies.
The paradigm of continuity applies here above all. I.e., a design of technological infrastructures that still follows the rules of the game of tightly “wired” automation systems located in the facility or process. And are designed to achieve as seamless an integration of the operator level as possible — that is, from the work process with its sensors, its equipment, all the way to the ERP system. This should result in uninterrupted processes and data flows, just as in quality testing, laboratory equipment, laboratory control and manufacturing systems.
This orientation is by no means obsolete and must continue to be a central design principle for digitization. However, their implementation must not ignore the benefits of a virtual infrastructure that adds edge components to the physical machines, facilities and equipment on the shop floor. These capture all relevant data and feed it into a virtual system, which in turn processes the data input according to the well-defined rules of preserving data integrity (ALCOA / see Figure 2). For the pharmaceutical industry, technologies that combine two general characteristics are particularly valuable here: first, they significantly improve man-machine interaction while adhering to certain rules, and second, they significantly reduce the proportion of manual work processes. Our voice-controlled, AI-based assistance system, Neo, demonstrates how this works in practice.
The Neo assistance system: AI support at all levels of automation
Neo interacts with its users across all technological infrastructures of the pharmaceutical industry.
A simple “Hello” on the screen of the first original Macintosh launched Apple’s success story. And to this day, every groundbreaking technology innovation in the consumer market has an element that makes it more emotional, more approachable, more “human”. This works particularly well when complex processes are reduced to simple conversations, the most natural form of interaction — for example, Siri, Alexa and Co. With this intention, we developed the voice-controlled assistant Neo, who can be addressed at any time via headset or keyboard. Neo uses the AI-supported Neo Enterprise Assistant Platform (NEAP) to, for example, guide employees in their production tasks or to achieve paperless documentation (see Figure 3). The software accesses the company’s IT systems, maps process logics, provides information and transfers data to downstream systems. This is how Neo resolves classic pain points in the pharmaceutical industry:
· Reducing manual work steps and securing documentation: In a strictly regulated environment, many manual steps are still necessary, some of them paper-based. Often, employees perform a work step manually at their station in the line and then change location to document this step, in the worst case, still using a piece of paper and pen. Neo takes over such tasks on demand, either via a headset or directly via a mobile device. This makes it possible to work “hands-free”, and all information can be documented in an audit-proof manner.
· Improving the flow of information: As soon as there is a high level of dependency on the expertise of individual experts in the organization or a very strict process control determines the processes in everyday work, the knowledge that is currently required is not always immediately available and on every occasion. Obtaining this knowledge leads to further manual effort or can represent a bottleneck process that results in losses of time and quality. Neo connects systems to automate and to streamline processes so that every user always receives the information that is relevant to them.
What distinguishes Neo from other assistance solutions is its applicability at all levels of the technological infrastructures of the pharmaceutical industry (see Figure 2). If the user carries out on-site measurements (ISA95 levels 1 & 0), for example, the assistant can be connected to the tools used via common interfaces, store the data obtained directly and even suggest actions or measures in the event of deviations and errors, for example. If ERP systems come into play (ISA95 level 4), Neo can transfer the data directly to them and retrieve data from them if necessary.
The assistant combines regulatory and technological aspects of process design and ensures data integrity.
The AI never changes the underlying processes or measurement data. The assistant takes over the prescribed process knowledge, which represents the guardrail — for example, in terms of compliance requirements — and transmits this to the employees. This not only means that the right actions or work steps are performed at the right time, but also that a faster response to measured values outside defined bandwidths is possible. In sum, then, Neo effectively addresses both the regulatory and technological perspectives on process-improving human-machine interaction and data integrity assurance in one fell swoop.
Information about the authors
Manfred Hörter is Senior Manager at msg industry advisors. His consulting focus is on GxP compliance and business process management in the pharmaceutical industry. In addition, he develops corresponding company-wide digitization concepts.
Philipp Csernalabics is Head of Digital Assistance Systems, Chat and Voicebots at msg systems. As co-founder of Neohelden, he co-developed the AI assistance system Neo, which reduces complex processes in the industrial sector to simple conversations.
