Conference
View in EDS

Influence of Gender-Specific Data Imbalance on scGPT Fine-Tuning for Single-Cell Genomics

Bibliographic Details
Title: Influence of Gender-Specific Data Imbalance on scGPT Fine-Tuning for Single-Cell Genomics
Authors: Al Amin, Mohammad Aman Ullah, Filienko, Daniil, Qin, Hong
Source: Computer Science Faculty Publications
Publisher Information: ODU Digital Commons
Publication Year: 2025
Collection: Old Dominion University: ODU Digital Commons
Subject Terms: Artificial intelligence—Sex differences, Bias in artificial intelligence, Big data—Sex differences, Bioinformatics, Cell populations—Sex differences, Cells—Classification—Data processing, Computational biology, Data imbalance (Machine learning), Gene expression—Analysis, Genomics—Data processing, Health disparities—Sex factors, Hematopoietic cells—Classification, Immune system—Sex differences, Machine learning—Sex differences, Medical informatics—Moral and ethical aspects, Monocytes—Sex differences, Natural killer cells—Sex differences, Precision medicine—Sex differences, Sex differences (Biology), Sex factors in disease, Single-cell RNA sequencing, T cells—Sex differences, Transformers (Machine learning), Artificial Intelligence and Robotics, Bioethics and Medical Ethics, Genetics and Genomics
Description: The transformer-based foundation model scGPT has demonstrated strong capabilities in analyzing high-dimensional single-cell RNA sequencing data. However, the impact of demographic factors, particularly gender, on model performance remains insufficiently understood. Gender is known to influence cell type compositions in the immune system. Here, using the gender-sensitive cell type composition in immune system, we comprehensively evaluate how the gender-sensitive imbalance of training data influences the performance of scGPT in cell type predictions. We fine-tune scGPT on male-only, female-only, and mixed-gender subsets from two large-scale datasets containing immune cells. We use a logit difference to measure the confidence gap between the true label and the actual model prediction. The confidence gap is zero for perfect classifications and negative for incorrect predictions. We find that training and testing configurations with aligned gender distributions generally show higher prediction confidence, while mismatched gender during training and testing, especially when training excludes one gender, leads to substantial confidence drops. We also find that training with mixed-gender data promotes more balanced generalization, but does not eliminate all biases. We conclude that gender-specific data imbalance, represented by immune cell type subpopulation variation between women and men, can influence fine-tuning of scGPT and its performance in cell type classification, highlighting the importance of addressing such demographic biases in biomedical AI models.
Document Type: conference object
Language: unknown
Relation: https://digitalcommons.odu.edu/computerscience_fac_pubs/419; https://doi.org/10.1609/aaaiss.v7i1.36914
DOI: 10.1609/aaaiss.v7i1.36914
Availability: https://digitalcommons.odu.edu/computerscience_fac_pubs/419
https://doi.org/10.1609/aaaiss.v7i1.36914
Rights: © 2023, Association for the Advancement of Artificial Intelligence. All rights reserved. "In the Returned Rights section of the AAAI copyright form, authors are specifically granted back the right to use their own papers for noncommercial uses, such as inclusion in their dissertations or the right to deposit their own papers in their institutional repositories, provided there is proper attribution. The published version is not available for posting outside the AAAI Digital Library."
Accession Number: edsbas.D7EC9F01
Database: BASE
FullText Text:
  Availability: 0
CustomLinks:
  – Url: https://digitalcommons.odu.edu/computerscience_fac_pubs/419#
    Name: EDS - BASE (ns324271)
    Category: fullText
    Text: View record from BASE
Header DbId: edsbas
DbLabel: BASE
An: edsbas.D7EC9F01
RelevancyScore: 985
AccessLevel: 3
PubType: Conference
PubTypeId: conference
PreciseRelevancyScore: 984.935607910156
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Influence of Gender-Specific Data Imbalance on scGPT Fine-Tuning for Single-Cell Genomics
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Al+Amin%2C+Mohammad+Aman+Ullah%22">Al Amin, Mohammad Aman Ullah</searchLink><br /><searchLink fieldCode="AR" term="%22Filienko%2C+Daniil%22">Filienko, Daniil</searchLink><br /><searchLink fieldCode="AR" term="%22Qin%2C+Hong%22">Qin, Hong</searchLink>
– Name: TitleSource
  Label: Source
  Group: Src
  Data: Computer Science Faculty Publications
– Name: Publisher
  Label: Publisher Information
  Group: PubInfo
  Data: ODU Digital Commons
– Name: DatePubCY
  Label: Publication Year
  Group: Date
  Data: 2025
– Name: Subset
  Label: Collection
  Group: HoldingsInfo
  Data: Old Dominion University: ODU Digital Commons
– Name: Subject
  Label: Subject Terms
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22Artificial+intelligence—Sex+differences%22">Artificial intelligence—Sex differences</searchLink><br /><searchLink fieldCode="DE" term="%22Bias+in+artificial+intelligence%22">Bias in artificial intelligence</searchLink><br /><searchLink fieldCode="DE" term="%22Big+data—Sex+differences%22">Big data—Sex differences</searchLink><br /><searchLink fieldCode="DE" term="%22Bioinformatics%22">Bioinformatics</searchLink><br /><searchLink fieldCode="DE" term="%22Cell+populations—Sex+differences%22">Cell populations—Sex differences</searchLink><br /><searchLink fieldCode="DE" term="%22Cells—Classification—Data+processing%22">Cells—Classification—Data processing</searchLink><br /><searchLink fieldCode="DE" term="%22Computational+biology%22">Computational biology</searchLink><br /><searchLink fieldCode="DE" term="%22Data+imbalance+%28Machine+learning%29%22">Data imbalance (Machine learning)</searchLink><br /><searchLink fieldCode="DE" term="%22Gene+expression—Analysis%22">Gene expression—Analysis</searchLink><br /><searchLink fieldCode="DE" term="%22Genomics—Data+processing%22">Genomics—Data processing</searchLink><br /><searchLink fieldCode="DE" term="%22Health+disparities—Sex+factors%22">Health disparities—Sex factors</searchLink><br /><searchLink fieldCode="DE" term="%22Hematopoietic+cells—Classification%22">Hematopoietic cells—Classification</searchLink><br /><searchLink fieldCode="DE" term="%22Immune+system—Sex+differences%22">Immune system—Sex differences</searchLink><br /><searchLink fieldCode="DE" term="%22Machine+learning—Sex+differences%22">Machine learning—Sex differences</searchLink><br /><searchLink fieldCode="DE" term="%22Medical+informatics—Moral+and+ethical+aspects%22">Medical informatics—Moral and ethical aspects</searchLink><br /><searchLink fieldCode="DE" term="%22Monocytes—Sex+differences%22">Monocytes—Sex differences</searchLink><br /><searchLink fieldCode="DE" term="%22Natural+killer+cells—Sex+differences%22">Natural killer cells—Sex differences</searchLink><br /><searchLink fieldCode="DE" term="%22Precision+medicine—Sex+differences%22">Precision medicine—Sex differences</searchLink><br /><searchLink fieldCode="DE" term="%22Sex+differences+%28Biology%29%22">Sex differences (Biology)</searchLink><br /><searchLink fieldCode="DE" term="%22Sex+factors+in+disease%22">Sex factors in disease</searchLink><br /><searchLink fieldCode="DE" term="%22Single-cell+RNA+sequencing%22">Single-cell RNA sequencing</searchLink><br /><searchLink fieldCode="DE" term="%22T+cells—Sex+differences%22">T cells—Sex differences</searchLink><br /><searchLink fieldCode="DE" term="%22Transformers+%28Machine+learning%29%22">Transformers (Machine learning)</searchLink><br /><searchLink fieldCode="DE" term="%22Artificial+Intelligence+and+Robotics%22">Artificial Intelligence and Robotics</searchLink><br /><searchLink fieldCode="DE" term="%22Bioethics+and+Medical+Ethics%22">Bioethics and Medical Ethics</searchLink><br /><searchLink fieldCode="DE" term="%22Genetics+and+Genomics%22">Genetics and Genomics</searchLink>
– Name: Abstract
  Label: Description
  Group: Ab
  Data: The transformer-based foundation model scGPT has demonstrated strong capabilities in analyzing high-dimensional single-cell RNA sequencing data. However, the impact of demographic factors, particularly gender, on model performance remains insufficiently understood. Gender is known to influence cell type compositions in the immune system. Here, using the gender-sensitive cell type composition in immune system, we comprehensively evaluate how the gender-sensitive imbalance of training data influences the performance of scGPT in cell type predictions. We fine-tune scGPT on male-only, female-only, and mixed-gender subsets from two large-scale datasets containing immune cells. We use a logit difference to measure the confidence gap between the true label and the actual model prediction. The confidence gap is zero for perfect classifications and negative for incorrect predictions. We find that training and testing configurations with aligned gender distributions generally show higher prediction confidence, while mismatched gender during training and testing, especially when training excludes one gender, leads to substantial confidence drops. We also find that training with mixed-gender data promotes more balanced generalization, but does not eliminate all biases. We conclude that gender-specific data imbalance, represented by immune cell type subpopulation variation between women and men, can influence fine-tuning of scGPT and its performance in cell type classification, highlighting the importance of addressing such demographic biases in biomedical AI models.
– Name: TypeDocument
  Label: Document Type
  Group: TypDoc
  Data: conference object
– Name: Language
  Label: Language
  Group: Lang
  Data: unknown
– Name: NoteTitleSource
  Label: Relation
  Group: SrcInfo
  Data: https://digitalcommons.odu.edu/computerscience_fac_pubs/419; https://doi.org/10.1609/aaaiss.v7i1.36914
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1609/aaaiss.v7i1.36914
– Name: URL
  Label: Availability
  Group: URL
  Data: https://digitalcommons.odu.edu/computerscience_fac_pubs/419<br />https://doi.org/10.1609/aaaiss.v7i1.36914
– Name: Copyright
  Label: Rights
  Group: Cpyrght
  Data: © 2023, Association for the Advancement of Artificial Intelligence. All rights reserved. "In the Returned Rights section of the AAAI copyright form, authors are specifically granted back the right to use their own papers for noncommercial uses, such as inclusion in their dissertations or the right to deposit their own papers in their institutional repositories, provided there is proper attribution. The published version is not available for posting outside the AAAI Digital Library."
– Name: AN
  Label: Accession Number
  Group: ID
  Data: edsbas.D7EC9F01
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=edsbas&AN=edsbas.D7EC9F01
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1609/aaaiss.v7i1.36914
    Languages:
      – Text: unknown
    Subjects:
      – SubjectFull: Artificial intelligence—Sex differences
        Type: general
      – SubjectFull: Bias in artificial intelligence
        Type: general
      – SubjectFull: Big data—Sex differences
        Type: general
      – SubjectFull: Bioinformatics
        Type: general
      – SubjectFull: Cell populations—Sex differences
        Type: general
      – SubjectFull: Cells—Classification—Data processing
        Type: general
      – SubjectFull: Computational biology
        Type: general
      – SubjectFull: Data imbalance (Machine learning)
        Type: general
      – SubjectFull: Gene expression—Analysis
        Type: general
      – SubjectFull: Genomics—Data processing
        Type: general
      – SubjectFull: Health disparities—Sex factors
        Type: general
      – SubjectFull: Hematopoietic cells—Classification
        Type: general
      – SubjectFull: Immune system—Sex differences
        Type: general
      – SubjectFull: Machine learning—Sex differences
        Type: general
      – SubjectFull: Medical informatics—Moral and ethical aspects
        Type: general
      – SubjectFull: Monocytes—Sex differences
        Type: general
      – SubjectFull: Natural killer cells—Sex differences
        Type: general
      – SubjectFull: Precision medicine—Sex differences
        Type: general
      – SubjectFull: Sex differences (Biology)
        Type: general
      – SubjectFull: Sex factors in disease
        Type: general
      – SubjectFull: Single-cell RNA sequencing
        Type: general
      – SubjectFull: T cells—Sex differences
        Type: general
      – SubjectFull: Transformers (Machine learning)
        Type: general
      – SubjectFull: Artificial Intelligence and Robotics
        Type: general
      – SubjectFull: Bioethics and Medical Ethics
        Type: general
      – SubjectFull: Genetics and Genomics
        Type: general
    Titles:
      – TitleFull: Influence of Gender-Specific Data Imbalance on scGPT Fine-Tuning for Single-Cell Genomics
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Al Amin, Mohammad Aman Ullah
      – PersonEntity:
          Name:
            NameFull: Filienko, Daniil
      – PersonEntity:
          Name:
            NameFull: Qin, Hong
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 01
              Type: published
              Y: 2025
          Identifiers:
            – Type: issn-locals
              Value: edsbas
            – Type: issn-locals
              Value: edsbas.oa
          Titles:
            – TitleFull: Computer Science Faculty Publications
              Type: main
ResultId 1