How Biopharmaceutical Companies Can Fish Relevant Information From A Sea Of Data

This article originally appeared in Bio-IT World

Content and data in the biopharmaceutical industry are complex and growing at an exponential rate. Terabytes from research and development, testing, lab reports, and patients reside in sources such as databases, emails, scientific publications, and medical records. Information that could be crucial to research can be found in emails, videos, recorded patient interviews, and social media.

school-of-fish

Extracting usable information from what’s available represents a tremendous opportunity, but the sheer volume presents a challenge as well. Add to that challenge the size of biopharmaceutical companies, with tens of thousands of R&D experts often distributed around the world, and the plethora of regulations that the industry must adhere to—and it’s difficult to see how anyone could bring all of that content and data together to make sense of it.

Information instrumental to developing the next blockbuster drug might be hidden anywhere, buried in a multitude of silos throughout the organization.

Companies that leverage automation to sift through all their content and data, in all its complexity and volume, to find relevant information have an edge in researching and developing new drugs and conducting clinical trials.

This is simply not a task that can be tackled by humans alone—there is just too much to go through. And common keyword searches are not enough, as they won’t tell you that a paper is relevant if the search terms don’t appear in it, or if a video has the answer unless the keywords are in the metadata of the video.

Today, companies can get help from insight engines, which leverage a combination of sophisticated indexing, artificial intelligence, and natural language processing for linguistic and semantic analyses to identify what a text is about, look for synonyms and extract related concepts. Gartner notes that insight engines, “enable richer indexes, more complex queries, elaborated relevancy methods, and multiple touchpoints for the delivery of data (for machines) and information (for people).” A proper insight engine does this at speed, across languages, and in all kinds of media.

For biopharmaceuticals, this is particularly powerful, allowing them to correlate and share research in all forms over widely distributed research teams. Here are several ways biopharma companies can use insight engines to accelerate their research.

Find A Network Of Experts

Many companies struggle to create the best teams for new projects because expertise is hidden in large, geographically-distributed organizations with multiple divisions. A drug repositioning project might require a range of experts on related drugs, molecules, and their mechanisms of action, medical experts, geneticists, and biochemists. Identifying those experts within a vast organization can be challenging. But insight engines can analyze thousands of documents and other digital artifacts to see who has experience with relevant projects.

The technology can go further, identifying which experts’ work is connected. If they appear together in a document, interact within a forum, or even communicate significantly via email, an insight engine can see that connection and deduce that the work is related. Companies can then create an “expert graph” of people whose work intersects to build future teams.

This technique can extend beyond the borders of the company, helping to identify the most promising collaboration partners outside the company in a given field, based on publicly available data, such as trial reports, patent filings and reports from previous collaboration projects.

Generate R&D News Alerts

Biopharma companies can also use insight engines to watch for new developments in drug research and stay on top of the latest trends. These news alerts can go beyond typical media sources to include scientific publications, clinical trial reports, and patent filings.

This capability can be used on SharePoint, Documentum, or other sources within a large company to surface relevant information. An insight engine ensures the right information gets to the right people in the right context, and in a timely way.

Optimize Clinical Trials

Clinical trials that stretch over many years generate millions of datasets for every drug and study provide a treasure trove of data. Biostatisticians can ensure they get a comprehensive list of patients having certain diseases within trials on a drug, something nearly impossible with traditional methods.

They can also search and analyze across many drugs and studies, across content and data silos. Over time, this allows biopharmaceutical companies’ growing number of clinical trials to become a valuable asset that can be easily leveraged across a growing number of use cases.

All of these uses can lead to biopharma companies developing new drugs more quickly and getting them to market faster—necessary as these companies face tremendous pressure to innovate quickly and develop new promising drugs as patents for older drugs expire. With insight engines, they can make every part of the journey more efficient, from research, to clinical trials, to regulatory processes, presenting incredible opportunities for everyone in this field.

 

+1Share on LinkedInShare on Twitter

3 Ways Manufacturers Can Leverage Insight Engines

This article was originally published on Manufacturing.net.

As distributed manufacturing gains adoption by some of the world’s largest companies, products are finding their way into the hands of customers faster than ever before. But in the process, companies are becoming increasingly disconnected and less efficient in new areas.  Unlike in traditional manufacturing, where materials are assembled in remote, centralized factories and shipped to customers, distributed manufacturing takes place at multiple, decentralized plants where products are assembled closer to the customer.

Global manufacturing

The process of leveraging a network of geographically dispersed manufacturing facilities connected over the Internet helps eliminate many of the inefficiencies of traditional manufacturing. However, distributed manufacturing creates new challenges in the form of separate entities, languages and processes.

Teams once gathered under a single roof are now spread across multiple sites, working on projects and programs unique to their local area. These distributed teams are inadvertently scattering knowledge and data across continents and systems, creating information silos that leave workers disconnected.

Content and data have been scattered across a myriad of applications and repositories without any means for end users to surface accurate information with any confidence. In some cases, access to critical institutional knowledge, insights and innovation is restricted to a small few. As a result, disparate teams are spending countless hours making and solving the same problem over and over again, lowering productivity and delaying projects.

A digital solutions manager at a global manufacturing firm once told me, “We had thousands and thousands and thousands of places where our documents were shared, managed by check-in, check-out. In fact, we once had thousands of applications with no search capabilities.”

But once information is accessible from a single place, there are no longer countless teams scavenging for data and insights. With insight engines, manufacturers are bringing the best people together for new projects. Critical parts are easier to locate instead of rebuilding or re-designing them. Even proposals and RFP responses are being developed faster with information found and shared from other RFPs.

These improvements are the result of insight engines that combine cognitive analytics, artificial intelligence and NLP (natural language processing) to helps computers understand, interpret and manipulate human languages, dialects and patterns.

This technology helps manufacturers quickly respond to changing conditions and identify products, parts and components across multiple data sources despite their distance from one another. Here are three ways manufacturers can leverage insight engines to rediscover their synergies and avoid mistakes.

Building Teams Through a Patchwork of Data

Many companies struggle to create the best teams for new projects because expertise is hidden in large organizations with multiple divisions and product groups. Sometimes this task is made more difficult by employees who fail to enter their background and experience in their HR profiles or update them with new skills or certifications.

AI-powered insight engines leverage NLP to surface information from large volumes of data stored in hundreds of millions of files, and thousands of repositories and databases to uncover work histories, experience on different projects, expertise, training and education in order to locate experts. Projects employees have worked on, languages they write in, and locations where they’ve traveled are all easily queried and analyzed to identify experts within the company.

Avoiding Duplication of New Parts

In a distributed environment, it’s difficult to find parts across systems and continents. Sometimes it’s easier to make new ones. That’s because parts are generally indexed by labels, which can be difficult to identify correctly. As a result, the exact same part is recreated, sometimes up to five times, each version with its own label.

This is a complete waste of time and money. And when a part is recreated, there’s a greater chance of a new flaw being introduced, which can be catastrophic in industries like transportation or medical devices.

With insight engines, parts data can be contextualized for engineers to identify duplicate parts and eliminate duplications. For any company with a broad base of engineers, the cost savings from this approach can be enormous.

Increasing the efficiency of proposal development

Companies with multiple departments and divisions often fail to share technical information for sales proposals and RFPs. Consequently, proposals for an opportunity in one part of the world with many of the same requirements as others in different markets or regions must be developed from scratch.

And, if questions arise during the proposal process, it is time-consuming and difficult to get answers from engineers. Acquiring new customers in a complex environment with a highly engineered, configurable project is time-consuming by its nature. The inaccessibility of data makes it worse.

Insight engines surface information in context across emails, databases, and records. With this information in hand, companies can respond faster to sales opportunities and present a common face for the company overall.

In all three of these scenarios, information and insights are securely surfaced from content and data in every application and every repository from every location across globally distributed companies, which enables teams to cultivate, improve quality and take advantage of new opportunities.

+1Share on LinkedInShare on Twitter

Sinequa 2018 Roundup… 2019 Here We Come!

2018Sometimes it helps to look at an entire year to gauge just how far you’ve come in a relatively short period of time. Sinequa experienced some very positive developments in 2018 that are worth highlighting. Our software platform evolved on several fronts to help us accelerate our mission to power the information-driven economy. In parallel, our customers demonstrated what the platform can do, even when stretched in creative and unexpected ways.

On the Technology Front

The Sinequa platform evolved with some very useful and powerful new capabilities in 2018.

Content-related Capabilities

Many of the new capabilities improved on the platform’s ability to integrate with even more enterprise applications and content formats, including:

  • New connectors to support the goal of ubiquitous connectivity across the enterprise. Among these were connectors to Atlassian products to incorporate information from software development projects, including source code files. Also addressed were new versions of popular repositories like SiteCore (a leading web content management platform according to Gartner), along with the likes of Azure storage, AODocs, Beezy, and Teamcenter.
  • New converters to index more formats like OCR on PDF and Images, AutoCAD and Windchill files, Visio, Improvements on PowerPoint, and a dedicated converter for source code files
  • Tighter integration with SalesForce.com
  • In a year full of major data privacy breaches being reported, the Sinequa platform continued to strengthen support of additional levels of encryption like in-flight encryption between all components in a distributed deployment and encryption at indexing time to secure the document cache, which contains elements like HTML preview and thumbnails to better serve customers operating in highly secure environments

Further Automation for the Interpretation of Meaning

The platform’s ability to interpret the meaning of content also evolved in 2018.

  • Query Intent: It is now possible to configure rules to be applied on queries to change the behavior of the underlying search process. This new query intent capability analyzes the query to detect certain words and entities and triggers actions based on the specified rules and classifications. New default entities were also introduced in the platform in 2018 that can be leveraged by the query intent capability and for enrichment during indexing.
  • Enhanced Linguistics: There were some language-specific improvements added to the platform to help automate the interpretation of meaning. These included things like enhanced linguistic processing for compound words in French, improved lexical disambiguation in English, enhanced detection of ordinal numbers for Danish & Swedish.

Improvements in Machine Learning

The year 2018 brought several significant improvements in the Sinequa platform’s ability to leverage machine learning, including:

  • The platform evolved to embed Online Machine Learning, applying machine learning models based on Spark or TensorFlow directly in the indexing pipeline. This represents the first of many new components that can serve machine learning models in real time. Deep learning is also used during indexing to detect new entities or concepts. These are immediately fed into machine learning algorithms, for example in the classification of incoming documents.
  • Packaged with the platform is a new unsupervised Deep Learning application for text analysis that detects the key words, key phrases, and key sentences of a document.
  • The platform now supports the Spark 2.3 implementation.
  • Packaged integration with 3rd party spark distribution providers – e.g. AWS EMR, HortonWorks.
  • Battle testing of supervised classification algorithms – i.e. Sinequa reached a threshold training set size over 10M documents
  • First machine learning customers are now in production
  • Packaging of hierarchical classification
  • Ongoing transition to Software 2.0 paradigm where software is effectively “trained” rather than manually programmed with the packaging of the lifecycle of the model and the model feedback from the search based applications into the Sinequa platform.

Presentation Enhancements for End Users and Admins

Sinequa invested significantly during 2018 to evolve the way the platform presents insights to end users as well as status information and optional settings to administrators. Here are a couple of the most significant developments:

  • A very exciting development from 2018 involved a complete overhaul of the user interface framework to a responsive design based on Angular 7. This will not only ensure optimal flexibility and performance for end users on all kinds of devices, but will open up Sinequa application development to a much wider audience.
  • On the Admin front, components have been reshaped to offer administrators of the platform more functionality and a better user experience for their work behind the scenes.

On the Customer Front

There were a few compelling themes driven by our customer base in 2018, each of which was rewarding in its own way.

Customer satisfaction and retention is a predominant theme for Sinequa. We are extremely pleased by the sheer number of existing satisfied customers that came back to us in 2018 with additional use cases to accelerate their information-driven journey. For instance, business drivers related to governance, risk and compliance with the advent of GDPR and related regulatory demands spurred a lot of activity this past year.

We also had a significant number of customers who experienced that “light bulb moment”, which often occurs when they realize their existing return on Sinequa investment could be significantly amplified by extending the use of the platform with information-driven applications in other parts of the business – e.g. areas like customer service, R&D, supply chain, and other knowledge-intensive arenas.

We even had a few long-time customers take a pause to re-evaluate their vendor choice and, without exception, decided to double-down on their commitment to Sinequa for years to come.

Of course, the disappearance of the Google Search Appliance brought some new customers into the fold, most of them fiercely determined to go beyond their previous use of a dying application and truly become information-driven.

Possibly the single most exciting development for Sinequa in 2018 was the surge in machine learning projects, which contributed significant business value back to the respective organizations, especially in the Financial Services industry. As the underlying technology matures, we see a steady trend for machine learning projects going from research to production stages. Some of the projects from 2018 focused on applying machine learning models to automate the curation of enterprise content and improve relevance. For example, one customer demonstrated how trained machine learning models could be used to make the enrichment of their enterprise corpus more efficient. It turns out that by proactively identifying what content qualifies as “scientific”, both time and money can be saved by preventing non-scientific content from even being considered for scientific enrichment during ingestion. Another customer took a completely different tack, using machine learning to automatically reproduce confidentiality policies to classify large volumes of banking documents with measurably higher quality at a fraction of the cost they would have spent to do it manually or even with a more traditional rules-based approach.

Now it’s on to 2019!

As we turn the corner into 2019, we are grateful for both the accomplishments of our R&D team and for all of our partners and customers, especially those who provide the challenges, creativity and critical feedback necessary for Sinequa to continue providing the leading platform for information-driven applications and solutions.

We wish you all the best and look forward to serving all of you in 2019 and beyond.

+1Share on LinkedInShare on Twitter

Ferring Pharmaceuticals selects Sinequa and Atos to boost Global Cognitive Search Capabilities for R&D

ferring-logoAs a testament of Sinequa’s fast growing footprint among leading life sciences organizations, we are very excited to announce that Ferring Pharmaceuticals selected Sinequa and Atos to boost its global cognitive search capabilities. Sinequa’s Cognitive Search & Analytics solution was recently deployed at Ferring Pharmaceuticals with Atos as the consulting and integration partner to empower the organization’s Global Pharmaceutical R&D group to look deeply into vast scientific research data sets in order to generate new insights and accelerate innovation

Headquartered in Switzerland, Ferring Pharmaceuticals is a research-driven, specialty biopharmaceutical group active in global markets. A leader in reproductive and maternal health, Ferring has been developing treatments for mothers and babies for over 50 years. Today, over one third of the company’s research and development investment goes towards finding innovative treatments to help mothers and babies, from conception to birth. Ferring has its own operating subsidiaries in nearly 60 countries and markets its products in 110 countries.

In today’s world, especially in the life sciences industry, it is impossible for humans alone to search, process and analyze all the world’s available scientific and research data, Sinequa’s Cognitive Search & Analytics platform  makes this scientific knowledge accessible any time by any given researchers. As Sinequa continues to expand its footprint in this very competitive industry, we are very pleased to count Ferring Pharmaceuticals among our customers. Together with our partner Atos, we are committed to help Ferring improve insights and facilitate innovation.

- Stéphane Kirchacker, vice president Sales, EMEA at Sinequa

test-tubes

Atos, Sinequa’s strategic global premium partner was selected to design, implement, support and operate the platform at Ferring Pharmaceuticals to deliver the highest possible relevancies on search for the R&D teams in different locations.

Sinequa’s solution is bringing the power of AI to Enterprise Search to provide Ferring a “future proof” solution that offers a whole range of opportunities for future innovations. The dilemma of pharmaceuticals is to find the needle in the haystack – scientists need to screen tens of millions of documents from internal and external sources, from structured and unstructured data for identifying relations between genes, drugs, Mechanism of Action (MoA) and finding the right skilled subject matter experts. Other departments like Regulatory & Compliance, Legal & IP, Marketing & Sales, Clinical Trials, HR and more can benefit from customized Search-based applications on the same platform – finding relevant information instantly for fact-based decisions – no waste of time anymore.

-Alex Halbeisen, Expert Sales Big Data & Analytics at Atos

 

+1Share on LinkedInShare on Twitter

How organizations can evolve from data-driven to information-driven

This article was originally published on Information Management.

Over the last several years, data analytics has become a driving force for organizations wanting to make informed decisions about their businesses and their customers.
With further advancements in open source analytic tools, faster storage and database performance and the advent of sensors and IoT, IDC predicts the big data analytics market is on track to become a $200 billion industry by the end of this decade.
+1Share on LinkedInShare on Twitter