NEW YORK (GenomeWeb) – Researchers in Philips' healthcare division are currently testing a software solution for combining and analyzing genomics and other kinds of data in both clinical and research contexts ahead of a full launch planned for the fall of this year.
The Philips Genome Informatics platform is part of the company's cloud-based HealthSuite Digital platform, a set of Philips-developed applications for capturing and automatically linking patients' healthcare and lifestyle data including laboratory results, pathology images, clinical history, and more, Nevenka Dimitrova, chief technology officer for genomics in the Philips Healthcare division, told GenomeWeb this week.
The genome analysis software is currently being tested as part of an ongoing early-access program in a number of unnamed hospitals and research labs. The program, which started earlier this year, is focused on evaluating and improving users' experience and analysis workflows as well information flow within the system, she said.
When it launches, Philips will offer two iterations of the software, one that provides capabilities specifically for clinicians, and another that provides supporting tools for research projects. For the clinical version of the software, Dimitrova said that Philips plans to charge per patient analyzed. For the research software, customers will have to purchase licenses and the exact costs will depend on usage. She could not provide specifics on the different price points, but she said that costs would be "competitive" with prices for other informatics tools available on the market.
The clinical version of the platform lets users combine, organize, and characterize genomic data with full audit, traceability, and reproducibility. Specifically, it includes tools for analyzing raw sequence through to variant calls and for annotating variants with functional, biological, and disease indication information from public and private repositories, Dimitrova said. It also includes bespoke natural language processing tools that let clinicians match patients to potential therapies and clinical trials. Moreover, clinicians can also connect with their peers through the software to share information and get second opinions and feedback on cases before generating the final clinical report for patients.
"Everything that we have on the clinical side is streamlined so that the doctors, pathologists, or medical experts don't have to deal with many discretized pieces of software," Dimitrova said. "All the different steps in running a pipeline [are] done automatically ... and all the clinical data that is pulled from the different information systems in the hospital [is] done automatically."
The research iteration of the software, on the other hand, offers computational workflows for digging deep into large quantities of genomic and phenotypic data. Users have access to supervised and unsupervised machine-learning methods for identifying genes that behave differently under specific conditions such as tumor versus normal or response versus non-response to therapy. There are also tools for analyzing disease pathways and for visualizing genomic data.
The software uses methods that Dimitrova and colleagues have developed and tested in studies done in collaboration with researchers at a number of institutions including Cold Spring Harbor Laboratory, Yale University, and Case Western University. One of these studies, which was published last year, describes a Bayesian method for predicting the functional effects of recurring mutations in cancer.
Those methods will soon be available to the community in an automated way and with full provenance. Like the clinical software, users will also be able to track the samples, datasets, and databases they use for their projects, as well as the analysis steps performed within the software, Dimitrova said. Researchers can also share workspaces with collaborators so that each investigator looks at the same samples, the analyses that have been run, and the results that were generated.
Dimitrova characterized the current iteration of the Philips Genome Informatics platform as an "extension" of a research software tool that Philips started developing in 2009 called the Platform for Personalized Analytics Applications, or PAPAYA. Initially, the tool was designed to analyze microarray data, she said, but as newer genomics technologies came on the market, it evolved to handle data from these platforms as well. In 2013, for instance, Philips' focus was on making PAPAYA available as a tool to help oncologists analyze and interpret next-generation sequencing data and use it to prioritize treatment for cancer patients.
"While we certainly could have deployed earlier versions of the software prior to now, we took what we believe is a significant step in deploying [it] on our flagship Philips HealthSuite Digital platform in order to solve a very pressing problem in making genomic information more usable in a clinical context," Dimitrova said in an email. "This was a significant effort, but I believe we will show very quickly how worthwhile it was."
The broader healthsuite platform will let users host and combine data from a variety of domains including genomics, imaging data, and clinical data from electronic medical records. The platform, which currently sits on Amazon Web Services but could be deployed on other cloud infrastructure, includes identity access management features, interoperability services, virtual health records, hybrid storage, and genomics storing and processing services. There are also administrative tools that let users create and modify their own pipelines. Dimitrova said that Philips is currently testing some of the applications that will be part of the full digital platform in pilot projects. One of those pilots, for example, is testing an application for collecting data from discharged patients, she said.
The platform also makes use of Philips' IntelliBridge enterprise solution that "closes the communication gap" between different information systems in the hospital, according to Dimitrova. For example, if a clinician wants to treat a cancer patient with a BRAF V600E mutation, they need information on whether the patient has melanoma and can be treated with Vemurafenib or if they have some other type of cancer that needs different treatment, she said. IntelliBridge makes it possible to consolidate information on disease diagnostics, cancer histology, and so on from different sources in an automated way instead of clinicians having to search for and collate that information themselves, she said.
When it launches, Dimitrova expects that Philips' genome informatics solution will be useful in multiple clinical and research domains. The firm expects initial customers will come from the oncology space, she said, but the company is seeing interest from cardiovascular and infectious disease researchers. In addition to connecting with Philips-developed clinical information technology systems, the software is also compatible with platforms from third-party healthcare information technology vendors such as electronic medical record systems from Epic and Cerner among others, she said.
It will compete with software solutions from various companies that offer a subset of some of its capabilities. That list includes smaller firms like Tute Genomics and Euformatics, both of which offer tools for variant annotation and interpretation; and larger companies like Qiagen, which offers sequence analysis and variant interpretation solutions. There's also RealTime Oncology, which offers tools for, among other things, matching oncology patients to effective therapies. The benefit of choosing Philips Genome Informatics is that it combines all these capabilities into a single integrated and automated solution, Dimitrova said.
A more direct competitor is SAP's Foundation for Health solution, which launched last year and provides infrastructure based on the SAP Hana proprietary in-memory database computing technology. That solution integrates, manages, and analyzes structured and unstructured datasets including genomic and electronic medical records data in both research and clinical contexts. It is not clear how the two systems differ, but Dimitrova believes that Philips' years of experience in the clinical informatics space and its clinician-centric approach to software development set it apart.
"We start with the clinical expert in mind," she said. "How would a pathologist [or] an oncologist use the system? What are their biggest problems and biggest pain points? That's how we started the design of the system. It's not technology-centric."