KASA or No, It’s Time to Embrace Structured Knowledge Management
“Pharmaceutical procedure” and “excitement” don’t typically go together. Outsiders wouldn’t exactly line up for “Pharma Six Flags”, but observers of ever-evolving pharmaceutical data have been on a rollercoaster ride for the past several decades. Henry Levy, chief strategy officer for Veeva sums it up: “Thirty years ago, electronic data wasn’t common and all of your data was in one place, and it was ugly, but it was centralized and managed in one place. And then EDC was a revolution… But it actually broke everything else.”
Image via The Medical University of South Carolina
What’s broken? Data management – the glue that holds medical data together; the common thread that makes clinical data points from sources like EDC (and now PRO, mHealth, lab data, and more) actually useful. First, pharma companies tried overly-structured data warehouses to reconnect the dots, then data lakes – but still, clinical information is disconnected, hard to find, and even harder to leverage. The problem isn’t unique to pharma; it hinders all aspects of healthcare and beyond.More often than not, the gap between theory and practice around data management is a full blown chasm. Despite the advanced technology at our fingertips, data management has been notoriously difficult since the days of vast paper record rooms and filing clerks. Your data, in total, should represent a collective record of your organization’s experience, it’s major failures and learnings. Ever tried to find a simple summary of something like this? The thought is laughable. Many conventional approaches have failed; maybe in order to unlock the value of an organization’s knowledge, we need something more powerful than data management. You need structured knowledge management.
What is Structured Knowledge Management?
The classic definition of knowledge management, or KM, is “the process of capturing, distributing, and effectively using knowledge”. Fierce competition (regardless of industry) forces today’s companies to use every tool at their disposal to leverage any competitive advantage. It’s common practice to search structured data like dashboards, spreadsheets, classic analytics tools for insight. But what about all the information hidden in emails, text-based PDFs, and powerpoint presentations? Structured data only makes up 20% of an organization’s information. More difficult to leverage unstructured data makes up the other 80%, yet it goes largely ignored today. Organizations recognize its value; it’s simply too difficult to access in any meaningful way.
Some industries are taking note, however. In pharma, where data is king, there is a joint push from the FDA and pharma giants to streamline the cumbersome regulatory reporting process through next-generation knowledge management. This effort, called KASA (Knowledge-aided assessment and structured applications), aims to better manage pharmaceutical knowledge throughout the entire drug lifecycle. All parties involved agree that structured knowledge management is the key to unlocking huge efficiencies in the regulatory process which will turn into better treatments for patients all over the world.
KASA requires a strong foundation: SKM for pharma
KASA is the natural evolution of the current eCTD format optimized for the 21st century. It tracks the drug development cycle better than eCTD and gets rid of unstructured data altogether, making information consistent across submissions. This gives the FDA more clarity and better risk mitigation and speeds up the review process. At a high level, the FDA will set rules and algorithms which it can use to standardize review and capture risk early-on. Better risk management upfront will reduce the iterations needed to reach approval, and because each submission will arrive in a standardized format, reviewers will be able to automate redundant work and pull better historical data.
KASA or no, the time has come for big pharma to embrace structured knowledge management. It’s a powerful tool for efficient operations, and now it’s a competitive advantage as well – giving an edge over companies still manually slogging through unstructured text-based documents. Drug approval submissions that are complete, transparent, and easy for the FDA to process will get approved faster and go to market faster. It’s as simple as that.
Outside of KASA, there are huge gains to be realized for pharmacos that embrace SKM. We’ve collected the top three.
1) Repetitive processes use identical data sets but cost big
Pharmaceutical data has many uses. From regulatory reports to clinical trial design documents, to molecule design during drug discovery – data is worth its weight in gold to pharmacos and they use it to every advantage.
Image via Coseer
With drug discovery and development processes stretching from 12-16 years from ideation to marketable drug, the same data sets are accessed and in some cases changed, moved, or added to countless times. Pharma processes like filing for regulatory approvals in different geographies require the exact same data sets, but often in different formats. As the figure suggests, such applications of managing repetitive data are important to every stage of drug life cycle.
The time lost finding relevant data, reformatting, and fixing errors/updating figures can be huge. Gartner estimates that bad data costs the average organization $9.7M per year. IBM paints an even bleaker picture; bad data costs the US economy over $3 trillion per year. The cost to pharma is likely higher than average, as those digging for answers aren’t exactly interns – they’re highly compensated scientists.
How can SKM help?
Structured knowledge management has the ability to strip data from its various sources and condense it to its essentials while keeping all context ready and instantly available to users with just a click. Data housed in a single easily-accessed and queried location could save scientists and medical writers months as they return to the same information from preclinical efficacy studies all the way through phase 3 clinical trials and regulatory review.
2) Multiple workflows handling the same data leads to multiple versions of the “truth”
Version control is an issue for small organizations, but the problem refracts in larger companies. Many groups of people work with the same or similar data sets, sometimes at the same time. Document management systems are great in theory, but when we are all asked to be file clerks, it shouldn’t be surprising that things tend to fall to chaos. As CMS wire puts it, “what we got instead [with document management systems] was an electronic file room with no file room clerks — or rather, we all became file room clerks. Except we had no training, no particular aptitude and no guidance on how to do that job. So, like a closed stack library suddenly opened to the public with no rules for how to manage the volumes, our electronic file room became a digital landfill, a dumping ground for more documents than we could ever use, in no particular order, that we struggle to work with.”
Because the same data sets are used so extensively in pharma applications, data integrity and version control are critical, but often messy. Let’s take the regulatory report-building process for example. Scientists and medical writers set to the task may not have been a part of the original research, and likely don’t know which document repository contains the best, most up-to-date information. Even in the best case scenario, a medical writer may spend weeks searching for the right information and completing the initial drafts.
EOP2 Report creation timeline at a top-20 pharmaco before SKM
Image via Coseer
How can SKM help?
Technologies such as Natural Language Search-based AI that enable SKM are already in place at pharma giants. The above graph tracks the EOP2 report creation timeline before this top-20 parmaco implemented an NLS-based solution. Now, because search has been significantly reduced, scientists save months creating their report. Additionally, some SKM tools enable simultaneous search, collaborative drafting, and data integrity review, cutting out even more waste from the process.
3) Data silos obscure the big picture
A data silo is a collection of isolated data not accessible by all in an organization who may need it. There are many root causes of data silos but regardless of whether the cause in your case is cultural, structural, or technological, the effect of a data silo is confusion, error, and redundancy and waste. These silos chip away at productivity bit by bit as each group works without an understanding of the big picture, and struggles to recreate already existing data or goes without and operates in the dark.
Image via Beyond PLM
Data silos compound storage costs as multiple versions of documents are stored in different places. There is a high risk that up-to-date information will be accidentally overwritten by an out of data document. And of course, scientists must allocate time and effort tracking down nuggets of information from vendors, team members, and distant colleagues. Waste compounds.
In pharma, the stakes are high – it costs billions and takes years to put a drug on the market, and each innovative treatment stands or falls on the quality of relevant clinical data. If a team of medical writers putting together an EOP2 document for the FDA accidentally use outdated clinical trial information, they risk rejection and months of delays in an already time-consuming process.
In the past, pharma used highly structured data warehouses to solve this issue. More recently, many pharmacos have adopted “data lakes” in which the goal is to capture all data, and find a use for it later – hopefully. That hasn’t worked. Some have commented that data lakes are “ like having the internet, with everything you’d ever want to know at your fingertips, but with no Google.” Generic tools seem great at the outset, but can’t handle the domain complexity of pharma, and highly customized tools cost too much and often don’t deliver.
How can SKM help?
Structured knowledge management done well, powered by AI that can “understand” natural language, could be the answer to the data lake problem. The key to data management success in pharma lies in a tool that is purpose-specific enough to be effective across functions without sacrificing ease of use. It must be accessible without hand-holding, and it must be able to ingest huge volumes of data, sometimes in real-time, without SMEs manually tagging and annotating. There’s simply too much information coming in for a manual curation strategy. Pharma demands automation that works.
KASA is only the tip of the iceberg
KASA looms large in pharma, but it’s just the beginning. Many questions are yet to be answered (will KASA be rolled out to more than just generics? What’s the compliance timeline?) But it’s clear that the time is ripe to move to a structured knowledge management system. Regardless of specific initiatives, regardless of competition moving in this direction, your organization has much to be gained from an effective, streamlined, easy-to-access knowledge management system. Instead of spending time searching through messy data repositories and huge documents, your scientists can spend their time developing life-saving drugs. They’ll thank you.