2022 News

A beta release of the MIMIC-IV-ECG module is now available to MIT Critical Data Consortium members. The MIMIC-IV-ECG module contains approximately 800,000 diagnostic electrocardiograms across nearly 160,000 unique patients. All of the ECGs for patients who appear in the MIMIC-IV Clinical Database are included. When a cardiologist report is available for a given ECG, it is also provided. 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.

A public version of this dataset will be released in approximately six months. During the embargo period we will be carrying out additional tests and data quality checks.

Congratulations to our colleagues who have been selected as finalists for the National Institutes of Health DataWorks Challenge!

Please support one of these teams working to share and reuse data in research and scientific discovery (The link will take you directly to the page to submit a vote). Voting is open until December 21, 2022.  Unfortunately, you may only vote for one team, but they can both be awarded prizes.  Please share and promote awareness to increase our colleagues’ chances!

MIT Critical Data
https://www.herox.com/dataworks/round/2457/entry/41104

MIT Critical Data builds communities across disciplines to derive knowledge from health records to understand health and disease better. Help them continue to build valuable research resources such as MIMIC  and freely accessible educational resources.

PhysioNet Challenges
https://www.herox.com/dataworks/round/2457/entry/41376

The PhysioNet Challenges are annual data science competitions that ask what we can learn from data to improve health and healthcare. Help the team draw out unrealized value from data and advance data reuse and algorithm development.

Background

The Federation of American Societies for Experimental Biology (FASEB) and the National Institutes of Health (NIH) are championing a bold vision of data sharing and reuse. The DataWorks! Prize fuels this vision with an annual challenge that showcases the benefits of research data management while recognizing and rewarding teams whose research demonstrates the power of data sharing or reuse practices to advance scientific discovery and human health.  The future of biological and biomedical research hinges on researchers’ ability to share and reuse data. Sharing and reuse had a sizable, catalytic impact on the development of COVID-19 vaccines and treatment protocols. The DataWorks! Prize is an opportunity for the research community to share their stories about the practices, big and small, that lead to scientific discovery.

The PhysioNet team were recipients of the inaugural MIT Prize for Open Data in recognition of their work to support health research and education. The award - established to highlight the value of open data at MIT - was presented by School of Science Dean Nergis Mavalvala and MIT Libraries Director Chris Bourg on October 28 in the MIT Hayden Library.

We are pleased to announce the publication of a dataset comprising multimodal physiologic, flight performance, and user interaction data streams, collected as participants performed virtual-reality flight tasks of varying difficulty. 

With over an hour of highly multimodal physiological and behavioral signals collected on each of the thirty-five participants, the dataset represents a unique opportunity to develop analytics and models linking an individual’s physiology to their behavior and performance in tasks of varying difficulty.

More data are being collected and will be uploaded to PhysioNet periodically. The data underpins the CogPilot Data Challenge, which explores how performance measurements and physiological data can be used to assess the competency of student pilots. To participate in the CogPilot Challenge, visit: https://pilotperformance.mit.edu/

We are pleased to announce an initial release of a version (0.1.0) of the MIMIC-IV-Waveform module. These waveforms are a rich source of patient information - including ECG, PPG, and Blood Pressure signals - and can be linked to the clinical information in MIMIC-IV. This initial release contains 200 records from 198 patients. An upcoming release will include around 10,000 records. 

The dataset was the subject of a workshop at IEEE EMBC in July of 2022, led by Peter Charlton, which demonstrated how to use the WFDB-Python package to extract and analyze waveform features. Executable notebooks and tutorial materials are available at: https://mimic.mit.edu/docs/iv/tutorials/waveform/ieee_workshop/ .

Gari Clifford and Matthew Reyna from Emory University and Elaine Nsoesie from Boston University recently published an invited viewpoint in The Journal of the American Medical Association on "Rethinking Algorithm Performance Metrics for Artificial Intelligence in Diagnostic Medicine". The viewpoint focuses on how we often use the wrong optimization targets when applying machine learning to medical data, and how we can address this issue, and is part of the Gordon and Betty Moore Foundation's broader series on Diagnostic Excellence. 

Our servers at MIT are undergoing and testing and maintenance work between Friday 15th and Monday 18th July 2022. PhysioNet (https://physionet.org/) will remain active during this period, but some services may be affected (for example, the archive website https://archive.physionet.org/).

Linking an ORCID to your PhysioNet profile can help us to quickly verify your identity, speeding up the process of gaining access to datasets such as MIMIC. 

When reviewing an ORCID profile, we look out for “Trust Markers” which are pieces of information added to the profile by groups such publishers and employers.

To find out more about how PhysioNet is working with ORCID, see their blog post on how we are using the Trust Markers to streamline the data credentialing process.

We are currently dealing with a high volume of applications for credentialed access to PhysioNet, so please expect significant delays (up to 45 days) in the review process. We are doing our best to deal with the waitlist quickly, handling applications in the order in which they are received. To help ensure that your application is successful, please remember to:

• Include a copy of your CITI training report (not the certificate).
• Remind your reference to reply promptly when contacted.
• Check your application details are correct before submitting.
• Add an institutional or educational email address as your primary email.

We apologize for the inconvenience. Please bear with us during this busy time!

Between Wednesday 30 March and Thursday 31 March the PhysioNet mail server suffered an interruption. As a result, if you attempted to register for a new account or reset a password, you may not have received an email.

We are now in the process of resending emails, so please check your mailbox if an expected email did not arrive. We apologize for the inconvenience and we will be taking steps to prevent similar issues issue in the future.

Applying for access to protected datasets such as MIMIC is now a three step process:

• Submit your personal details via the "Credentialing" page of your user profile.
• Submit your training report via the "Training" page of your user profile.
• Once the steps above are complete, sign the Data Use Agreement in the "Files" section of the relevant project. You can view your signed agreements on the "Agreements" page of your user profile.

The change helps us to review data access requests more efficiently and gives us the flexibility to support additional training courses in future.

When applying for access to datasets such as MIMIC, you are currently required to submit a single "credentialing" application that allows us to check both CITI training status and identity. 

We will soon be releasing updates to PhysioNet that will require training and identity details to be submitted independently. As a result, a new "Training" tab will be added to your user profile. 

This change will allow us to support different types of training for different datasets (for example, an Australian dataset may require training on specific issues relating to Australia). The change will also help us to streamline the review process.

Dear Community,

We have two big announcements today. First, at last year’s Computing in Cardiology (CinC) Conference, the Board voted to rename the PhysioNet/Computing in Cardiology Challenges in honor of George Moody, and his life-long contributions to the field, and specifically PhysioNet, the Challenges and CinC. The Challenge is now called the “The George B. Moody PhysioNet Challenge”. For consistency, we will still abbreviate this to the PhysioNet Challenge. We continue to partner with CinC, and will be awarding prizes in Finland in September this year.

Second, we are delighted to also announce the opening of the George B. Moody PhysioNet Challenge 2022. This year’s Challenge aims to identify the presence, absence, or unclear cases of murmur waves in heart sound recordings collected from multiple auscultation locations on the body using a digital stethoscope. Building on our successful Challenge from 2016, together with our generous collaborators at the Universidade Portucalense and Universidade do Porto, we have sourced a database of 5272 recordings from 1568 inhabitants of Pernambuco state, Brazil during two independent cardiac screening campaigns which were designed to support the development of a telemedicine network. More details on the data can be found in a recent publication in the IEEE Journal of Biomedical and Health Informatics.

Demo code in Python and MATLAB is available, and the scoring system will be open later in February - Look out for the announcement very soon!

See here for more information, rules and deadlines: https://moody-challenge.physionet.org/2022/

As with previous years, we have divided the Challenge into two phases. The first (unofficial phase) is to enable the research community (i.e. you) to help us improve the Challenge. We value your input, and it is therefore compulsory to enter the Challenge in this unofficial phase (before 9th April) to be eligible for a prize. This is very important - we need your help! Please enter early and often - the score doesn’t matter and will be wiped from the board for the official phase. We need you to look for quirks in our scoring system, data formats, or anomalies in the raw data that do not make sense to you. We are not perfect, and are bandwidth-limited, and so we rely on the peer-review of our community to improve the quality of the Challenge every year. In particular, this year we strongly encourage suggestions for modifying our scoring function, which we are sure will be controversial.

Please note that you *must* re-register for this Challenge (using the link below) even if you competed last year and have no team changes, and to be eligible for a prize, you must submit at least one successful entry during the unofficial phase.

Quick links for this year's Challenge:

• Registration form
• Public discussion forum
• Rules and deadlines
• Data and Data Description

More information will be posted on the PhysioNet Challenge website as it is available. Please check the Challenge forum for real-time updates. Please also post questions and comments in the forum. However, if your question reveals information about your entry, then please email Info [at] physionetchallenge.org. We may post parts of our reply publicly if we feel that all Challengers should benefit from the information contained in our responses. We will not answer emails about the Challenge sent to other email addresses.

Many thanks again for your continued support of this event, and we hope that you enjoy the 2022 Challenge

-Gari, Matt, Ali and the PhysioNet Challenge Team

January 26, 2022: Both the 2020 Challenge and the 2021 Challenge, which extended the 2020 Challenge, are now complete. The CinC articles for both Challenges are available on the CinC website here and here. The final scores can be found here. Please cite Perez Alday EA, Gu A, J Shah A, Robichaux C, Ian Wong AK, Liu C, Liu F, Bahrami Rad A, Elola A, Seyedi S, Li Q, Sharma A, Clifford GD^* Reyna MA^*. Classification of 12-lead ECGs: The PhysioNet/Computing in Cardiology Challenge 2020. Physiol. Meas. 2021 Jan 1;41(12):124003. doi: 10.1088/1361-6579/abc960 to refer to the 2020 Challenge. Please also cite Reyna MA, Sadr N, Perez Alday EA, Gu A, Shah AJ, Robichaux C, Rad AB, Elola A, Seyedi S, Ansari S, Ghanbari H, Li Q, Sharma A, Clifford GD. Will Two Do? Varying Dimensions in Electrocardiography: The PhysioNet/Computing in Cardiology Challenge 2021. Computing in Cardiology 2021; 48: 1-4 and Reyna MA, Sadr N, Perez Alday EA, Gu Annie, Shah AJ, Robichaux C, Rad AB, Elola A, Seyedi S, Ansari S, Ghanbari H, Li Q, Sharma A, Clifford GD. Issues in the automated classification of multilead ECGs using heterogeneous labels and populations. Preprint. 2022 to refer to the 2021 Challenge. Finally, please also cite the standard PhysioNet citation. You can find followup articles to the 2020 Challenge and the 2021 Challenge in the Journal of Physiological Measurement Focus Issue on Classification of Multilead ECGs.

We are pleased to announce publication of BRAX, a Brazilian chest x-ray dataset labelled with 14 radiological findings derived from Portuguese medical reports using NLP. BRAX contains 24,959 chest radiography exams and 40,967 images acquired in a large general Brazilian hospital.

Dr Eduardo Pontes Reis, co-author and Radiologist at Hospital Israelita Albert Einstein, notes the importance of datasets from underrepresented regions for evaluating how well clinical applications of deep learning can generalize to new populations and for reducing the lack of geographic diversity in publicly available chest x-ray data.