Safety Data Integration and Search

Safety Data Integration & Search

PointCross's Safety Data Integration & Search (SDIS™) empowers scientists to search, navigate and "meta-analyze" normalized studies to uncover hidden patterns, highlight safety indicators, and identify potential drug candidates early to reduce late-stage attrition.

With SDIS™, scientists can perform meta-analysis across normalized in vivo, in vitro, clinical, CRO, and other partner study data; it also allows users to search and navigate 'contextualized' emails, documents, meta-data and other qualitative content related to each study or from external sources. SDIS™ interoperates with LIMS, other data acquisition systems, and CDMS to integrate both legacy as well as contemporaneous drug safety data in a single “dynamic repository.”

Watch SDIS™ in Action in the below demo

SDIS™ empowers scientists to

Search, navigate and "meta-analyze" semantically harmonized studies, emails, documents, meta-data and more across the entire organization, CROs and other partners
Normalize study data, units and taxonomies (including controlled vocabularies) using a flexible semantic data exchanger
Find correlations, patterns, trends, and outliers in multiple dimensions, including study types, findings, compounds, and animals/patients
Perform meta-analysis that controls for measurement and sampling errors which may otherwise distort signals within individual studies
Interact with search results through tabular views, graphical visualizations and thematic data browsers
Collaborate securely and contextually across teams and multiple organizations to accelerate early safety assessments and generate new scientific insights
Make better and faster strategic and pipeline decisions, identify successful drug candidates earlier, and reduce costly late-state attrition

To schedule a complimentary live overview of SDIS™, contact Dr. Shree Nath, VP of Pharmaceuticals, at Shree.Nath@pointcross.com or call us at (650) 350-1900.

SDIS™ Components & Attendant Solutions

Semantic Data Exchanger (SDE™) an any-to-any data converter that semantically normalizes and unifies data from multiple sources (systems, formats and geographies) into a common Universal Operational Data Model (UODM). SDE ensures a common unified information base for search and orienteering. Learn more in our white paper on PointCross's Semantic Data Exchanger™, including its relevancy to CDISC data standards adoption, as well as collaborative data sharing initiatives like IMI.

Contextualization of Legacy Content with Free Text Search for qualitative data like animal/patient details, study/trial data, site visit observations, as well as unstructured content (protocols, reports and related collateral). Facilities are provided to contextualize legacy content including electronic and paper versions from internal/external sources, and to harvest related data and metadata through federation. Also enabled is the ability to meta-tag all such information using a rich set of taxonomies including folksonomies, A.K.As, and controlled or industry standard terminologies. Learn more about this capability in our white paper on Contextualizing Legacy Information for Pharma Companies.

Metadata character search for rapid queries that scan metadata within contexts (business topics), email, document, meetings and selected processes for matching character string patterns. Results are automatically contextually organized along multiple facets.

Full Content search through embedded Lucene/SOLR search capabilities that harvest information across multiple content types; other search engines including Microsoft FAST, Google and others are readily supported.

Parametric Search across quantitative data such as animal/patients observations, clinical pathology, hematology, other laboratory measurements and others. This includes search across not only raw data, but automatically extracted statistical metadata and associated patterns within and across studies. These statistical indexes are created using a wide variety of statistical analytics including linear and nonlinear modeling, regression analysis, classical statistical tests, time-series analysis, classification, and clustering.

Security and Authorization for managing people and role access on a granular, need-to-know basis. Rather than relying solely on Access Control Lists for individual applications, the solution uses a context-based approach to harmonize authorization in line with business needs.

Blind Search facilities trigger workflows to authorized users who can choose to grant information access to others who may not otherwise be able to tap into it.

Collaborative controls to interact with peers in context with live and saved search results. All collaborative facilities directly integrate with common enterprise applications such as Microsoft SharePoint and Office.

Governance and Compliance Services for importing and normalizing legacy data and documents including dashboards and controls for reconciling data discrepancies, applying transformation rules, and capturing annotations to ensure full audit traceability back to the original source.

Benefits of SDIS™

Empowers scientists and pharmaceutical companies to identify potential drug candidates, and failures, in a fraction of time and drive better decision making and reduced cost by lowing the attrition rate in later drug developmental phases;
Identifies and establishes common objects and concepts, and resolves conflicts between different sources by meta-tagging based on Controlled Terminologies;
Leverages contextualized meta-data concepts while integrating free text search to provide visual cues of context for organization of results as well as for subsequent queries.
Dashboards reduce time that researchers spend searching for information, and allow them instead to better understand patterns and correlations;
Combines search with statistical analytics across raw data and automatically extracted meta-data to improve research productivity and serendipitous discovery of secondary beneficial effects
Expedites therapeutic efficacy of a drug molecule from among a heap of studies by enabling scientists to perform meta-analysis and draw more definitive conclusions sooner.

March 3rd 2010 Webinar Q&A - Cross-Study Search and Meta-Analysis with SDIS™

Question: Do you have examples on how such a system could assist sponsor companies?

Answer: Yes, a specific case regarding an anti-viral drug candidate comes to mind where safety data most of the animal models suggested no safety implications. Not having the ability to go across silos of data and across all animal models caused the drug candidate to move forward at an eventual cost of several million dollars, not to mention lost time and opportunity. As it turned out, signs of nephro-toxicity were discernible in the in vivo studies for rabbits and had to be extracted manually and with difficulty during retrospective analyses. A system like SDIS™ with built-in controls for multiple biomarker panels would have helped reduced risks, cost and time sooner in the drug development program.

Question: Relative to the last example of species specific renal toxicity, how readily can data from in-license compounds, where access to source data maybe be limited, be added to this database?

Answer: Adding in-licensed drug data into the environment is certainly possible; the same approach of extracting, normalizing and indexing such source data would apply. The degrees of freedom on searchability will naturally be restricted to what is available. However, by combining structured search along with unstructured (documents, other collateral) search and specifically looking for patterns, it should be possible for scientists to better connect the dots while evaluating the potential in-licensing candidate.

Question: How does SDIS™ fit in with the SEND initiative, and specifically with regards to data exchange strategies moving forward?

Answer: SDIS™ is a natural fit with SEND since it provides a much broader opportunity to not only comply with anticipated study data submissions to the FDA, but to provide sponsors with decision support tools to look across studies within a particular drug candidate program (and/or related programs) to pick up safety signals sooner, to design new studies where necessary, and possibly to pick up beneficial secondary effects sooner. In particular, connecting clinical, pharmaco-vigilance and non-clinical data in the future, across both structured and unstructured content, can be very powerful to drive drug development strategies and support “fail fast” initiatives for portfolio optimization.

SDIS™ coupled with SEND data submissions also offers regulatory agencies an opportunity to improve and accelerate review processes.

With regards to data exchange with CROs, PointCross believes that SEND is an important first step in creating a dialog between sponsors and their CROs. However, it is unlikely to meet the true business intent of sponsors due to the following reasons:

The SEND data model (currently version 3.0a) will continue to evolve based on Agency, sponsors, CROs and CDISC discussions. The model at the current moment does not encapsulate all of the data being captured in studies today, nor does it handle analytics data. Limiting data exchange to the extant SEND data model may cause less study data to be transferred to sponsors than may be required in the future for a subsequent regulatory submission. In practice, this will require a repeat submission with necessary quality assurance and compliance checks, not to mention lost time and opportunity. Instead, sponsors should consider a more universal data model embracing all of the study data which can then be re-purposed with traceability into the necessary SEND data model OR any other data exchange models specified by regulatory agencies. This will ensure that sponsors have access to all the data for search and analytics, both to support regulatory interactions, as well as for e-Discovery, Audit Traceability and Records Management initiatives.
The SEND data model continues to be based on a legacy SAS transport (XPT) format which has well known limitations. It is far more practical to embrace an XML format for CRO to sponsor data exchange, and again re-purpose this universal and open format into XPT or other formats as required. In all instances, a validation layer should be provided to ensure minimum compliance with requirements of SEND (or other regulatory data model specifications).

Question: Does the system interface with statistical tools like SAS?

Answer: Yes, SDIS™ provides interfaces to SAS, and can export data in formats that can be accepted by SAS (or other statistical tools). The underlying Orchestra platform on which SDIS™ is based also includes the open source R Project statistical toolkit that can be used as an alternative. This approach has the advantage of combining search, data exchange, and analytics facilities with traceability within one unified environment.

Question: Can SDIS™ provide the user with an indication of matches as queries/search criteria are entered? Specifically, do you get an idea of the size of the result set WHILE you're creating the search criteria? (e.g., it is no use to be more precise of you already have 0 results; but if you still have 700 results, it's worthwhile proceeding to further narrow down the results)

Answer: Yes, this kind of meta-data is already available. Some of the meta-data is already used to drive the Results Cluster (left side tree) for quick navigation and orienteering. We will shortly be releasing (late Q2, 2010) a version of SDIS™ where new usability and ergonomic controls for displaying search matches will be provided to the user for increased awareness as s/he build the search criteria.

In this context, we would like to point out that meta-data is continuously being created and added into SDIS™ with new data and as people execute, use and collaborate around search results. For instance, SDIS™ automatically extracts trends and correlations as new content is populated into the environment.

It is our opinion that the key to advanced search and discovery facilities lies in the meta-data layer which allows otherwise hidden patterns to surface as "meta-analysis" across studies is being conducted.

Question: Can SDIS™ provide summary “bread crumbs” (snippets of matched search criteria) and enhanced relevancy for unstructured document and other text search?

Answer: Yes, these facilities along with concept-driven controls are being added to the search results along with controls to specifically navigate through results sets using the same results cluster approach demonstrated for structured search. These controls and views are readily configurable based on business needs.

Question: Are there plans to integrate with Google Scholar (the SCI Google) so that you can also look at external search results and compare with internal ones?

Answer: We have not specifically looked at Google Scholar. However, as demonstrated, the platform provides standard connectors to external sources and we can certainly integrate with Google Scholar or other external sources where necessary.

Question: To pull up sub-chronic studies, was it a text search that assumed the study title had "sub-chronic" in it, or would it also find a study that was titled, e.g., "a 3-month study..."?

Answer: Search is not limited to the study title. The content of the document, matches to local or global/controlled vocabularies, and other meta-data (content usage, communications around a document, and users who accessed the document) are all indexed and are searchable.

Question: Does the technology provide compliance controls with the ability to identify issues in the data sources? In other words, is it a read only system or a read/write or update system?

Answer: All data extraction, normalization and transformation facilities are audit traceable and the underlying Orchestra platform on which SDIS™ is built provides the necessary controls to ensure requisite validation and compliance, along with links to the original data source.

Additionally, annotation capabilities are available to allow exclusion or other flags to be applied to the source data as meta-data, keeping the original data intact. In short, the original data is ALWAYS kept intact, and addition facilities are available to annotate or transform the data with traceability.

Question: I am interested in how you upload/compile data into the system, and effort spent to upload data?

Answer: We use a set of extensible toolkits to manage the semantic data exchange process, as well as all the necessary normalization that is required for data, taxonomies and units. The toolkits also permit extensions to the underlying XML based universal data model and validation rules for specific systems.

The level of effort depends on the type of data/system, availability of source system dictionaries and some expertise to support questions that arise. As an example, initial integration with new systems can take in the order of 2-4 weeks; additional quality assurance, bulk import and clean-up activities can take anywhere from 12-16 weeks for each major system. The size of the teams involved can provide further acceleration (e.g,., by combining multiple systems in parallel tracks). Further controls for compliance validation are client-specific and it is difficult to predict the required duration. Our experience has been that we execute these types of programs anywhere from 5X to 10X faster compared to standard technology providers because of the extensive use of tools instead of just people.

Question: Does PointCross offer additional data visualization tools to look at other types of data and clinical outcomes not related to safety?

Answer: Yes, we do have some initiatives on this front for various kinds of R&D, competitive intelligence, analytics and other types of information. We also do quite a bit of visualization on the Oil and Gas side of our global business with Exploration and Production companies. Please contact shree.nath@pointcross.comto discuss further.

Contact

For more information, and to schedule a complimentary live demonstration of SDIS™, contact:

Dr. Shree Nath, VP of Pharmaceuticals,

E-mail: Shree.Nath@pointcross.com
or call us at (650) 350-1900.