Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

Bias01:22

Bias

4.3K
Bias refers to any tendency that prevents a question from being considered unprejudiced. In research, bias occurs when one outcome or answer is selected or encouraged over others in sampling or testing. Bias can occur during any research phase, including study design, data collection, analysis, and publication.
In statistics, a sampling bias is created when a sample is collected from a population, and some members of the population are not as likely to be chosen as others (remember, each member...
4.3K
Confirmation Biases01:31

Confirmation Biases

5.5K
The confirmation bias is the tendency to focus on information that confirms our existing beliefs and ignore information that is inconsistent with our expectations. For example, if you think that your professor is not very nice, you notice all of the instances of rude behavior exhibited by the professor while ignoring the countless pleasant interactions he is involved in on a daily basis. Have you ever fallen prey to the confirmation bias, either as the source or target of such bias?
5.5K
Stereotype Content Model02:16

Stereotype Content Model

14.8K
The Stereotype Content Model (SCM) was first proposed by Susan Fiske and her colleagues (Fiske, Cuddy, Glick & Xu, 2002; see also Fiske, 2012 and Fiske, 2017). The SCM specifies that when someone encounters a new group, they will stereotype them based on two metrics: warmth—or that group’s perceived intent, and how likely they are to provide help or inflict harm—and competence—or their ability to carry out that objective. Depending on the warmth-competence...
14.8K
Stereotypes, Prejudice, and Discrimination02:55

Stereotypes, Prejudice, and Discrimination

90.3K
Humans are very diverse and although we share many similarities, we also have many differences. The social groups we belong to help form our identities (Tajfel, 1974). These differences may be difficult for some people to reconcile, which may lead to prejudice toward people who are different. Prejudice is a negative attitude and feeling toward an individual based solely on one’s membership in a particular social group (Allport, 1954; Brown, 2010). Prejudice is common against people who...
90.3K
Bias in Epidemiological Studies01:29

Bias in Epidemiological Studies

391
Biases can arise at various stages of research, from study design and data collection to analysis and interpretation. Recognizing and addressing these biases is essential to ensure the validity and reliability of epidemiological findings.Broadly speaking, biases in epidemiology fall into three main categories: selection bias, information bias, and confounding. A more detailed description of possible biases is:  
391
Regression Toward the Mean01:52

Regression Toward the Mean

6.3K
Regression toward the mean (“RTM”) is a phenomenon in which extremely high or low values—for example, and individual’s blood pressure at a particular moment—appear closer to a group’s average upon remeasuring. Although this statistical peculiarity is the result of random error and chance, it has been problematic across various medical, scientific, financial and psychological applications. In particular, RTM, if not taken into account, can interfere when...
6.3K

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Predictors of Repeat Nondiagnostic Cytology in Thyroid Fine-Needle Aspiration.

Canadian Association of Radiologists journal = Journal l'Association canadienne des radiologistes·2026
Same author

Evaluating data heterogeneity's impact on convolutional neural network performance in medical imaging.

International journal of computer assisted radiology and surgery·2026
Same author

Cognitively Biased Prompt Effects on Large Language Model Accuracy for Radiology Board-style Examination Questions.

Radiology. Artificial intelligence·2026
Same author

Deep learning for pediatric synovial recess distension detection in hemophilia: synthetic image augmentation with styleGAN2-ADA.

International journal of computer assisted radiology and surgery·2026
Same author

Evaluating the impact of sex bias on AI models in musculoskeletal ultrasound of joint recess distension.

PloS one·2025
Same author

Promising initial evaluation: the Shock Thyroid Imaging Score (STIS) could predict mortality in hemodynamically unstable trauma patients.

Emergency radiology·2025

Related Experiment Video

Updated: Jul 29, 2025

Assessment of Mouse Judgment Bias through an Olfactory Digging Task
12:10

Assessment of Mouse Judgment Bias through an Olfactory Digging Task

Published on: March 4, 2022

2.7K

Can Machine Learning Be Better than Biased Readers?

Atsuhiro Hibi1,2, Rui Zhu3, Pascal N Tyrrell1,2,4

  • 1Institute of Medical Science, University of Toronto, Toronto, ON M5S 1A8, Canada.

Tomography (Ann Arbor, Mich.)
|May 23, 2023
PubMed
Summary

Machine learning (ML) models can overcome reader bias in medical imaging datasets. Using regularized loss functions effectively mitigates biased labels when multiple annotators work without consensus.

Keywords:
annotation biaschest X-rayconvolutional neural networklabeling consensusmachine learning

More Related Videos

A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
12:18

A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment

Published on: January 11, 2020

7.6K
Eye-tracking to Distinguish Comprehension-based and Oculomotor-based Regressive Eye Movements During Reading
05:54

Eye-tracking to Distinguish Comprehension-based and Oculomotor-based Regressive Eye Movements During Reading

Published on: October 18, 2018

6.3K

Related Experiment Videos

Last Updated: Jul 29, 2025

Assessment of Mouse Judgment Bias through an Olfactory Digging Task
12:10

Assessment of Mouse Judgment Bias through an Olfactory Digging Task

Published on: March 4, 2022

2.7K
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
12:18

A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment

Published on: January 11, 2020

7.6K
Eye-tracking to Distinguish Comprehension-based and Oculomotor-based Regressive Eye Movements During Reading
05:54

Eye-tracking to Distinguish Comprehension-based and Oculomotor-based Regressive Eye Movements During Reading

Published on: October 18, 2018

6.3K

Area of Science:

  • Medical Imaging
  • Machine Learning
  • Artificial Intelligence

Background:

  • Training machine learning (ML) models in medical imaging necessitates extensive labeled data.
  • Dividing annotation tasks among multiple readers without consensus can introduce dataset bias and degrade ML model performance.
  • Addressing reader variability is crucial for reliable AI in healthcare.

Purpose of the Study:

  • To evaluate the capability of ML algorithms to overcome biases introduced by multiple annotators lacking consensus.
  • To investigate methods for improving ML model prediction accuracy on potentially biased medical imaging datasets.

Main Methods:

  • A publicly available pediatric pneumonia chest X-ray dataset was utilized.
  • Artificial random and systematic errors were introduced to simulate a non-consensus labeling scenario.
  • A Resnet18 convolutional neural network (CNN) was employed as a baseline, with a second model incorporating a regularized loss function for comparison.

Main Results:

  • Biases from false positive, false negative, and random errors (5-25%) led to a decrease in Area Under the Curve (AUC) by 0-14% for the baseline CNN classifier.
  • The CNN model with a regularized loss function demonstrated improved AUC, ranging from 75-84%, compared to the baseline model's 65-79%.

Conclusions:

  • ML algorithms possess the potential to surmount individual reader biases in the absence of consensus labeling.
  • Regularized loss functions are recommended for their ease of implementation and effectiveness in mitigating biased labels from multi-reader annotation tasks.