GenMab
AI assisted web service providing clinical trial documentation oversight.
Transforming regulatory trial compliance checks & processes.
The mission
GenMab is leveraging AI & ML tech as part of their digital transformation journey to improve the efficiency & quality of clinical trial documentation management in order to improve regulatory filing compliance.
The goal was to deliver major improvements against 3 key performance indicators (KPI’s):
Completeness:
All required documents are present and in the right placeTimeliness
All required documents are filed within specified timescalesQuality
The content, data and naming for each document follows mandatory specifications & rules
Known issues & challenges
Bi-annual regulatory compliance is labour intensive and depends almost entirely on offline tools and manual processes for the checking & validation of data, files & reports across disparate systems. Due to the shear volume of documents generated it’s not practical to check all files so a method was introduced for randomly checking approx 10% during ‘Spot checks’ involving many individuals. The most time-consuming aspect was quality checking as this required each document to be opened and visually scanned.
Overview
The original intent was for me to review existing AI / ML initiatives, overlay with a consistent experience and guide future development inline with business ambitions and actual user needs.
In parallel I had to learn Pharmaceutical GxP, regulatory compliance and understand existing systems and processes.
From my involvement with the main initiative I found the existing application wasn’t fit for purpose & wouldn’t meet business ambitions without significant investment. That, plus licence issues, determined the need to build a bespoke application and which would also replace the existing application.
I then had to design an entire product that fully automated the current processes and leveraged AI to meet each KPI.
My contributions
UX / UI design
Protoypes
Design System
User research
Business analysis & requirements capture
Workshops
User story creation
Risk assessment
Acceptance criteria creation
UAT validation plan & methods
Team role and responsibility planning
Backlog management & prioritisation
Sprint planning
Roadmaps and future release planning
Role, duration & client
Consultant - Senior Scientist, Data Science
Feb 2023 to Dec 2023
GenMab
Tools
Word, Excel, Miro, Conluence, Figma, Jira
Phase 1: Existing processes & practices
Deep dive to understand how compliance analysis was performed and provide a blueprint for the future.
Challenge
Baseline the as-is experience & define the to-be state for common use cases
Essential Document Lists (EDL) process mapping
Contribution
Interviewed SME’s and mapped out each core process ‘as is’, including time on task and pain points.
Workshops to define and stress test solutions.
Built out a roadmap based on business priority and feasibility
Outcome
User journeys &/or service blueprints to support optimal experience
Wire frames for report layouts, information hierarchy & data presentation
Contribution
Establish calculations to measure completeness & timeliness and present the data in simplistic and meaningful ways
Outcome
Detailed specification documents to explain the relationships between data points and the required computations
Phase 2: Dynamic report generation
Automate the collection and verification of compliance data across multiple systems and combine into individual reports delivered through a single web application.
Challenge
Create digital experiences to provide business insight
Application user journey
Application dashboard
Contribution
Designed the product application UI & prototypes
Contribution
Produced a design system & component library to support company wide initiatives and ensure common look, feel and experience for all future products.
Component library samples
AI powered analysis chart rendering
Contribution
Provide oversight, design direction & baseline experience for generative AI tools & initiatives
Phase 3: Trust but verify
Define a method to check the data presented by the application was accurate and aligned with business requirements.
Challenge
How to test the application vs. real / expected results when no current method or processes for doing so exsited.
Data validation test
Application simulator & business requirments
Contribution
Created a single excel file that contained 3 linked worksheets, primarly to manage data validation testing during UAT (User Accetance Testing):
Key business requirements & the core values required
Template for manual data input of actual trial data
A simulation of the application UI that showed the data pestented as expected
The file expanded to include explicit instruction relating to data origins, required computations, dates, values, end points etc.
It also became the defacto source of truth for the business, Data Engineers, developers & test teams.
Key learnings & challenges
Mind the gaps
Almost immediately I became aware that significant experience & skill gaps within the core team (POD) existed in relation to product and web application development.
This necessitated the need for me to adopt additional roles & responsibilities throughout the project lifecycle as required:
Product designer
Business translator & analyst
Technical Product Owner
Scrum master
Business goals, requirements & intent
The basic purpose of the product and the business goal was distributed across various confluence pages, Jira stories, UX designs, business requirements etc.
Understanding the 'need' proved challenging to both new & exisiting team members and necessitated frequent, and repeated, UI presentations and requirement explanations.
The later creation of a manual application simulation proved to be an invaluable, but unintentional, instructional tool in that regard. It effectively showcased the intent, desired outcomes and the logic required.