Related Experiment Video
Updated: Aug 9, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.9K
Algorithmic encoding of protected characteristics in chest X-ray disease detection models
Ben Glocker1, Charles Jones1, Mélanie Bernhardt1
1Department of Computing, Imperial College London, London, SW7 2AZ, UK.
Ebiomedicine
|February 15, 2023
Summary
AI in healthcare can worsen health disparities. This study introduces a subgroup analysis methodology to detect and understand how protected characteristics influence AI disease detection models, ensuring fairer clinical decision-making.
Area of Science:
- Artificial Intelligence in Medicine
- Medical Imaging Analysis
- Health Disparities Research
Background:
- Artificial intelligence (AI) in clinical decision-making risks amplifying health disparities due to biases in historical training data.
- It remains challenging to determine if algorithms utilize protected characteristics, leading to disparate performance, especially with limited data from underserved populations.
- Understanding how dataset biases manifest in predictive models is crucial for equitable AI deployment in healthcare.
Purpose of the Study:
- To explore methodologies for subgroup analysis in image-based disease detection models.
- To investigate performance disparities across demographic subgroups (race, biological sex) in deep learning models for chest X-ray analysis.
- To assess the encoding of protected characteristics and their impact on disease detection performance.
Main Methods:
- Utilized two public chest X-ray datasets: CheXpert and MIMIC-CXR.
- Employed test set resampling, transfer learning, multitask learning, and model inspection techniques.
- Assessed the relationship between protected characteristics and disease detection performance across subgroups.
Main Results:
- Confirmed subgroup performance disparities, including shifted true and false positive rates, partially mitigated by correcting for population and prevalence shifts.
- Transfer learning alone was insufficient to determine if protected patient information influenced predictions.
- A combined approach of test-set resampling, multitask learning, and model inspection provided insights into how protected characteristics are encoded in deep neural networks.
Conclusions:
- Subgroup analysis is essential for identifying AI performance disparities in disease detection.
- Statistical differences across subgroups must be considered when analyzing potential biases.
- The proposed comprehensive framework facilitates further research into the root causes of AI-driven health disparities.
Related Concept Videos
01:20Imaging Studies for Cardiovascular System III: X-Ray
229
The most common cardiovascular diagnostic test is an X-ray. It produces images of the heart, blood vessels, and adjacent structures.
Definition and Purpose
An X-ray, or radiograph, is a non-invasive method that uses ionizing radiation to take images of internal structures. It is mainly used in cardiac imaging to examine the heart, lungs, and major blood vessels, aiming to identify abnormalities in the heart's size, shape, and position, such as heart failure, congenital defects, and vascular...
Definition and Purpose
An X-ray, or radiograph, is a non-invasive method that uses ionizing radiation to take images of internal structures. It is mainly used in cardiac imaging to examine the heart, lungs, and major blood vessels, aiming to identify abnormalities in the heart's size, shape, and position, such as heart failure, congenital defects, and vascular...
229
01:24X-ray Imaging
5.7K
German physicist Wilhelm Röntgen (1845–1923) was experimenting with electrical current when he discovered that a mysterious and invisible "ray" would pass through his flesh but leave an outline of his bones on a screen coated with a metal compound. In 1895, Röntgen made the first durable record of the internal parts of a living human: an "X-ray" image (as it came to be called) of his wife’s hand. Scientists worldwide quickly began their own experiments with...
5.7K
01:30Radiological Investigation I: X-ray and CT
328
Radiological investigations, including X-rays and computed tomography (CT) scans, are critical for diagnosing and evaluating various medical conditions. These imaging techniques provide valuable insights into the body's internal structures, aiding in the detection of abnormalities, assessment of disease progression, and development of treatment strategies. This article delves into two primary radiological investigations, chest X-rays and CT scans, outlining their purpose, procedures, and...
328
01:25Imaging Studies for Cardiovascular System VI: Calcium -Scoring CT
55
Calcium-Scoring CT ScanA calcium-scoring CT scan, also known as coronary artery calcium (CAC) scan, detects calcium deposits in the coronary arteries. This test assesses the risk of coronary artery disease (CAD), which can lead to cardiovascular events such as angina, heart failure, and sudden cardiac arrest.A calcium-scoring CT scan is generally recommended for individuals at intermediate risk of CAD without symptoms. It includes:Men aged 40-75 and women aged 50-75: Especially those with a...
55
01:15Receiver Operating Characteristic Plot
294
A ROC (Receiver Operating Characteristic) plot is a graphical tool used to assess the performance of a binary classification model by illustrating the trade-off between sensitivity (true positive rate) and specificity (false positive rate). By plotting sensitivity against 1 - specificity across various threshold settings, the ROC curve shows how well the model distinguishes between classes, with a curve closer to the top-left corner indicating a more accurate model. The area under the ROC curve...
294