|
|
| (41 intermediate revisions by 3 users not shown) |
| Line 1: |
Line 1: |
| − | Cancer is a genomic disease, with enormous heterogeneity in its behavior. In the past, our methods for categorization, prediction of outcome, and treatment selection have relied largely on a morphologic classification of Cancer. But new technologies are fundamentally reframing our views of cancer initiation, progression, metastasis, and response to treatment; moving us towards a molecular classification of Cancer. This transformation depends not only on our ability to deeply investigate the cancer genome, but also on our ability to link these specific molecular changes to specific tumor behaviors. As sequencing costs continue to decline at a supra-Moore’s law rate, a torrent of cancer genomic data is looming. However, our ability to deeply investigate the cancer genome is outpacing our ability to correlate these changes with the phenotypes that they produce. Translational investigators seeking to associate specific genetic, epigenetic, and systems changes with particular tumor behaviors, lack access to detailed observable traits about the cancer (the so called ‘deep phenotype’), which has now become a major barrier to research.
| + | '''Welcome to the Cancer Deep Phenotype Extraction (DeepPhe) project. |
| | + | ''' |
| | | | |
| − | We propose the advanced development and extension of a software platform for performing deep phenotype extraction directly from medical records of patients with Cancer, with the goal of enabling translational cancer research and precision medicine. The work builds on previous informatics research and software development efforts from Boston Children’s Hospital and University of Pittsburgh groups, both individually and together. Multiple software projects developed by our groups (some initially funded by NCI) that have already passed the initial prototyping and pilot development phase (eMERGE, THYME, TIES, ODIE, Apache cTAKES) will be combined and extended to produce an advanced software platform for accelerating cancer research. Previous work in a number of NIH-funded translational science initiatives has already demonstrated the benefits of these methodologies (e.g. Electronic Medical Record and Genomics (eMERGE), PharmacoGenomics Research Network (PGRN), SHARPn, i2b2). However, to date these initiatives have focused exclusively on select non-cancer phenotypes and have had the goal of dichotomizing patients for a particular phenotype of interest (for example, Type II Diabetes, Rheumatoid Arthritis, or Multiple Sclerosis). In contrast, our proposed work focuses on extracting and representing multiple phenotype features for individual patients, to build a cancer phenotype model, relating observable traits over time for individual patients.
| + | Cancer is a genomic disease, with enormous heterogeneity in its behavior. In the past, our methods for categorization, prediction of outcome, and treatment selection have relied largely on a morphologic classification of cancer. But new technologies are fundamentally reframing our views of cancer initiation, progression, [https://en.wikipedia.org/wiki/Metastasis metastasis], and response to treatment; moving us towards a molecular classification of cancer. This transformation depends not only on our ability to deeply investigate the cancer genome, but also on our ability to link these specific molecular changes to specific tumor behaviors. As sequencing costs continue to decline at a supra-[https://en.wikipedia.org/wiki/Moore%27s_law Moore’s law] rate, a torrent of cancer genomic data is looming. However, our ability to deeply investigate the cancer genome is outpacing our ability to correlate these changes with the phenotypes that they produce. [https://en.wikipedia.org/wiki/Translational_science Translational investigators] seeking to associate specific genetic, epigenetic, and systems changes with particular tumor behaviors, lack access to detailed observable traits about the cancer (the so called ‘deep phenotype’), which has now become a major barrier to research. |
| | | | |
| − | ==Goals==
| + | We propose the advanced development and extension of a software platform for performing deep phenotype extraction directly from medical records of patients with cancer, with the goal of enabling translational cancer research and [https://en.wikipedia.org/wiki/Precision_medicine precision medicine]. The work builds on previous informatics research and software development efforts from Boston Children’s Hospital and University of Pittsburgh groups, both individually and together. Multiple software projects developed by our groups (some initially funded by NCI) that have already passed the initial prototyping and pilot development phase ([https://emerge.mc.vanderbilt.edu/about-emerge/ eMERGE], [https://clear.colorado.edu/TemporalWiki/index.php/Main_Page THYME], [http://ties.dbmi.pitt.edu/ TIES], ODIE, [http://ctakes.apache.org/ Apache cTAKES]) will be combined and extended to produce an advanced software platform for accelerating cancer research. Previous work in a number of NIH-funded [https://en.wikipedia.org/wiki/Translational_science translational science] initiatives has already demonstrated the benefits of these methodologies (e.g. [https://emerge.mc.vanderbilt.edu/about-emerge/ Electronic Medical Record and Genomics (eMERGE)], [http://www.pgrn.org/ PharmacoGenomics Research Network (PGRN)], [http://informatics.mayo.edu/sharp/index.php/Main_Page SHARPn], [https://www.i2b2.org/ i2b2]). However, to date these initiatives have focused exclusively on select non-cancer phenotypes and have had the goal of dichotomizing patients for a particular phenotype of interest (for example, [https://en.wikipedia.org/wiki/Diabetes_mellitus_type_2 Type II Diabetes], [https://en.wikipedia.org/wiki/Rheumatoid_arthritis Rheumatoid Arthritis], or Multiple Sclerosis). In contrast, our proposed work focuses on extracting and representing multiple phenotype features for individual patients, to build a cancer phenotype model, relating observable traits over time for individual patients. |
| | | | |
| | + | '''News - DeepPhe v0.2.0 Released |
| | + | ''' |
| | | | |
| − | Our first four development specific aims significantly extend the capability of our current software, focusing on challenging problems in biomedical information extraction. These aims support the development and evaluation of novel methods for cancer deep phenotype extraction:
| + | Visit the [https://github.com/DeepPhe/DeepPhe-Release/blob/master/README.md DeepPhe project] on github to download the software. |
| | | | |
| − | '''Specific Aim 1''': Develop methods for extracting phenotypic profiles. Extract patient’s deep phenotypes, and their attributes such as general modifiers (negation, uncertainty, subject) and cancer specific characteristics (e.g. grade, invasion, lymph node involvement, metastasis, size, stage)
| + | For questions, contact us via the [https://groups.google.com/forum/#!forum/deepphe DeepPhe group] |
| − | | + | |
| − | '''Specific Aim 2''': Extract gene/protein mentions and their variants from the clinical narrative
| + | |
| − | | + | |
| − | '''Specific Aim 3''': Create longitudinal representation of disease process and its resolution. Link phenotypes, treatments and outcomes in temporal associations to create a longitudinal abstraction of the disease
| + | |
| − | | + | |
| − | '''Specific Aim 4''': Extract discourses containing explanations, speculations, and hypotheses, to support explorations of causality
| + | |
| − | | + | |
| − | Our last two implementation specific aims focus on the design of the software to support the cancer research community, ensuring the usability and utility of our software. These aims support the design, dissemination and sharing of the products of this work to maximize impact on cancer research:
| + | |
| − | | + | |
| − | '''Specific Aim 5''': Design and implement a computational platform for deep phenotype discovery and analytics for translational investigators, including integrative visual analytics.
| + | |
| − | | + | |
| − | '''Specific Aim 6''': Advance translational research in driving cancer biology research projects in breast cancer, ovarian cancer, and melanoma. Include research community throughout the design of the platform and its evaluation. Disseminate freely available software.
| + | |
| − | | + | |
| − | '''''Impact:''''' The proposed work will produce novel methods for extracting detailed phenotype information directly from the EMR, the major source of such data for patients with cancer. Extracted phenotypes will be used in three ongoing translational studies with a precision medicine focus. Dissemination of the software will enhance the ability of cancer researchers to abstract meaningful clinical data for translational research. If successful, systematic capture and representation of these phenotypes from EMR data could later be used to drive clinical genomic decision support.
| + | |
| − | | + | |
| − | ==Who We Are==
| + | |
| − | | + | |
| − | '''Boston Childrens Hospital/Harvard Medical School'''
| + | |
| − | | + | |
| − | *Guergana Savova (MPI)
| + | |
| − | | + | |
| − | *Dmitriy Dligach
| + | |
| − | | + | |
| − | *Timothy Miller
| + | |
| − | | + | |
| − | *Sean Finan
| + | |
| − | | + | |
| − | *David Harris
| + | |
| − | | + | |
| − | *Chen Lin
| + | |
| − | | + | |
| − | | + | |
| − | '''University of Pittburgh'''
| + | |
| − | | + | |
| − | *Rebecca Crowley Jacobson (MPI)
| + | |
| − | | + | |
| − | *Harry Hochheiser
| + | |
| − | | + | |
| − | *Roger Day
| + | |
| − | | + | |
| − | *Adrian Lee
| + | |
| − | | + | |
| − | *Robert Edwards
| + | |
| − | | + | |
| − | *John Kirkwood
| + | |
| − | | + | |
| − | *Kevin Mitchell
| + | |
| − | | + | |
| − | *Eugene Tseytlin
| + | |
| − | | + | |
| − | *Girish Chavan
| + | |
| − | | + | |
| − | *Liz Legowski (through Jan 2015)
| + | |
| − | | + | |
| − | *Melissa Castine
| + | |
| − | | + | |
| − | ==Publications and Presentations==
| + | |
| − | | + | |
| − | The following are publications and presentations crediting the Cancer Deep Phenotype Extraction ('''DeepPhe''') project:
| + | |
| − | | + | |
| − | * Hochheiser H; Jacobson R; Washington N; Denny J; Savova G. 2015. Natural language processing for phenotype extraction: challenges and representation. AMIA Annual Symposium. Nov 2015, San Francisco, CA.
| + | |
| − | * Dmitriy Dligach, Timothy Miller, Guergana K. Savova. 2015. Semi-supervised Learning for Phenotyping Tasks. AMIA Annual Symposium. Nov 2015, San Francisco, CA.
| + | |
| − | * Chen, Lin; Dligach, Dmitriy; Miller, Timothy; Bethard, Steven; Savova, Guergana. 2015. Layered temporal modeling for the clinical domain. Journal of the American Medical Informatics Association. http://jamia.oxfordjournals.org/content/early/2015/10/31/jamia.ocv113
| + | |
| − | | + | |
| − | | + | |
| − | ==Funding==
| + | |
| − | | + | |
| − | The project described is supported by Grant Number 1U24CA184407-01 from the National Cancer Institute at the US National Institutes of Health. This work is part of the NCI's Informatics Technology for Cancer Research (ITCR) Initiative (http://itcr.nci.nih.gov/) The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
| + | |
| − | | + | |
| − | The project period is May 2014 - April, 2019.
| + | |
| − | | + | |
| − | ==Stakeholder Information==
| + | |
| − | | + | |
| − | ===User Personae===
| + | |
| − | [[User Personae]]
| + | |
| − | | + | |
| − | ===Contextual Interview Protocol===
| + | |
| − | During the interview, we will ask the informant to conduct relevant tasks, explaining what is being done, why it must be done, and how it is done. As the informant continues through the task, the interviewer(s) will ask for clarification, pose questions about contextual factors, and generally attempt to build an understanding of the work. Therefore, the interview will be unstructured and much of the discussion will be determined by the events that occur as the informant completes their tasks.
| + | |
| − | | + | |
| − | Below is some broad structure for the interview. However, no ordering of these questions is implied - rather, they can be asked when convenient during the course of the discussion.
| + | |
| − | | + | |
| − | ====Recording====
| + | |
| − | All sessions should be audio-recorded with the participant’s permission, either via web conference software or portable digital audio recorder. Written notes will be taken as well. If interviews are conducted via web-conference, screen captures will be recorded as well. We will use GoToMeeting for web conferencing. Artifacts such as data files would also be useful when available.
| + | |
| − | | + | |
| − | Steps will be taken to avoid protected health information (PHI). We will try to remove PHI data entirely from consideration. We will only observe what participants do if it does not involve any PII. If any displays or artifacts involving PHI are inadvertently introduced into the discussion, any screen capture or recording modes that might capture the PHI will be disabled immediately.
| + | |
| − | | + | |
| − | ====Interview Introduction====
| + | |
| − | The introductory component of the discussion is designed to provide basic context and to start conversation:
| + | |
| − | | + | |
| − | Introductory script:
| + | |
| − | | + | |
| − | Hello. My name is XXXX. As you know, you’re here today to participate in a contextual inquiry study to understand the requirements for the visual analytics tools we will be developing. These tools will help researchers explore and interpret phenotype data that will be extracted with our software. We would like to understand users’ information and workflow needs to maximize the usefulness of the software.
| + | |
| − | | + | |
| − | During the session, you will be questioned and observed. We will ask some general questions about your work, and more detailed questions about specific tasks. Data will be collected in the form of notes and artifacts. With your permission, the session will be audio-recorded and captured via screen capture software. The session will last approximately one hour.
| + | |
| − | | + | |
| − | In this session, we would like to understand your broad research and clinical goals, and then explore in detail how you currently identify, store, and manipulate specific phenotype data and link this to genotype data in your work. We’d like to understand how this work is done, any challenges you face, and what you would hope for in a new tool to make work more efficient and easier to accomplish. If possible, we’d like you to demonstrate the tools that you currently use on some meaningful data. We’d like you to identify a representative dataset and task that you are currently working on or have worked on previously to demonstrate how you currently do things.
| + | |
| − | | + | |
| − | Do you have any questions?
| + | |
| − | | + | |
| − | Before we begin, do we have your permission to record this session? (IF YES, BEGIN RECORDING)
| + | |
| − | | + | |
| − | To start off, we’d like to get a sense of your background and research.
| + | |
| − | | + | |
| − | =====Respondent background:=====
| + | |
| − | # What is your training? What degrees have you obtained, and in what areas?
| + | |
| − | # What is your current position? How long have you been at this position?
| + | |
| − | # How much of your time do you spend on cancer data analytics?
| + | |
| − | # If applicable: How much of your time is spent on clinical work?
| + | |
| − | | + | |
| − | =====Problem Description=====
| + | |
| − | # What questions are you currently trying to answer related to cancer analytics?
| + | |
| − | # What sort of data are you dealing with?
| + | |
| − | # Is this a research or clinical effort?
| + | |
| − | # Who is this effort supported by?
| + | |
| − | | + | |
| − | =====Data=====
| + | |
| − | # Which tools do you currently work with for your data?
| + | |
| − | # What are the major data challenges that you struggle with?
| + | |
| − | | + | |
| − | =====Major Non-data Challenges:=====
| + | |
| − | # What are the major challenges that you struggle with that are not data related?
| + | |
| − | | + | |
| − | ===Information modeling interview protocol===
| + | |
| − | | + | |
| − | ===Competency questions===
| + | |
| − | | + | |
| − | ===Contextual Design Information===
| + | |
| − | See presentation here: [[https://healthnlp.hms.harvard.edu/cancer/wiki/images/3/35/Cd-quick-intro-201408110918.pdf]]
| + | |
| − | | + | |
| − | ==Licensing==
| + | |
| − | | + | |
| − | ===Goals===
| + | |
| − | | + | |
| − | * Encourage the broad use of DeepPhe tools and models, while maintaining openness and attribution.
| + | |
| − | * Comply with resource sharing plan as submitted to NIH. See, for example [ITCR policies](http://itcr.nci.nih.gov/about-itcr)
| + | |
| − | | + | |
| − | ===Distinction===
| + | |
| − | | + | |
| − | We are reading "software" as referring to executable components, while "models" refers to the OWL data models developed for representation of cancer phenotypes.
| + | |
| − | | + | |
| − | ===Licensing of Software===
| + | |
| − | | + | |
| − | * cTAKES is licensed under the Apache Source License 2.0
| + | |
| − | * DeepPhe cTAKES components will be similarly licensed
| + | |
| − | * Other DeepPhe components will be licensed by some open source license, potentially (but not necessarily) the Apache license.
| + | |
| − | | + | |
| − | ===Licensing of Other Models===
| + | |
| − | | + | |
| − | * List of non-software licenses -- https://en.wikipedia.org/wiki/List_of_free_content_licenses
| + | |
| − | ** Licensing for content are appropriate
| + | |
| − | * Principles
| + | |
| − | ** receive credit for what we do
| + | |
| − | ** allow other people to contribute (not just take it and use it) -> share alike
| + | |
| − | * Creative Commons
| + | |
| − | ** Creative Commons does not recommend the use of Creative Commons licenses for software.
| + | |
| − | ** https://en.wikipedia.org/wiki/Creative_Commons_license
| + | |
| − | ** CC BY 2.0 - https://creativecommons.org/licenses/by/2.0/
| + | |
| − | *** you are free to share and adapt when you give attribution (You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.)
| + | |
| − | ** CC BY 3.0 - http://creativecommons.org/licenses/by/3.0/
| + | |
| − | ** is it compatible with Apache license?
| + | |
| − | *** CC BY 2.5 and 3.0 are in Category B - http://www.apache.org/legal/resolved.html
| + | |
| − | *** Unmodified media under the Creative Commons Attribution-Share Alike 2.5 and Creative Commons Attribution-Share Alike 3.0 licenses may be included in Apache products, subject to the licenses attribution clauses which may require LICENSE/NOTICE/README changes. For any other type of CC-SA licensed work, please contact the Legal PMC.
| + | |
| − | ** CY BY 4.0 http://creativecommons.org/licenses/by/4.0/
| + | |
| − | *** is it compatible with the Apache license? We believe so.. see [this thread](http://apache.markmail.org/search/?q=legal-discuss#query:legal-discuss%20order%3Adate-backward+page:1+mid:3whmjdfqnvukcg2s+state:results)
| + | |
| − | * Open Content
| + | |
| − | ** https://en.wikipedia.org/wiki/Open_Content_License -- The Open Content License is a share-alike public copyright license which can be applied to a work to make it open content. This license is not compatible with any other license in that it requires derivative works to be licensed under the Open Content License.
| + | |
| − | ** probably not compatible with Apache.
| + | |
| − | * Art Libre
| + | |
| − | ** https://en.wikipedia.org/wiki/Free_Art_License - a copyleft license that grants the right to freely copy, distribute, and transform creative works without the author's explicit permission
| + | |
| − | | + | |
| − | ===DeepPhe Ontology Models===
| + | |
| − | | + | |
| − | * DeepPhe Ontology models will be licensed using CC-BY 4.0
| + | |
| − | * This is consistent with plans to use CC-BY 4.0 to license the schema ontology upon which the DeepPhe model is based.
| + | |
Cancer is a genomic disease, with enormous heterogeneity in its behavior. In the past, our methods for categorization, prediction of outcome, and treatment selection have relied largely on a morphologic classification of cancer. But new technologies are fundamentally reframing our views of cancer initiation, progression, metastasis, and response to treatment; moving us towards a molecular classification of cancer. This transformation depends not only on our ability to deeply investigate the cancer genome, but also on our ability to link these specific molecular changes to specific tumor behaviors. As sequencing costs continue to decline at a supra-Moore’s law rate, a torrent of cancer genomic data is looming. However, our ability to deeply investigate the cancer genome is outpacing our ability to correlate these changes with the phenotypes that they produce. Translational investigators seeking to associate specific genetic, epigenetic, and systems changes with particular tumor behaviors, lack access to detailed observable traits about the cancer (the so called ‘deep phenotype’), which has now become a major barrier to research.
We propose the advanced development and extension of a software platform for performing deep phenotype extraction directly from medical records of patients with cancer, with the goal of enabling translational cancer research and precision medicine. The work builds on previous informatics research and software development efforts from Boston Children’s Hospital and University of Pittsburgh groups, both individually and together. Multiple software projects developed by our groups (some initially funded by NCI) that have already passed the initial prototyping and pilot development phase (eMERGE, THYME, TIES, ODIE, Apache cTAKES) will be combined and extended to produce an advanced software platform for accelerating cancer research. Previous work in a number of NIH-funded translational science initiatives has already demonstrated the benefits of these methodologies (e.g. Electronic Medical Record and Genomics (eMERGE), PharmacoGenomics Research Network (PGRN), SHARPn, i2b2). However, to date these initiatives have focused exclusively on select non-cancer phenotypes and have had the goal of dichotomizing patients for a particular phenotype of interest (for example, Type II Diabetes, Rheumatoid Arthritis, or Multiple Sclerosis). In contrast, our proposed work focuses on extracting and representing multiple phenotype features for individual patients, to build a cancer phenotype model, relating observable traits over time for individual patients.
