SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2307.02791 · 2023

The Role of Subgroup Separability in Group-Fair Medical Image Classification

arXiv · PDF · Open in the Atlas

Code that ran

4 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 3 failures of our instrument; 0 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2307.02791.json · badge: /api/ran/2307.02791.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
biomedia-mira/subgroup-separability canonical (classifier label)
classifier basis: census
4 ran · 0 no recorded run
FunctionStatusWhere it lives
compute_fairness_metrics Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
biomedia-mira/subgroup-separability/models/metrics.py
code served (permissive licence) · get_code("a47e2fd30b169ba6")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Set random seed for reproducibility
np.random.seed(0)

# Create small sample data
N = 10  # Number of samples

# scores: predicted probability scores in [0, 1], shape (N,)
scores = np.random.rand(N)

# labels: binary labels (0 or 1), shape (N,)
labels = np.random.randint(0, 2, size=N)

# attributes: subgroup labels (0 or 1), shape (N,)
attributes = np.random.randint(0, 2, size=N)

# threshold: float value for converting scores to predictions
threshold = 0.5

ARGS = {
    'scores': scores,
    'labels': labels,
    'attributes': attributes,
    'threshold': threshold
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_overall_metrics Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
biomedia-mira/subgroup-separability/models/metrics.py
code served (permissive licence) · get_code("169381619abe2aef")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Set random seed for reproducibility
np.random.seed(0)

# Create sample data for binary classification
# scores: predicted probability scores in range [0, 1], shape (N,)
N = 10
scores = np.random.rand(N).astype(np.float32)

# labels: binary labels (0 or 1), shape (N,)
labels = np.random.randint(0, 2, size=N).astype(int)

# threshold: default value for converting scores to predictions
threshold = 0.5

ARGS = {
    'scores': scores,
    'labels': labels,
    'threshold': threshold
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_transforms_for_train Ran
ran; our contract draft was wrong, not the code
biomedia-mira/subgroup-separability/models/transforms.py
code served (permissive licence) · get_code("c135e4280abaa186")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Sequential. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
import torchvision.transforms as transforms

torch.manual_seed(0)

IMAGENET_MEAN = [0.485, 0.456, 0.406]
IMAGENET_STD = [0.229, 0.224, 0.225]
normalise_2d = transforms.Normalize(mean=IMAGENET_MEAN, std=IMAGENET_STD)

def get_transforms_for_eval() -> nn.Module:
    """Evaluation (val/test) transforms. Input to the pipeline is a
    RGB uint8 Tensor, output is a normalised float Tensor.

    Returns:
        nn.Module: Sequential set of transforms. Call this on the images during the
            training step on GPU.
    """
    transform_list = [transforms.ConvertImageDtype(torch.float32), transforms.Resize(256), transforms.CenterCrop(224), normalise_2d]
    return nn.Sequential(*transform_list)

def get_transforms_for_train(augment: bool=True) -> nn.Module:
    """Transforms for training data. If augment is True, apply autoaugment policy.

    Input to the pipeline is a RGB uint8 Tensor, output is a normalised float Tensor.

    Args:
        augment (bool, optional): Whether to augment data. If False, simply return
            the eval transforms. Defaults to True.

    Returns:
        nn.Module: Sequential set of transforms. Call this on the images during the
            training step on GPU.
    """
    if not augment:
        return get_transforms_for_eval()
    transform_list = [transforms.ConvertImageDtype(torch.float32), transforms.RandomResizedCrop(224), transforms.RandomRotation(15), normalise_2d]
    return nn.Sequential(*transform_list)

ARGS = {
    "augment": True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
is_valid_sensitive_attribute Ran
ran, violated the contract we drafted
biomedia-mira/subgroup-separability/datasets/dataset.py
code served (permissive licence) · get_code("3e1d7db04da370ac")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

from enum import Enum

class SensitiveAttribute(Enum):
    """All datasets use one or more of these attributes and are binarized in the
    following way. We expect the csv files to have a column with the same name as
    the Key field in the mapping below."""
    sex = {0: 'Male', 1: 'Female', 'Key': 'Sex'}
    age = {0: '0-60', 1: '60+', 'Key': 'Age'}
    race = {0: 'White', 1: 'Non-White', 'Key': 'Race'}
    skintype = {0: 'Types 1-3', 1: 'Types 4-6', 'Key': 'SkinType'}

class AvailableDataset(Enum):
    """Register datasets by adding them and their available sensitive attributes here."""
    chexpert = [SensitiveAttribute.sex, SensitiveAttribute.age, SensitiveAttribute.race]
    mimic = [SensitiveAttribute.sex, SensitiveAttribute.age, SensitiveAttribute.race]
    ham10000 = [SensitiveAttribute.sex, SensitiveAttribute.age]
    papila = [SensitiveAttribute.sex, SensitiveAttribute.age]
    fitzpatrick17k = [SensitiveAttribute.skintype]

# Valid case: sex attribute is available in chexpert dataset
ARGS = {
    'a': SensitiveAttribute.sex,
    'ds': AvailableDataset.chexpert
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
typing_extensions — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 4 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We investigate performance disparities in deep classifiers. We find that the ability of classifiers to separate individuals into subgroups varies substantially across medical imaging modalities and protected characteristics; crucially, we show that this property is predictive of algorithmic bias. Through theoretical analysis and extensive empirical evaluation, we find a relationship between subgroup separability, subgroup disparities, and performance degradation when models are trained on data with systematic bias such as underdiagnosis. Our findings shed new light on the question of how models become biased, providing important insights for the development of fair medical imaging AI.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2307.02791")
get_code_for_paper("2307.02791")
have("2307.02791")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2307.02791.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2307.02791.svg)](https://syntology.ai/paper/2307.02791)

Connect an agent — have() is free.