One of the cool things about being part of technology industry is living through certain hype cycles. I have experienced, in past 20 years of my professional life, ups and downs that tend to get everybody riled up about ‘this is going to end the world’ dooms day scenarios like Y2K to ‘this can do everything’ conversations about things like Artificial Intelligence, Virtual Reality etc. I still remember back in the day when my friends in the Computer Science branch in college had subjects like Virtual Reality, Neural Networks and Artificial Intelligence. I am talking about 1994 – 1998, which is almost 2 decades+ to the day. Not many of them pursued their careers in those areas, but it is amazing to think that it took as long as it did to get to current developments in this space that is revolutionizing the consumer and enterprise software industry.
There have been many developments along the way that got us here from Internet to client / server technologies to cloud to big data to NLP, to name a few. Today we can genuinely claim there are things that machines can do, faster and cheaper and more than anything else, SMARTER than humans with very little to no intervention from us. As BigData hype reaches a plateau, Deep Learning is picking up steam and more and more companies are investing in this area to genuinely unleash the power of data through smarter analysis with help from neural networks, NLP, Deep Learning and the likes.
Having spent considerable amount of time dealing with Life Sciences industry for the past 14 years, I can speak to the utmost conservative approach these companies take when it comes to technology adoption. It is a heavily regulated industry and rightfully so since it deals with human lives. Life Sciences companies can release life saving ‘Elixirs’ but can also unleash ‘Drug from the Devil’. In my experience Life Sciences companies typically are 2 to 3 years behind in terms of technology adoption. This may change depending on the department within the value chain but this tends to be the average duration before they use the latest version of Windows or IE or Office.
I hope to highlight some of the use cases where newer technology developments can be leveraged in Drug/Device/Vaccine development, specifically in the areas of Regulatory and Safety in a series of posts starting with this one. I will try to prioritize areas where there is a lot of manual intervention (Compliance) as well as areas that could leverage technology to deliver faster ROI (increase Revenue) and improve Operational Excellence (Reduce Cost).
One of the first areas that I thought could benefit from these technological advances is Regulatory Intelligence. The EU Regulatory Intelligence Network Group (RING Europe) defines it as “Regulatory intelligence is the act of processing targeted information and data from multiple sources, analysing the data in its relevant context and generating a meaningful output – e.g. outlining risks and opportunities – to the regulatory strategy. The process is driven by business needs and linked to decisions and actions.” RI is a key part of Life Sciences industry primarily for three reasons:
- It is a heavily regulated industry
- If companies operate globally they ought to comply with ever changing regulations and
- Influence policy and advocacy of future development
Please refer to this presentation from Carol Hynes of GSK for more details on “Regulatory Intelligence: Implications for product development“.
Many organizations have built Regulatory Intelligence Repositories by collating information from various sources. The diagram below represents various sources of RI data (courtesy : Regulatory Intelligence 101 By Meredith Brown-Tuttle).
These repositories cannot be built overnight. They have to be collated piece-by-piece over a period of time. The sources could go beyond the ones identified in the above diagram. Also, the repository may contain structured as well as unstructured content and data. Extracting information from such repositories is typically not a straight forward process. It definitely will not be as easy as asking a colleague who would then manually conduct the research needed and collate the information that can then be circulated to one or more individuals in the Reg Affairs organization for consumption and decision making . Therefore leveraging Automation, Machine Learning and Natural Language Processing in order to glean into the information in such repositories will make the life of Regulatory Intelligence colleagues lot easier. They can easily query the repository in their language of preference (for regular users with NLP capabilities) or write No-SQL and Semantic queries (experienced/super users) to extract the relevant information.
Information thus obtained can be leveraged to put together documents, newsletters and other communication vehicles which in turn could be stored back in the repository thus continually enriching and expanding the wealth of information available. This idea can be extended to create a federation of such repositories (internal, external, partners, vendors etc.) that can be scoured for the necessary information. Also leveraging even more advanced technology advances like Deep Learning might enhance the effectiveness and Return on Investment even more.
As always, your feedback and comments are most welcome. Thank you.
I was looking at the FDA report from Sep 2014 titled “Standardizing and Evaluating Risk Evaluation and Mitigation Strategies (REMS)” to better understand the need for the new DRAFT guidance to submitting REMS using SPL, as outlined in the guidance document issued in Sep 2017 titled “Providing Regulatory Submissions in Electronic Format — Content of the Risk Evaluation and Mitigation Strategies Document Using Structured Product Labeling”. I felt it would be easier for people to understand the need for the guidance if they understand the backstory leading up to the new guidance. Please refer to the original report and subsequent guidance for more details.
Feedback to FDA, through various stakeholder engagement sessions participated by vendors, healthcare community and other impacted groups:
- Stakeholders are not uniformly impacted by REMS requirements
- Communication about REMS requirements should be improved
- There should be flexibility to implement a REMS program based on the nature and variety of health care settings.
- REMS are vital tools that will be increasingly necessary, and content delivery must be streamlined without compromising the content itself
- FDA should standardize REMS across platforms, media, and delivery technologies and work to fully integrate them into health care systems—which will increase access by both health care providers and patients, and enable improved assessments to further advance standardization.
- FDA should use human factor evaluation approaches like Failure Mode and Effects Analysis (FMEA) to support and standardize REMS program design.
- FDA can improve REMS assessments with a variety of tools and techniques
- FDA should structure and standardize REMS information.
Based on engagement with industry stakeholders and the feedback received, FDA prioritized 4 projects in four different areas:
- Patient Benefit/Risk Information under REMS
- Providing Patient Benefit/Risk Information by Improving Tools for Prescriber-to-Patient Counseling
- Prescriber Education under REMS:
- Prescriber Education—REMS and Continuing Education (CE) for Health Care Providers
- Pharmacy Systems under REMS:
- Standardizing REMS Information for Inclusion into Pharmacy Systems Using Structured Product Labeling (SPL)
- Practice Settings under REMS:
- Providing a Central Source of REMS Information for Practice Settings
Of these projects, the project titles “Pharmacy Systems under REMS” led to issuing the guidance that is issued in September 2017 and distributed for public comments. The objectives of this project were:
- Make structured REMS information available to health care providers, patients, and FDA.
- Provide a single conduit of comprehensive information about REMS programs.
- Facilitate the integration of REMS into pharmacy systems and health information technology, including systems for electronic prescribing.
- Improve the efficiency of FDA’s review of proposed REMS by allowing the Agency to receive REMS submissions in a consistent format.
- Support FDA’s ongoing REMS standardization efforts by enabling the cataloging of similarities and differences between REMS programs.
The project’s final deliverable was set to be a revised SPL Implementation Guide that describes how sponsors, health care information system developers, and other stakeholders can share REMS information leveraging the existing SPL standard.
To help address the concerns expressed by various stakeholders, FDA wants to take steps to streamline the access to REMS information
- FDA intends to require applicants of NDAs, ANDAs, and BLAs to submit the content of their REMS documents in Structured Product Labeling (SPL) format
- SPL can be used to capture and present REMS information in a format that is:
- Easily shared with stakeholders and
- Readily incorporated into health information technology
- Twenty-four months after the final version of the guidance is published in the Federal Register, applicants must submit REMS documents in electronic format consistent with the requirements set forth below:
- Types of submissions : NDAs, ANDAs and certain BLAs
- Also applies to all subsequent submissions including amendments, supplements, and reports, to the submission types identified above