search

justin13601 / ACES

ACES: Automatic Cohort Extraction System for Event-Streams
https://eventstreamaces.readthedocs.io/en/latest/
MIT License
21 stars 1 forks source link
cohort-selection electronic-health-records machine-learning reproducible-research

readme

ACES

Python PyPI Hydra Codecov Tests Code Quality Documentation Contributors Pull Requests License

Automatic Cohort Extraction System for Event-Streams

Updates

Automatic Cohort Extraction System (ACES) is a library that streamlines the extraction of task-specific cohorts from time series datasets formatted as event-streams, such as Electronic Health Records (EHR). ACES is designed to query these EHR datasets for valid subjects, guided by various constraints and requirements defined in a YAML task configuration file. This offers a powerful and user-friendly solution to researchers and developers. The use of a human-readable YAML configuration file also eliminates the need for users to be proficient in complex dataframe querying, making the extraction process accessible to a broader audience.

There are diverse applications in healthcare and beyond. For instance, researchers can effortlessly define subsets of EHR datasets for training of foundation models. Retrospective analyses can also become more accessible to clinicians as it enables the extraction of tailored cohorts for studying specific medical conditions or population demographics. Finally, ACES can help realize a new era of benchmarking over tasks instead of data - please check out MEDS-DEV!

Currently, two data standards are directly supported: the Medical Event Data Standard (MEDS) standard and the EventStreamGPT (ESGPT) standard. You must format your data in one of these two formats by following instructions in their respective repositories. ACES also supports any arbitrary dataset schema, provided you extract the necessary dataset-specific plain predicates and format it as an event-stream. More information about this is available below and here.

This README provides a brief overview of this tool, instructions for use, and a description of the fields in the task configuration file (see representative configs in sample_configs/). Please refer to the ACES Documentation for more detailed information.

Installation

For MEDS v0.3.3

pip install es-aces

For ESGPT

  1. Install EventStreamGPT (ESGPT):

Clone EventStreamGPT:

git clone https://github.com/mmcdermott/EventStreamGPT.git

Install with dependencies from the root directory of the cloned repo:

pip install -e .

Note: To avoid potential dependency conflicts, please install ESGPT first before installing ACES. This ensures compatibility with the polars version required by ACES.

Instructions for Use

  1. Prepare a Task Configuration File: Define your predicates and task windows according to your research needs. Please see below or here for details regarding the configuration language.
  2. Prepare Dataset & Predicates DataFrame: Process your dataset according to instructions for the MEDS or ESGPT standard so you can leverage ACES to automatically create the predicates dataframe. Alternatively, you can also create your own predicates dataframe directly (more information below and here).
  3. Execute Query: A query may be executed using either the command-line interface or by importing the package in Python:

Command-Line Interface:

aces-cli data.path='/path/to/data/directory/or/file' data.standard='<meds|esgpt|direct>' cohort_dir='/directory/to/task/config/' cohort_name='<task_config_name>'

For help using aces-cli:

aces-cli --help

Python Code:

from aces import config, predicates, query
from omegaconf import DictConfig

# create task configuration object
cfg = config.TaskExtractorConfig.load(config_path="/path/to/task/config.yaml")

# get predicates dataframe
data_config = DictConfig(
    {
        "path": "/path/to/data/directory/or/file",
        "standard": "<meds|esgpt|direct>",
        "ts_format": "%m/%d/%Y %H:%M",
    }
)
predicates_df = predicates.get_predicates_df(cfg=cfg, data_config=data_config)

# execute query and get results
df_result = query.query(cfg=cfg, predicates_df=predicates_df)
  1. Results: The output will be a dataframe of subjects who satisfy the conditions defined in your task configuration file. Timestamps for the start/end boundaries of each window specified in the task configuration, as well as predicate counts for each window, are also provided. Below are sample logs for the successful extraction of an in-hospital mortality cohort:
aces-cli cohort_name="inhospital_mortality" cohort_dir="sample_configs" data.standard="meds" data.path="MEDS_DATA"
2024-09-24 02:06:57.362 | INFO     | aces.__main__:main:153 - Loading config from 'sample_configs/inhospital_mortality.yaml'
2024-09-24 02:06:57.369 | INFO     | aces.config:load:1258 - Parsing windows...
2024-09-24 02:06:57.369 | INFO     | aces.config:load:1267 - Parsing trigger event...
2024-09-24 02:06:57.369 | INFO     | aces.config:load:1282 - Parsing predicates...
2024-09-24 02:06:57.380 | INFO     | aces.__main__:main:156 - Attempting to get predicates dataframe given:
standard: meds
ts_format: '%m/%d/%Y %H:%M'
path: MEDS_DATA/
_prefix: ''

2024-09-24 02:07:58.176 | INFO     | aces.predicates:generate_plain_predicates_from_meds:268 - Loading MEDS data...
2024-09-24 02:07:01.405 | INFO     | aces.predicates:generate_plain_predicates_from_esgpt:272 - Generating plain predicate columns...
2024-09-24 02:07:01.579 | INFO     | aces.predicates:generate_plain_predicates_from_esgpt:276 - Added predicate column 'admission'.
2024-09-24 02:07:01.770 | INFO     | aces.predicates:generate_plain_predicates_from_esgpt:276 - Added predicate column 'discharge'.
2024-09-24 02:07:01.925 | INFO     | aces.predicates:generate_plain_predicates_from_esgpt:276 - Added predicate column 'death'.
2024-09-24 02:07:07.155 | INFO     | aces.predicates:generate_plain_predicates_from_esgpt:279 - Cleaning up predicates dataframe...
2024-09-24 02:07:07.156 | INFO     | aces.predicates:get_predicates_df:642 - Loaded plain predicates. Generating derived predicate columns...
2024-09-24 02:07:07.167 | INFO     | aces.predicates:get_predicates_df:645 - Added predicate column 'discharge_or_death'.
2024-09-24 02:07:07.772 | INFO     | aces.predicates:get_predicates_df:654 - Generating special predicate columns...
2024-09-24 02:07:07.841 | INFO     | aces.predicates:get_predicates_df:681 - Added predicate column '_ANY_EVENT'.
2024-09-24 02:07:07.841 | INFO     | aces.query:query:76 - Checking if '(subject_id, timestamp)' columns are unique...
2024-09-24 02:07:08.221 | INFO     | aces.utils:log_tree:57 -

trigger
┣━━ input.end
┃   ┗━━ input.start
┗━━ gap.end
    ┗━━ target.end

2024-09-24 02:07:08.221 | INFO     | aces.query:query:85 - Beginning query...
2024-09-24 02:07:08.221 | INFO     | aces.query:query:89 - Static variable criteria specified, filtering patient demographics...
2024-09-24 02:07:08.221 | INFO     | aces.query:query:99 - Identifying possible trigger nodes based on the specified trigger event...
2024-09-24 02:07:08.233 | INFO     | aces.constraints:check_constraints:110 - Excluding 14,623,763 rows as they failed to satisfy '1 <= admission <= None'.
2024-09-24 02:07:08.249 | INFO     | aces.extract_subtree:extract_subtree:252 - Summarizing subtree rooted at 'input.end'...
2024-09-24 02:07:13.259 | INFO     | aces.extract_subtree:extract_subtree:252 - Summarizing subtree rooted at 'input.start'...
2024-09-24 02:07:26.011 | INFO     | aces.constraints:check_constraints:176 - Excluding 12,212 rows as they failed to satisfy '5 <= _ANY_EVENT <= None'.
2024-09-24 02:07:26.052 | INFO     | aces.extract_subtree:extract_subtree:252 - Summarizing subtree rooted at 'gap.end'...
2024-09-24 02:07:30.223 | INFO     | aces.constraints:check_constraints:176 - Excluding 631 rows as they failed to satisfy 'None <= admission <= 0'.
2024-09-24 02:07:30.224 | INFO     | aces.constraints:check_constraints:176 - Excluding 18,165 rows as they failed to satisfy 'None <= discharge <= 0'.
2024-09-24 02:07:30.224 | INFO     | aces.constraints:check_constraints:176 - Excluding 221 rows as they failed to satisfy 'None <= death <= 0'.
2024-09-24 02:07:30.226 | INFO     | aces.extract_subtree:extract_subtree:252 - Summarizing subtree rooted at 'target.end'...
2024-09-24 02:07:41.512 | INFO     | aces.query:query:113 - Done. 44,318 valid rows returned corresponding to 11,606 subjects.
2024-09-24 02:07:41.513 | INFO     | aces.query:query:129 - Extracting label 'death' from window 'target'...
2024-09-24 02:07:41.514 | INFO     | aces.query:query:142 - Setting index timestamp as 'end' of window 'input'...
2024-09-24 02:07:41.606 | INFO     | aces.__main__:main:188 - Completed in 0:00:44.243514. Results saved to 'sample_configs/inhospital_mortality.parquet'.

Task Configuration File

The task configuration file allows users to define specific predicates and windows to query your dataset. Below is a sample generic configuration file in its most basic form:

predicates:
  predicate_1:
    code: ???
  ...

trigger: ???

windows:
  window_1:
    start: ???
    end: ???
    start_inclusive: ???
    end_inclusive: ???
    has:
      predicate_1: (???, ???)

    label: ???
    index_timestamp: ???
  ...

Sample task configuration files for 6 common tasks are provided in sample_configs/. All task configurations can be directly extracted using 'direct' mode on sample_data/sample_data.csv as this predicates dataframe was designed specifically to capture concepts needed for all tasks. However, only inhospital_mortality.yaml and imminent-mortality.yaml would be able to be extracted on sample_data/esgpt_sample and sample_data/meds_sample due to a lack of required concepts in the datasets (predicates are defined as per the MEDS sample data by default; modifications will be needed for ESGPT).

Predicates

Predicates describe the event at a timestamp. Predicate columns are created to contain predicate counts for each row of your dataset. If the MEDS or ESGPT data standard is used, ACES automatically computes the predicates dataframe needed for the query from the predicates fields in your task configuration file. However, you may also choose to construct your own predicates dataframe should you not wish to use the MEDS or ESGPT data standard.

Example predicates dataframe .csv:

subject_id,timestamp,death,admission,discharge,covid,death_or_discharge,_ANY_EVENT
1,12/1/1989 12:03,0,1,0,0,0,1
1,12/1/1989 13:14,0,0,0,0,0,1
1,12/1/1989 15:17,0,0,0,0,0,1
1,12/1/1989 16:17,0,0,0,0,0,1
1,12/1/1989 20:17,0,0,0,0,0,1
1,12/2/1989 3:00,0,0,0,0,0,1
1,12/2/1989 9:00,0,0,0,0,0,1
1,12/2/1989 15:00,0,0,1,0,1,1

There are two types of predicates that can be defined in the configuration file, "plain" predicates, and "derived" predicates.

Plain Predicates

"Plain" predicates represent explicit values (either str or int) in your dataset at a particular timestamp and has 1 required code field (for string categorical variables) and 4 optional fields (for integer or float continuous variables). For instance, the following defines a predicate representing normal SpO2 levels (a range of 90-120 corresponding to rows where the lab column is O2 saturation pulseoxymetry (%)):

normal_spo2:
  code: lab//O2 saturation pulseoxymetry (%)     # required <str>//<str>
  value_min: 90                                  # optional <float/int>
  value_max: 120                                 # optional <float/int>
  value_min_inclusive: true                      # optional <bool>
  value_max_inclusive: true                      # optional <bool>
  other_cols: {}                                 # optional <dict>

Fields for a "plain" predicate:

Derived Predicates

"Derived" predicates combine existing "plain" predicates using and / or keywords and have exactly 1 required expr field: For instance, the following defines a predicate representing either death or discharge (by combining "plain" predicates of death and discharge):

# plain predicates
discharge:
  code: event_type//DISCHARGE
death:
  code: event_type//DEATH

# derived predicates
discharge_or_death:
  expr: or(discharge, death)

Field for a "derived" predicate:

A special predicate _ANY_EVENT is always defined, which simply represents any event, as the name suggests. This predicate can be used like any other predicate manually defined (ie., setting a constraint on its occurrence or using it as a trigger - more information below!).

Special Predicates

There are also a few special predicates that you can use. These do not need to be defined explicitly in the configuration file, and can be directly used:

_ANY_EVENT: specifies any event in the data (ie., effectively set to 1 for every single row in your predicates dataframe)

_RECORD_START: specifies the beginning of a patient's record (ie., effectively set to 1 in the first chronological row for every subject_id)

_RECORD_END: specifies the end of a patient's record (ie., effectively set to 1 in the last chronological row for every subject_id)

Trigger Event

The trigger event is a simple field with a value of a predicate name. For each trigger event, a prediction by a model can be made. For instance, in the following example, the trigger event is an admission. Therefore, in your task, a prediction by a model can be made for each valid admission (ie., samples remaining after extraction according to other task specifications are considered valid). You can also simply filter to a cohort of one event (ie., just a trigger event) should you not have any further criteria in your task.

predicates:
  admission:
    code: event_type//ADMISSION

trigger: admission                    # trigger event <predicate>

Windows

Windows can be of two types, a temporally-bounded window or an event-bounded window. Below is a sample temporally-bounded window configuration:

trigger: admission

input:
  start: NULL
  end: trigger + 24h
  start_inclusive: True
  end_inclusive: True
  has:
    _ANY_EVENT: (5, None)

In this example, the window input begins at NULL (ie., the first event or the start of the time series record), and ends at 24 hours after the trigger event, which is specified to be a hospital admission. The window is inclusive on both ends (ie., both the first event and the event at 24 hours after the admission, if any, is included in this window). Finally, a constraint of 5 events of any kind is placed so any valid window would include sufficient data.

Two fields (start and end) are required to define the size of a window. Both fields must be a string referencing a predicate name, or a string referencing the start or end field of another window. In addition, it may express a temporal relationship by including a positive or negative time period expressed as a string (ie., + 2 days, - 365 days, + 12h, - 30 minutes, + 60s). It may also express an event relationship by including a sequence with a directional arrow and a predicate name (ie., -> predicate_1 indicating the period until the next occurrence of the predicate, or <- predicate_1 indicating the period following the previous occurrence of the predicate). Finally, it may also contain NULL, indicating the first/last event for the start/end field, respectively.

start_inclusive and end_inclusive are required booleans specifying whether the events, if present, at the start and end points of the window are included in the window.

The has field specifies constraints relating to predicates within the window. For each predicate defined previously, a constraint for occurrences can be set using a string in the format of (<min>, <max>). Unbounded conditions can be specified by using None or leaving it empty (ie., (5, None), (8,), (None, 32), (,10)).

label is an optional field and can only exist in ONE window in the task configuration file if defined (an error is thrown otherwise). It must be a string matching a defined predicate name, and is used to extract the label for the task.

index_timestamp is an optional field and can only exist in ONE window in the task configuration file if defined (an error is thrown otherwise). It must be either start or end, and is used to create an index column used to easily manipulate the results output. Usually, one would set it to be the time at which the prediction would be made (ie., set to end in your window containing input data). Please ensure that you are validating your interpretation of index_timestamp for your task. For instance, if index_timestamp is set to the end of a particular window, the timestamp would be the event at the window boundary. However, in some cases, your task may want to exclude this boundary event, so ensure you are correctly interpreting the timestamp during extraction.

FAQs

Static Data

In MEDS, static variables are simply stored in rows with null timestamps. In ESGPT, static variables are stored in a separate subjects_df table. In either case, it is feasible to express static variables as a predicate and apply the associated criteria normally using the patient_demographics heading of a configuration file. Please see here and here for examples and details.

Complementary Tools

ACES is an integral part of the MEDS ecosystem. To fully leverage its capabilities, you can utilize it alongside other complementary MEDS tools, such as:

Alternative Tools

There are existing alternatives for cohort extraction that focus on specific common data models, such as i2b2 PIC-SURE and OHDSI ATLAS.

ACES serves as a middle ground between PIC-SURE and ATLAS. While it may offer less capability than PIC-SURE, it compensates with greater ease of use and improved communication value. Compared to ATLAS, ACES provides greater capability, though with slightly lower ease of use, yet it still maintains a higher communication value.

Finally, ACES is not tied to a particular common data model. Built on a flexible event-stream format, ACES is a no-code solution with a descriptive input format, permitting easy and wide iteration over task definitions. It can be applied to a variety of schemas, making it a versatile tool suitable for diverse research needs.

Future Roadmap

Usability

Coverage

Generalizability

Causal Usage

Additional Tasks

Natural Language Interface

Acknowledgements

Matthew McDermott, PhD | Harvard Medical School

Alistair Johnson, DPhil | Independent

Jack Gallifant, MD | Massachusetts Institute of Technology

Tom Pollard, PhD | Massachusetts Institute of Technology

Curtis Langlotz, MD, PhD | Stanford University

David Eyre, BM BCh, DPhil | University of Oxford

For any questions, enhancements, or issues, please file a GitHub issue. For inquiries regarding MEDS or ESGPT, please refer to their respective repositories. Contributions are welcome via pull requests.