Palantir » Analysis Blog

Palantir Pharma: mitigating R&D risk through data fusion

Andrew — Mon, 11 Feb 2013 17:49:19 +0000

With development times of ten years or more and costs of over $1 billion per new medicine, pharmaceutical R&D is an expensive, lengthy, and risky process. Enterprises can minimize the risk associated with drug development programs by better understanding the universe of data relevant to particular diseases, targets, and drug candidates.

Unfortunately, achieving this level of understanding is technically challenging, as the data is typically scattered across public and private databases and stored in a wide range of formats. Furthermore, novel research data is noisy and can be fraught with contradictions. Serious data security and access control concerns only add to this complexity.

The Palantir Gotham data fusion platform helps pharmaceutical companies overcome these R&D challenges by integrating public and private data of nearly any type and empowering secure collaboration both within and beyond the enterprise.

In this three-part series of videos, you will see how the Palantir Gotham platform facilitates rapid, intuitive, and comprehensive exploration across a variety of data sets. We focus on just a handful of ways in which Palantir removes the friction between pharmaceutical researchers and their data, allowing them to (1) discover connections between assays performed by different teams, (2) evaluate evidence for a drug-protein interaction, and (3) capture the investigative process for future use.

For these demonstrations we have integrated data from public sources including ChEMBL, PubMed, KEGG, PDB, and PubChem. From high-level literature reviews to the analysis of specific variations in a genetic sequence, Palantir Gotham increases efficiency in the drug development process and amplifies signal strength for key pipeline decisions.

Part 1: To start, we use Palantir Gotham to collaboratively model biological relationships and assess their certainty in order to better understand the risk associated with targeting a particular pathway.

Part 2: Next, we explore new experimental results that were produced by an external team. We investigate their underlying scientific rationale and apply these insights to my own drug development research.

Part 3: Finally, we investigate structured assay data and quickly drill down on a compound of interest. We use a variety of techniques to make sense of complex experimental results in both structured and unstructured formats.

Drug development is just one of many potential pharmaceutical applications of the Palantir Gotham platform. Our goal is to solve your organization’s most challenging data problems across the pharmaceutical pipeline. To learn more, read about our Pharma solution and contact our health team: helix@palantir.com.

Palantir Cyber: Uncovering malicious behavior at petabyte scale

Ryan — Thu, 20 Dec 2012 18:48:44 +0000

One of the most difficult challenges for cyber security analysts is navigating through vast quantities of network data, which can approach petabyte scales and is often distributed across many disconnected systems. In this demonstration, we show how an analyst can use the Palantir Cyber solution to detect beaconing, a network behavior suggestive of malware, by querying multiple databases at a large institution in a matter of seconds. As fraudulent patterns are uncovered, analysts can automate these searches into regularly run jobs, serving as proactive alerts of malicious activity that are fed into our new prioritized inbox interface, powered by Hadoop. Finally, these alerts can be shared between analysts through Palantir Gotham’s collaboration application, which enables the rapid exchange of information within and across institutions to diminish cyber security threats.*

*While this demonstration is based on a typical investigation workflow, the data is simulated and names were randomly generated. Any resemblance to real people or entities is coincidental.

Adaptive Management and the Analysis of California’s Water Resources

Dustin — Fri, 07 Sep 2012 01:08:37 +0000

Water resource management in California is a precarious and costly balancing act. Various federal, state, and municipal organizations have a stake in the management of California’s water resources. In the case of the Sacramento River Delta, they all compete to manage a single resource. Decisions made about the Delta affect millions of Californians, as well as the endangered species in the Delta’s delicate estuarial ecosystem, such as the Delta smelt. It is therefore critical that these decisions be based on transparent, reproducible, and comparable analyses of the best available data.

In this demonstration developed by Palantir and environmental consultants from NewFields, we show how Palantir’s data fusion platforms can help tackle different facets of the adaptive resource management problem. We use the Palantir Gotham platform to map out the relationships of the various organizations managing the Delta, as well as the documents they publish and the data sources they maintain. With Palantir Metropolis, we use data from monitoring stations that are scattered throughout the Delta to analyze relationships between smelt abundance, salinity (and an associated metric called X2), and other physical factors in the Delta such as temperature and turbidity (cloudiness of the water). The Palantir Metropolis platform offers a means to compare scientific analyses at the high level of granularity needed to make critical management decisions. Users can conduct and modify competing analyses side by side to easily see where different models or underlying data diverge and lead to different conclusions.

The Chart application in Palantir Metropolis allows users to share their analysis and conclusions, quickly and easily. In this case, an analyst displays the changes in water salinity over time.

This kind of analysis can give policy makers maximum insight into the relationships between the variables that affect the Delta’s health and allow them to make decisions that appropriately weigh the interests of all parties involved.

Securely collaborating across the enterprise and with external partners to expose cyber fraud

Pooja — Mon, 20 Aug 2012 18:54:56 +0000

In an earlier demonstration on this blog, we showed how a single analyst used Palantir Metropolis to uncover an actual cyber threat at one of Palantir’s largest commercial deployments. However, in many large financial institutions, detecting complicated schemes requires the work of multiple analysts across the enterprise. Collaboration is critical, but the need to enforce data access restrictions can impede cooperative analysis across groups. In response to this need, Palantir has made secure information sharing a possibility within the organization and with external community members. Watch as we demonstrate how multiple analysts at one of the world’s largest financial institutions can collaborate to expose cyber fraud.*

*While this demonstration is based on a real investigation workflow, the data has been anonymized, and any resemblance to real people or entities is coincidental.

Palantir Legal Intelligence: analyzing the Enron emails

Kyle — Wed, 11 Jul 2012 19:30:48 +0000

On December 2, 2001 Enron Corporation entered the largest bankruptcy in US history at the time, and left behind a collection of hundreds of thousands of e-mails, which is the largest legal data set currently available to the public. Here we present Palantir Gotham’s ability to combine structured data (in this case financial transaction records and company org charts) with unstructured data (in the form of e-mails and documents) in an analysis of accusations of insider trading against the CEO of Enron Energy Services.

Using Palantir to Explore Prescription Drug Safety

Jesse — Mon, 26 Mar 2012 18:36:27 +0000

Drug safety is a serious concern in the United States with adverse drug events contributing to over 770,000 injuries and deaths per year. Cost estimates range from $1.5 to $5.6 billion annually. The FDA closely monitors these adverse events and releases communications and advisories depending on the severity and frequency of the events. The FDA released such a communication regarding the drug Simvastatin in June 2011. Simvastatin, which is used to treat hyperlidemia, is one of the most heavily prescribed medications in the world, and nearly 100 million prescriptions were written for patients in 2010.

This demonstration draws on data from the FDA, Practice Fusion, and the National Library of Medicine to uncover how a drug communication can efficiently be applied to electronic medical records (EMR) within a health network. Through a variety of techniques, at-risk patients can be quickly found and notified using the Palantir platform.

Palantir as a Program Management Platform: Examining Hurricane Katrina Acquisition Spending Data

Jesse — Fri, 09 Mar 2012 01:24:57 +0000

Hurricane Katrina caused 1,833 deaths and $108 billion in damage, making it the deadliest and costliest hurricane in American history. When a collection of federal, state, and local agencies converged to respond to the crisis, they found they lacked analytic tools capable of tracking and responding to the dramatic volume and scope of relief needs as they developed.

Using only open source spending data from Congress’ Federal Procurement Data System (FPDS), a Palantir analyst spent one week examining Katrina-related spending from 2005 to the present, with particular focus on the spending of the Federal Emergency Management Agency (FEMA). The analyst used Palantir’s geo-spatial, temporal, and relational analysis to add entirely new layers of depth to the FPDS Data.

OSINT Analysis of Sudan and South Sudan

Jason — Mon, 28 Nov 2011 21:59:35 +0000

Less than four months ago, the Southern portion of Sudan seceded and formed South Sudan, only the 5th country to be created this century. In this session, we will demonstrate how Palantir can draw from a plethora of Open Source Intelligence (OSINT) data sources (including academic research, blogs, news media, NGO reports and United Nations studies) to rapidly construct an understanding of the conflict underlying this somewhat anomalous 21st Century event. Using a suite of Palantir Helpers developed for OSINT analysis, the video performs relational, temporal, statistical, geospatial, and social network analysis of over a dozen open sources of data.

Note: The key analysis portion of the video begins at about 2:10.

Using Palantir to Address Information Security and Insider Threat in the Enterprise

Brandon — Wed, 19 Oct 2011 19:08:52 +0000

In this demonstration, we will show how an analyst can use Palantir to uncover insider threat. Using the integrated data and a combination of analysis tools available on the platform, the investigator can seamlessly cross data sets, examine activity, and rapidly reveal suspicious employee behavior in an information security fraud case.

Note: Data used in this video is simulated; any connection to actual individuals is purely coincidental.

Uncovering Cyberfraud at a Large Financial Institution

Brandon — Mon, 12 Sep 2011 20:25:28 +0000

This cyberfraud workflow is based on an actual case discovered at one of Palantir’s largest and most successful commercial deployments. In this demonstration, we will show how an investigator uses Palantir to rapidly surf across data from multiple lines of business generated through customer interactions via multiple channels. The investigator is able to use a combination of analysis tools available on the platform to quickly trace the origin of a reported threat and protect the bank’s assets from further exfiltration.