2023 News

We are pleased to announce the launch of the SNOMED CT Entity Linking Challenge in collaboration with SNOMED International, Veratai, and DrivenData. Sponsored by SNOMED International, the challenge seeks to advance the development of Entity Linking models that operate on unstructured clinical text. 

Participants are tasked with developing a model to annotate MIMIC-IV-Note discharge summaries with SNOMED CT concepts. Training data, comprising ~300 annotated discharge summaries, has been made available on PhysioNet at https://doi.org/10.13026/s48e-sp45.

Sign up now at https://snomed.drivendata.org!

INSIDE-HEART brings together universities, companies and hospitals from Italy, Finland, France, Israel, Netherlands, Spain, and Sweden to establish a multi-disciplinary network to tackle the design and early-phase validation of digital biomarkers targeting the diagnosis of supraventricular arrhythmias (SVAs) and their associated potential for adverse risk assessment.

Our colleagues in the network are looking for 10 motivated PhD candidates, funded by the European Union’s Horizon Europe program under the Marie Skłodowska Curie Actions. For further information and details on how to apply, see: https://www.inside-heart.eu/recruitment/. The call for applications is open until 31 January 2024.

The INSIDE-HEART project is coordinated by the Politecnico di Milano. Please direct questions to insideheart@polimi.it

Qualification for the DARPA Triage Challenge has been extended through November 27, 2023 at 23:59. We encourage you to join the challenge as a self-funded team for the Systems, Virtual and Data Competitions. You may compete in one or more challenge tracks, where qualification must be entered for each track individually. 

To register, please visit the Team Qualification Portal at: https://events.sa-meetings.com/DTCTeamPortal/.

For more information on the challenge, see: https://triagechallenge.darpa.mil/ or contact TriageChallenge@darpa.mil.

The Defense Advanced Research Projects Agency (DARPA), a research and development agency within the Department of Defense, is seeking competitors for a new medical response challenge. The DARPA Triage Challenge aims to drive breakthrough innovations that improve medical response time during mass casualty incidents in complex military and civilian settings, especially when medical resources are limited relative to the need.

The challenge includes a series of technical challenge events to drive breakthrough innovations in the identification of physiological features (signatures) of injury, and help medical responders perform scalable, timely, and accurate triage. The challenge has two primary triage competitions – Systems and Virtual – and a secondary triage Data competition. The Systems and Virtual competitions focus on stand-off sensing of physiological data using autonomous platforms – uncrewed aerial and ground vehicles – during primary triage. Competitors will conduct real-time sensor data analysis to identify casualties for urgent hands-on evaluation by medical personnel.

Relevant to the PhysioNet community, the Data competition seeks to identify physiological signatures of injury derived from data captured by non-invasive sensors (contact-based or stand-off). Such advances could accelerate responders’ anticipatory decisions and prioritization for medical care during secondary triage. Competitors will attempt to develop algorithms that detect signatures in these data streams to provide decision support appropriate for austere and complex pre-hospital settings. Of particular interest are early signatures indicating a need for life-saving interventions against conditions that medics are trained and equipped to treat during secondary triage, such as hemorrhage and airway injuries.

The Data competition will use DARPA-provided de-identified, multi-modal physiological data from trauma patients across diverse settings and cohorts provided by the DARPA Research Infrastructure for Trauma with Medical Observations effort. Data types include, but are not limited, to: photoplethysmography (PPG) waveforms, medical procedures, imaging results and video footage during prehospital helicopter transport and in the trauma bay.

Prizes for year one:

• Systems Competition: Up to $200K Prize pool
• Virtual Competition: Up to $100K Prize pool
• Data Competition: Up to $200K Prize pool

Total Prizes $7M over three challenges

DARPA is currently seeking self-funded competitors. Join us by registering on the Qualification Portal, now through Nov. 13, 2023.

For more information visit the DARPA Triage Challenge website.

The Google Research team is looking for partners to understand the needs and requirements for synthetic data. They have capabilities to generate both structured and unstructured patient data as well as images for infrastructure testing and medical research. Please contact us if you are interested in the partnership.

The MIMIC-IV-ECG module is now available. This module contains approximately 800,000 diagnostic electrocardiograms across nearly 160,000 unique patients. The vast majority of ECGs for patients who appear in the MIMIC-IV Clinical Database are included. The patients in MIMIC-IV-ECG have been matched against the MIMIC-IV Clinical Database, making it possible to link to information across the MIMIC-IV modules. When a cardiologist report is available for a given ECG, we provide information for linking to it.

The authors of the FFA-IR dataset have asked for downloads to be disabled until further notice to adhere with local policy changes. We apologize for the inconvenience and hope to make the files available again in the future.

I-CARE v.2.0 is now available on Google Cloud. For details on downloading the dataset or working with it directly in the cloud, see the Files section of the project description.

We are pleased to announce CXR-LT, a competition on Multi-Label Long-Tailed Classification on Chest X-Rays. Many real-world problems, including diagnostic medical imaging exams, are “long-tailed”: there are a few common findings followed by more relatively rare conditions. This competition will provide a challenging large-scale multi-label long-tailed learning task on chest X-rays (CXRs), encouraging community engagement with this emerging interdisciplinary topic.

CXR-LT is organized as a shared task for the workshop on Computer Vision for Automated Medical Diagnosis (CVAMD) held in association with the International Conference on Computer Vision (ICCV) 2023. Participants will be invited to submit their solutions for publication presentation at CVAMD 2023 and publication in the ICCV 2023 workshop proceedings.

The challenge uses an expanded version of MIMIC-CXR-JPG v2.0.0, a large benchmark dataset for automated thorax disease classification. Each CXR study in the dataset was labeled with 12 newly added disease findings extracted from the associated radiology reports. The resulting long-tailed (LT) dataset contains 377,110 CXRs, each labeled with at least one of 26 clinical findings (including a "No Finding" class).

Important dates

05/01/2023: Development Phase begins. Participants can begin making submissions and tracking results on the public leaderboard.
07/14/2023: Testing Phase begins. Unlabeled test data will be released to registered participants. The leaderboard will be kept private for this phase.
07/17/2023: Competition ends. Participants are invited to submit their solutions as 8-page papers to ICCV CVAMD 2023!
07/28/2023: ICCV CVAMD 2023 submission deadline. (Competition participants may receive an extension if needed.)
08/11/2023: ICCV CVAMD 2023 acceptance notification.
10/06/2023: ICCV CVAMD 2023 workshop.

This competition is supported in part by the Artificial Intelligence Journal (AIJ). For any questions, please contact cxr.lt.competition.2023@gmail.com.

We are aware that downloading the I-CARE dataset from PhysioNet is currently slow and we apologize for the inconvenience. To resolve this issue, we are currently transferring the dataset to Google Cloud. Once the transfer is complete, the dataset can be analyzed directly in the cloud or downloaded using Google Cloud Utilities. Please check here for updates.

We are excited to announce the release of Medical AI Research Foundations — a repository of open-source medical foundation models and a collaboration between Google Health and PhysioNet. Our goal in releasing this collection of resources is to accelerate medical AI research and to democratize access to foundational medical AI models.

We are seeding Medical AI Research Foundations with REMEDIS models for chest X-ray and pathology (with related Github code). We expect to add more models and resources for training medical foundation models such as datasets and benchmarks in the future. We also welcome contributions from the medical AI research community.

We have received inquiries regarding the use of credentialed data (MIMIC-III, MIMIC-IV, MIMIC-CXR) with online services such as GPT. The PhysioNet Credentialed Data Use Agreement explicitly prohibits sharing access to the data with third parties, including sending it through APIs provided by companies like OpenAI, or using it in online platforms like ChatGPT.

If you are interested in using the GPT family of models, we suggest using one of the following services:

• Azure OpenAI service. You'll need to opt out of human review of the data, as (1) you are processing sensitive data where the likelihood of harmful outputs and/or misuse is low, and (2) you do not have the right to permit Microsoft to process the data for abuse detection due to the data use agreement you have signed. The form for opting out of the review process is available here: https://aka.ms/oai/additionalusecase
• Amazon Bedrock. Bedrock provides options for fine-tuning foundation models using private labeled data. After creating a copy of a base foundation model for exclusive use, data is not shared back to the base model for training.

If you have any questions about this policy, feel free to reach out: https://physionet.org/about/#contact_us

The EnPRO Lab in the Department of Engineering and School of Biomedical Engineering & Imaging Sciences at King's College London is looking to fill a music-physiology data science doctoral scholarship which is now open to international applicants for October 2023 entry. The successful applicant will join the research team of the ERC COSMOS project (cosmos.isd.kcl.ac.uk).

The research investigates the impact of music expressivity on the autonomic nervous system. The project focuses on developing individualized, explanatory computational models for modulating autonomic responses through music that can be used in digital therapeutics for cardiovascular health. The scientific approach will be based on studying the interactions between musical prosody (acoustic variations introduced in musical communication) and autonomic parameters such as heart rate, heart rate variability, respiration, and blood pressure. The methods build on software tools developed in the COSMOS project.

The research activities will include study design, ethics application, data collection, data processing, computational modelling, and analysis and interpretation of results, and disseminating results through publications and conference presentations. The ideal candidate will be knowledgeable in Python, Matlab, or R, and have a Bachelors or Masters degree in biomedical engineering, mathematical and computational sciences, music information research, or a related discipline. Experience in analysis of biosignals and/or music signals, or industrial experience is desirable. 

Funding is available for 3.5 years, covers fees and stipend, and standard computing/travel support. For further details, please see the job posting.

An estimated 7 to 10 million people around the world have Parkinson’s disease, many of whom suffer from freezing of gait (FOG). FOG are unpredictable, unexpected, involuntary episodic events. During a FOG episode, patients report that their feet are inexplicably “glued” to the ground, preventing them from moving forward despite their attempts.

PhysioNet contributor Jeff Hausdorff and his colleagues at the Tel Aviv Sourasky Medical Center, KU Leuven, and Harvard Medical School, have contributed a large dataset to a machine learning contest that was recently launched to automatically detect FOG episodes and to address the shortcomings of existing methods.

This work has the potential to help advance the evaluation, understanding, and treatment of FOG, and, ultimately, to improve the lives of the many people who suffer from this debilitating Parkinson’s disease symptom. To join the competition, visit Kaggle.

Timeline

• June 1, 2023: Entry Deadline. You must accept the competition rules before this date in order to compete.
• June 1, 2023: Team Merger Deadline. This is the last day participants may join or merge teams.
• June 8, 2023: Final Submission Deadline.

Prizes

• 1st Place: $40,000
• 2nd Place: $25,000
• 3rd Place: $20,000
• 4th Place: $10,000
• 5th Place: $5,000

Over 45 students, academics, clinicians, and engineers gathered at the Google Canada offices on 23-24 February for the Toronto Health Datathon 2023. Participants used anonymized real-world data from Health Data Nexus to develop machine learning models aimed at solving real-world problems facing Canadian healthcare.

Over the past two years, PhysioNet has been collaborating with Temerty Centre for Artificial Intelligence Research and Education in Medicine (T-CAIREM) at the University of Toronto to develop the software that underpins both Health Data Nexus and PhysioNet. We look forward to continuing this collaboration, working towards tight integration between the two platforms.

We are excited to be supporting the 3rd Tokyo Datathon on Machine Learning in Healthcare, co-organized by Tokyo Medical and Dental University and MIT Critical Data. The event will be held on 1-3 September 2023 and will bring together experts from across healthcare and data science to tackle clinical questions.

Registration and event details will be posted on the Datathon Website. If you are a Japanese-language speaker with experience with the MIMIC dataset and would be interested in helping as a mentor at the event, please reach out to Leo Anthony Celi.

We are delighted to announce the opening of the George B. Moody PhysioNet Challenge 2023. This year’s Challenge invites teams to use electroencephalogram (EEG) recordings to predict the neurological recovery of patients from coma in the hours following resuscitation from cardiac arrest. This Challenge leverages a novel database of over 1,000 subjects from seven hospitals who together underwent over 50,000 hours of EEG monitoring. As always, the team with the best score for this task on the hidden test set wins the Challenge.

We have shared data, example code, and scoring code in both MATLAB and Python, and we will open the scoring system in the coming weeks. As in previous years, we have divided the Challenge into two phases: an unofficial phase and an official phase. The unofficial phase solicits feedback from the research community (i.e., you) to help us to improve the Challenge for the official phase, so we require teams to register and participate in the unofficial phase of the Challenge to be eligible for a prize. Please enter early and often – we need you to look for and share the quirks in our data, our scoring system, and otherwise. 

Please see the Challenge website and the Challenge forum for more information, rules and deadlines: https://physionetchallenges.org/2023/
https://groups.google.com/g/physionet-challenges/

Many thanks again for your continued support of this event, and we hope that you enjoy the 2023 Challenge.

We are delighted to announce that the George B. Moody PhysioNet Challenges were awarded the "Distinguished Achievement Award for Data Reuse, as part of the DataWorks! Prize, while MIT Critical Data was awarded "Significant Achievement Award for Data Sharing". 

Launched on May 11, 2022, the Data Works! Prize was created in partnership between the NIH Office of Data Science Strategy and the Federation of American Societies for Experimental Biology (FASEB) to highlight the critical role of data sharing and reuse in scientific discovery.

George B. Moody designed and led the Challenges from 2000 to 2015. Prof. Clifford has led the Challenges since 2015 and has been a key contributor to its parent resource, PhysioNet (The Research Resource for Complex Physiologic Signals), for over two decades. Prof. Reyna has co-led the PhysioNet Challenges since 2019, and has been instrumental in the development of its repeatable science standards. 

MIT Critical Data, led by the Laboratory for Computational Physiology, builds communities around the world across disciplines to derive knowledge from data routinely collected in the process of care in order to understand health and disease better, and in the local context. Its flagship project is the Medical Information Mart for Intensive Care, or the MIMIC database.

More on the DataWorks! Prize here: https://datascience.nih.gov/director/directors-blog-dataworks-winners-2023 and the PhysioNet Challenges here: https://physionetchallenges.org/faq and MIT Critical Data here: https://criticaldata.mit.edu/.

PhysioNet and MIMIC are supported by the National Institute of Biomedical Imaging and Bioengineering.

Our colleagues at UCSF Department of Anesthesia, The UCSF Hypoxia Lab (hypoxialab.org) and the UCSF Center for Health Equity in Surgery and Anesthesia (chesa.ucsf.edu) are seeking a full-time Data Analyst (with data engineering skills) to join their Open Oximetry project.

This project seeks to understand the potential impact of skin color on accuracy of pulse oximetry and other medical diagnostic devices. The Data Analyst will work at the intersection of health diagnostics, health equity and AI in the world’s leading lab (hypoxialab.org) for this type of research along with a team of experts who have published some of the seminal research on this topic.

As part of the project, the team will be setting up robust data collection systems in the lab and the hospital settings as well as creating and managing an open access data repository for diagnostics device data. This repository will contain data from our lab as well as data from collaborating study groups. The data will be shared via portals that facilitate raw data utilization for researchers and industry, as well as visualized data to help lay persons and consumers better understand device technology performance and standards.

The Data Analyst will work closely with the laboratory-based clinical research team to gather, analyze, and interpret a wide variety of research data; Design and conduct research including selecting data, developing research instruments, analyzing collected information according to established statistical methods, and developing recommendations based on research findings; Prepare reports, charts, tables, and other visual aids to interpret and communicate data and results; Create and manage data repositories; work with our AI/ML team to test novel analytic methods with our data. The ideal candidate will be knowledgeable in SQL and Python or R and have a Bachelors or Masters degree in Statistics, Data Science, or adjacent technical field.  

This position is for a period of 12 months, though may be longer contingent on further funding. For further details, see the job posting.

We are excited to announce the launch of a shared task on problem list summarization at the BioNLP Workshop 2023. The goal for participants is to generate a list of diagnoses and problems in a patient’s daily care plan using input from the provider’s progress notes during hospitalization. The task contains 768 progress notes for training, and 300 progress notes for evaluation. The goal of this shared task is to attract future research efforts in building NLP models for real-world decision support applications, where a system generating relevant and accurate diagnoses will assist the healthcare providers’ decision-making process and improve the quality of care for patients.

Participants will be tasked with developing NLP systems for EHR summarization. Participants who design novel systems and achieve competitive performance in the shared task, running from January to April 2023, will be invited to present their results at the BioNLP Workshop, which will be held in Toronto, Canada and co-located with ACL 2023. The challenge is open to anyone interested in clinical NLP and medical AI. We encourage individuals, teams, and organizations to participate.

To register for the challenge, please visit: https://forms.gle/geTXN6Z1pyfC55Fn8. More information about the challenge, including the official rules and guidelines, can be found at: https://physionet.org/content/bionlp-workshop-2023-task-1a/. You are welcome to join our google discussion group for newest update: https://groups.google.com/g/bionlp2023problemsumm

The SOAP Note Tagging and Problem List Summarization dataset dataset is temporarily unavailable as it is part of an ongoing shared task of BioNLP Workshop 2023: 1A (Problem List Summarization). The dataset will be made available on July 13th, 2023. More details about the workshop and shared task can be found at: https://doi.org/10.13026/s8wk-ja78

We apologize for any inconvenience this may cause and appreciate your understanding. We will provide updates as soon as more information becomes available. A new test set with 300 progress notes will be released along with the original set of 768 notes when the embargo is lifted. If you are interested in signing up the shared task, register here: https://forms.gle/geTXN6Z1pyfC55Fn8