Strategies to Mitigate Age-Related Bias in Machine Learning: Scoping Review

Background Research suggests that digital ageism, that is, age-related bias, is present in the development and deployment of machine learning (ML) models. Despite the recognition of the importance of this problem, there is a lack of research that specifically examines the strategies used to mitigate age-related bias in ML models and the effectiveness of these strategies. Objective To address this gap, we conducted a scoping review of mitigation strategies to reduce age-related bias in ML. Methods We followed a scoping review methodology framework developed by Arksey and O’Malley. The search was developed in conjunction with an information specialist and conducted in 6 electronic databases (IEEE Xplore, Scopus, Web of Science, CINAHL, EMBASE, and the ACM digital library), as well as 2 additional gray literature databases (OpenGrey and Grey Literature Report). Results We identified 8 publications that attempted to mitigate age-related bias in ML approaches. Age-related bias was introduced primarily due to a lack of representation of older adults in the data. Efforts to mitigate bias were categorized into one of three approaches: (1) creating a more balanced data set, (2) augmenting and supplementing their data, and (3) modifying the algorithm directly to achieve a more balanced result. Conclusions Identifying and mitigating related biases in ML models is critical to fostering fairness, equity, inclusion, and social benefits. Our analysis underscores the ongoing need for rigorous research and the development of effective mitigation approaches to address digital ageism, ensuring that ML systems are used in a way that upholds the interests of all individuals. Trial Registration Open Science Framework AMG5P; https://osf.io/amg5p


Rationale 3
Describe the rationale for the review in the context of what is already known.Explain why the review questions/objectives lend themselves to a scoping review approach.
Page 4: "The topic of digital ageism is gaining prominence in scholarly discussions, leading researchers to investigate these phenomena from various perspectives [19,20,24,25].Prior investigations have focused on developing conceptual frameworks to comprehend and define the nature and implications of digital ageism [13].
Previous reviews of facial-image datasets have also found that older adults, particularly older adults from the 85+ demographic, are under-represented in a majority of datasets [18].While this research has been foundational in identifying and characterizing these biases, there is now a critical need to focus on the mitigation strategies that can address age-related bias in AI systems.The purpose of this scoping review is to advance this crucial discussion by shedding light on the mitigation strategies currently being used to address age-related bias in AI.By bridging the gap between theory and practice, this research aims to pave the way for meaningful and impactful interventions that can rectify biases and promote inclusivity in the digital age" Objectives 4 Provide an explicit statement of the questions and objectives being addressed with reference to their key elements (e.g., population or participants, concepts, and context) or other relevant key elements used to The research questions for this review can be found on page 4: "1) Which mitigation strategies have been employed to address age-related bias in artificial intelligence, and how successful were these strategies?2) Specifically, what types of biases Describe the methods of charting data from the included sources of evidence (e.g., calibrated forms or forms that have been tested by the team before their use, and whether data charting was done independently or in duplicate) and any processes for obtaining and confirming data from investigators.
Pages 6/7: Studies were selected if they acknowledged the presence of any bias against older adults in either their data or results and the researchers then took any action to correct that bias, regardless of its effectiveness.For example, publications were selected based on whether authors attempted to enhance the performance of their model on older demographics, regardless of the success of their efforts.Biweekly meetings were held to discuss progress of the charting process.Disagreements were resolved via discussion or by having the first author (CC) act as a third reviewer.The extracted information was converted into table format which allowed the authors to develop a narrative description according to the type of mitigation strategy (

Data items 11
List and define all variables for which data were sought and any assumptions and simplifications made.
Datasets were analyzed for the age categories used to categorize their subjects, as well as the number of subjects in each category.
Publications were analyzed according to the framework for bias in Machine Learning, created by Mehrabi et al.
Papers were included if they contained significant discussion of a specific type of bias, or whether or not they demonstrated said bias in their methods (for example, using a biased dataset would expose a paper's methods to representation bias).
Papers were included if their authors a) acknowledged that a certain type of bias against older adults was present in their paper, and then b) made a conscious effort to mitigate that bias.The effort did not need to be successful for the paper to be included, the authors only needed to make the attempt.For example, in Liang et a., the first effort at mitigating bias by balancing the dataset was unsuccessful, but the effort made by altering the algorithm was successful.

Critical appraisal of individual sources of evidence § 12
If done, provide a rationale for conducting a critical appraisal of included sources of evidence; describe the methods used and how this information was used in any data synthesis (if appropriate).
Papers were critically appraised for strategies used to mitigate bias against older adults specifically.Simply acknowledging that a bias affecting older adults was present in the data or outcomes of their publication was not enough, researchers had to attempt to mitigate it in some way.

Synthesis of results 13
Describe the methods of handling and summarizing the data that were charted.
Page 6, Section: "Selection of Sources of Evidence and Charting the Data" RESULTS

Selection of sources of evidence 14
Give numbers of sources of evidence screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage, ideally using a flow diagram.
See Figure 1 for a PRISMA flowsheet of the results.

Characteristics of sources of evidence 15
For each source of evidence, present characteristics for which data were charted and provide the citations.For each included source of evidence, present the relevant data that were charted that relate to the review questions and objectives.
See Tables 1 and 2  † A more inclusive/heterogeneous term used to account for the different types of evidence or data sources (e.g., quantitative and/or qualitative research, expert opinion, and policy documents) that may be eligible in a scoping review as opposed to only studies.This is not to be confused with information sources (see first footnote).‡ The frameworks by Arksey and O'Malley (6) and Levac and colleagues (7) and the JBI guidance (4, 5) refer to the process of data extraction in a scoping review as data charting.§ The process of systematically examining research evidence to assess its validity, results, and relevance before using it to inform a decision.This term is used for items 12 and 19 instead of "risk of bias" (which is more applicable to systematic reviews of interventions) to include and acknowledge the various sources of evidence that may be used in a scoping review (e.g., quantitative and/or qualitative research, expert opinion, and policy document).
See Table 1 in the Multimedia Appendix