APR Nov/Dec 2022 - 40

» MANUFACTURING
»
Implementation in our department
In the past experience of one author (P.M.P.), one cause for slow
acceptance of DT was excessive focus on top-down corporate
strategies. At the shop floor, these were perceived as too remote from
daily concerns, both in content and accessibility. To increase interest
and acceptance of DT in our own department, an intern position was
therefore funded in 2020. It was framed as a scientific hypothesis: a
test of the value of embedding a data science/IT worker in a physical
science environment to implement a variety of digital solutions. The
approach was successful and the internships have continued every
year since 2020.
The " Swarm " Principle
Organizational context
The Pharmaceutical R&D (PTDC-F) department is part of the small
molecules CMC (Chemistry, Manufacturing, Controls) division at F.
Hoffmann-La Roche. Specifically, the department is responsible for
formulation and process development of solid dosage forms, clinical
trial materials manufacture, and internal/external technical transfer to
commercial production. The division launched a formal DT initiative
in 2019. It includes formally resourced division-scale activities for
infrastructure development (such as the setup of a central data lake
or data harmonization and integration), as well as various activity
streams targeting data use.
For a successful culture change, however, this necessary top-down
effort must be complemented by quickly helping many people
with small requests. Our " Swarm " principle (Figure 1) is a bottomup
approach, where projects are led by internally embedded data
scientists. These iteratively develop prototypes through continuous
partner engagement with science or business subject matter experts
from the department. This principle enables data scientists to deliver
a large number, a " swarm, " of small- to medium-scale collaborative
projects, with fast response times due to low overhead. As applications
mature and grow in complexity, a progressive handover process
to permanent staff ensures long-term maintenance. The " Swarm "
principle accelerates culture change by demonstrating early yet
tangible value delivery to a number of department members, thus
raising general awareness and interest.
Historical perspective and strategy development
The first internship focused on the assembly of a portfolio of potential
projects, together with its associated project management machinery.
It identified sufficient numbers of valuable opportunities even just
in our Sciences section. The delivery grew further the following year,
raising connections and interest across the department, and producing
quite a few useful tools. With the increase of the department maturity,
more usage cases and project ideas from other sections appeared.
By the third year, these successes had generated a very diverse set
of projects in terms of subjects, partners and impact areas. More
formal consideration became necessary regarding tool embedding
into the data flow, their maintenance, and handover, leading to
stronger collaborations with the division's growing community of data
scientists. Further connection of top-down and bottom-up approaches
will be crucial in the future to integrate small-scale projects in larger DT
visions, and to better tailor division-wide solutions to end-users.
Supporting frameworks
We have developed many useful supporting frameworks to apply our
" Swarm " principle effectively (Table 1).
Table 1. Overview of supporting frameworks,
their purpose and contents.
Supporting
framework
Purpose
Content
* Project description including tool business and
code owners
* Users' stories
Project charter
Facilitate project
scoping
* Scoping and project objective including current
state, resources, intended outcomes, foreseen
issues and project validation techniques
* Future project state, maintenance
* Action plan including project tasks, timelines
and expected partner engagement
* Value statements with quality and time
improvements, number of users
* Track project
progress
Portfolio data
table
Figure 1. Visualization of the " Swarm " principle. Division-scale
infrastructure development (represented as pillars) serve as
the basis for the individual projects realized in the company
(depicted as colored patches in the bee hive). Data Scientists
(bright blue bees) work as a swarm in collaboration with natural
scientists (gray bees) on short iterative projects to show a fast
benefit to partners.
* Support
resource
allocation
* Enable portfolio
analysis
Project
documentation
* Facilitate project
handover
and error
troubleshooting
* Serve as user
guidance
* Description of end product functionalities,
testing, implementation strategy and problems
that occurred
* All relevant project links, the project charter,
definition of code and tool business owner
* Supporting best practice documents and
user manuals
* Project progress, resource, priority and
urgency estimations
* Software used, main activities, tool philosophy,
quality and time improvements, number of users
40 |
| November/December 2022
APR Nov/Dec 2022

Table of Contents for the Digital Edition of APR Nov/Dec 2022
