text stringlengths 0 41M |
|---|
The PhysioNet Credentialed Health Data License |
Version 1.5.0 |
Copyright (c) 2024 MIT Laboratory for Computational Physiology |
The MIT Laboratory for Computational Physiology (MIT-LCP) wishes |
to make data available for research and educational purposes to |
qualified requestors, but only if the data are used and protected |
in accordance with the terms and conditions stated in this License. |
It is hereby agreed between the data requestor, hereinafter referred to |
as the "LICENSEE", and MIT-LCP, that: |
1. The LICENSEE will not attempt to identify any individual or |
institution referenced in PhysioNet restricted data. |
2. The LICENSEE will exercise all reasonable and prudent care to avoid |
disclosure of the identity of any individual or institution |
referenced in PhysioNet restricted data in any publication or other |
communication. |
3. The LICENSEE will not share access to PhysioNet restricted data |
with anyone else. |
4. The LICENSEE will exercise all reasonable and prudent care to |
maintain the physical and electronic security of PhysioNet restricted |
data. |
5. If the LICENSEE finds information within PhysioNet restricted data |
that he or she believes might permit identification of any individual |
or institution, the LICENSEE will report the location of this |
information promptly by email to PHI-report@physionet.org, citing the |
location of the specific information in question. |
6. The LICENSEE will use the data for the sole purpose of lawful use in |
scientific research and no other. |
7. The LICENSEE will be responsible for ensuring that he or she |
maintains up to date certification in human research subject |
protection and HIPAA regulations. |
8. The LICENSEE agrees to contribute code associated with publications |
arising from this data to a repository that is open to the research community. |
9. This agreement may be terminated by either party at any time, but the |
LICENSEE's obligations with respect to PhysioNet data shall continue |
after termination. |
THE DATA ARE PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL |
THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR |
OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, |
ARISING FROM, OUT OF OR IN CONNECTION WITH THE DATA OR THE USE OR OTHER |
DEALINGS IN THE DATA. |
# EHR-DS-QA: A synthetic QA Dataset Derived from Medical Discharge Summaries for Enhanced Medical Information Retrieval Systems |
This repository contains a dataset designed for training large language models specialized in retrieving information from discharge summaries taken from electronic health records through context-augmented generation. The data is extracted and structured from the MIMIC-IV-Note database and is available in two formats: J... |
## Dataset Files |
1. **mimic_note_qa.json** |
2. **mimic_note_qa.csv** |
3. **mimic_note_qa_verified.json** |
4. **mimic_note_qa_verified.csv** |
## File Structure and Descriptions |
### JSON Structure |
The JSON file is structured as a list of dictionaries where each dictionary represents a medical discharge note. Here's the structure of each entry along with an example entry with placeholder data: |
```json |
{ |
"original_id": "placeholder_original_id", |
"qa_pairs": [ |
{ |
"question": "placeholder_question_text", |
"answer": "placeholder_answer_text", |
"correct": "placeholder_boolean" |
} |
], |
"extended_context": "placeholder_boolean", |
"note_id": "placeholder_note_id", |
"human_verified": "placeholder_boolean" |
} |
``` |
#### Fields: |
- **original_id**: (Integer) A unique identifier assigned during the creation of this dataset. |
- **qa_pairs**: (List of Dictionaries) Each dictionary contains: |
- **question**: (String) A question related to the patient's case. |
- **answer**: (String) An answer to the corresponding question. |
- **correct**: (Boolean, optional) Indicates the verified accuracy of the Q&A pair. |
- **extended_context**: (Boolean) Indicates whether the note contains more than 3000 tokens. |
- **note_id**: (String) A unique identifier for each note. |
- **human_verified**: (Boolean, optional) Denotes if all QA pairs associated with the note have been verified by humans. |
### CSV Structure |
The CSV file is generated from the JSON data, where each key from the JSON dictionary becomes a column in the CSV file. Nested structures such as `qa_pairs` (a list of dictionaries in the JSON) will be converted to a string representation and stored in a single cell in the CSV file. This is generated using a Python scr... |
Here's an example of how the data would be structured in the CSV: |
```csv |
End of preview. Expand in Data Studio
README.md exists but content is empty.
- Downloads last month
- 20