Development of a culturally targeted chatbot to inform living kidney donor candidates of African ancestry about APOL1 genetic testing: a mixed methods study

Clinical chatbots are increasingly used to help integrate genetic testing into clinical contexts, but no chatbot exists for Apolipoprotein L1 (APOL1) genetic testing of living kidney donor (LKD) candidates of African ancestry. Our study aimed to culturally adapt and assess perceptions of the Gia® chatbot to help integrate APOL1 testing into LKD evaluation. Ten focus groups and post-focus group surveys were conducted with 54 LKDs, community members, and kidney transplant recipients of African ancestry. Data were analyzed through thematic analysis and descriptive statistics. Key themes about making Gia culturally targeted included ensuring: (1) transparency by providing Black LKDs’ testimonials, explaining patient privacy and confidentiality protections, and explaining how genetic testing can help LKD evaluation; (2) content is informative by educating Black LKDs about APOL1 testing instead of aiming to convince them to undergo testing, presenting statistics, and describing how genetic discrimination is legally prevented; and (3) content avoids stigma about living donation in the Black community. Most agreed Gia was neutral and unbiased (82%), trustworthy (82%), and words, phrases, and expressions were familiar to the intended audience (85%). Our culturally adapted APOL1 Gia chatbot was well regarded. Future research should assess how this chatbot could supplement provider discussion prior to genetic testing to scale APOL1 counseling and testing for LKD candidate clinical evaluation. Supplementary Information The online version contains supplementary material available at 10.1007/s12687-024-00698-8.


Introduction
As personalized medicine gains greater momentum, the proliferation of genetic testing has enabled more patients to make value-based preventive treatment decisions.An essential component of genetic testing is the provision of counseling by genetic counselors before and after the test.However, the limited number of genetic counselors (National Society of Genetic Counselors 2022) and gap in physician knowledge of genetics to make referrals to genetic counselors (Rasouly et al. 2023) hinder patients' access to genetic services.Increasing patients' access to genetics expertise can be addressed by alternate delivery models that leverage innovative technologies to scale genetic information delivery, such as clinical chatbots (Schmidlen et al. 2019;Nazareth et al. 2021).
Clinical chatbots, a type of "conversational agent," use a text-based interface to simulate conversation with the user Extended author information available on the last page of the article via smartphones, or other digital devices.(Allouch et al. 2021) Chatbots are invaluable for streamlining the repetitive foundational information-giving responsibilities of providers, thereby providing more time for interpreting test results and addressing emotional dimensions of patients' decision-making (National Society of Genetic Counselors 2022; Nazareth et al. 2021).Chatbots are particularly useful for delivering information about genetic conditions given patients' limited health literacy in genetics, the heightened challenge of making preference-based treatment decisions in the context of uncertainty, and the extensiveness of genetic information delivered to patients (Bombard et al. 2022).
Chatbot use can be applied across the genomics pathway, including pre-and post-test counseling, and management (Bombard et al. 2022).Chatbots have been used for many purposes (e.g., to aid in decision-making about genetic testing (Siglen et al. 2022), provide secondary findings to patients undergoing genomic sequencing (Ireland et al. 2021), provide informed consent for genetic testing (Schmidlen et al. 2019), and enable self-diagnosis (Fan et al. 2021)).Chatbot use can improve knowledge of genetic testing, informed decision-making, physical activity, self-care, and health literacy (Ireland et al. 2021;Luca et al. 2023;Owens et al. 2019;Bickmore et al. 2013;Dworkin et al. 2019;Pereira and Diaz 2019).
Given their utility, healthcare institutions are increasingly using chatbots in genetic services delivery, like Gia® (Genetic Information Assistant), the Invitae® Health Insurance Portability and Accountability Act (HIPAA)-compliant chatbot (Schmidlen et al. 2019;Allouch et al. 2021;Ray 2018).However, integrating chatbots into the clinical context may be challenging to embed within the workflow and ensure patients understand the content and usability (Nazareth et al. 2021).Research shows that patients embrace chatbots (Schmidlen et al. 2018;Ponathil et al. 2020).
In the context of kidney transplantation, which is predominated by organ scarcity, evaluating patients' eligibility is essential for providing safe and equitable treatment.Genetic testing comprises an important component of the evaluation process through screening, diagnosis, management, and prognostication of kidney disease.It also provides helpful information for risk stratification in potential living kidney donor (LKD) candidates (Caliskan et al. 2022).Given the risk of inherited kidney disease in LKDs post-donation, guidelines recommend that LKD candidates undergo genetic testing in certain clinical situations (Lentine et al. 2017).
Testing for Apolipoprotein L1 (APOL1) risk variants in LKD candidates is one such consideration currently under debate (Freedman and Poggio 2021;Kumar and Locke 2021).The presence of two risk variants is significantly associated with kidney failure in the general population, prominently among individuals of African ancestry (Parsa et al. 2013).Evidence from observational studies and case reports suggests that genetic factors contribute to an elevated risk of kidney failure post-donation among LKDs of African ancestry (Muzaale et al. 2014).APOL1 risk variants are most prevalent among sub-Saharan African, West African, and Black/African American populations, while also found among Brazilian, Jamaican, Granadian populations (11-22%), and least prevalent in Native American, Peruvian, Belizean populations (1-5%), among other populations (Nadkarni et al. 2018).APOL1 risk variants occur in Hispanic/Latino populations at < 1% (Kramer et al. 2017)."Black" and "African American" comprise terms of social identity and, as social constructions, such terms may not align precisely with one's genetic ancestry, as research shows (Lazarin et al. 2013).However, approximately 36% of all individuals identifying as Black or African American in the USA carry APOL1 risk variants, which is larger than any other ethnic or racial group in the USA (Friedman and Pollack 2011).We use "African ancestry" to refer to individuals with African descent, including Black and African American individuals who are born outside of or born inside the USA, respectively.We henceforth use "Black" inclusively to designate a shared cultural and historical global identity (Simms 2018).
The lack of evidence demonstrating a causal link between APOL1 risk variants and LKDs' graft outcomes and ethical issues of APOL1 testing have given some transplant professionals pause.Given the available evidence, the importance the transplant community places on protecting living donor's safety (Tan et al. 2015), and the 2017 KDIGO Clinical Practice Guideline on the Evaluation and Care of Living Kidney Donors which recommends "…considering APOL1 genotyping in LD candidates with sub-Saharan African ancestors" have justified the implementation of some degree of APOL1 testing into clinical practice for many surveyed transplant centers (Gordon et al. 2018;Doshi et al. 2021).Accordingly, it is essential that centers performing APOL1 genetic testing do so in an ethically and culturally targeted manner to avoid reinforcing systemic racism (Smith et al. 2023).
The presence of two APOL1 risk variants can comprise a contraindication to living donation, requiring careful counseling and consent of LKD candidates (Riella and Sheridan 2015).However, some transplant physicians lack the knowledge to educate LKDs about APOL1 testing and results for donation, and few transplant programs leverage genetic counselors to aid in counseling (Rasouly et al. 2023;Gordon et al. 2018).The optimal implementation of genetic counseling and testing in LKD evaluation remains to be developed (Caliskan et al. 2022).To date, no chatbot has been used in the LKD context or adapted to APOL1.
Underserved populations have disproportionately less access to genetic services and digital health technologies, like chatbots.Relatively little is known about how Black patients perceive chatbots for genetic testing.Studies examining chatbot use in Black individuals demonstrated acceptability and efficacy (Owens et al. 2019;Bickmore et al. 2013;Dworkin et al. 2019).Increasing equity in accessibility requires designing chatbots that foster genetic literacy and trustworthiness (Bombard et al. 2022;Fan et al. 2021).To date, no studies have adapted chatbots to Black individuals who may share cultural worldview, to enhance receptivity to their use.Adapting evidence-based interventions to the self-ascribed Black population, an underrepresented population that comprises the majority of those who have APOL1 risk variants and identify as having African ancestry, requires addressing culturally shared beliefs regarding distrust of healthcare providers and genetic testing (Gordon et al. 2018;Gordon et al. 2019).Culturally targeted care can increase Black patients' knowledge about donation and LKDs rates (Arriola et al. 2014).In this study, we adapted Gia to the APOL1 genetic testing context, and culturally adapted (Barrera et al. 2017).Gia to LKD candidates of African ancestry.Our intent was to scale genetic testing by making Gia available to LKD candidates of African ancestry to facilitate making informed decisions about APOL1 testing and ultimately living donation.

Study design
This sequential mixed methods study involved focus groups and post-focus group surveys about our APOL1 Gia chatbot (Creswell and Plano Clark 2007).Data collection occurred in an iterative process, whereby changes based on focus group feedback were applied to Gia for review in subsequent focus groups to maximize user-centeredness.Focus groups and surveys involved LKDs, community members, and living donor kidney transplant recipients of African ancestry.
Our study was conducted as part of a larger study evaluating the integration of a culturally competent APOL1 genetic testing program into LKD candidate evaluation at Northwestern University (NU) in Chicago, IL and Georgetown University (GU) in Washington, DC (Smith et al. 2023).The study was approved by the Institutional Review Boards at NU (STU00214038) and GU (STUDY00003752); NU served as the Institutional Review Board of record for GU.We used the Consolidated Criteria for Reporting Qualitative Research for quality reporting of qualitative studies (Tong et al. 2007).

Research team
Our research team adapted Gia to present information on APOL1 testing for LKDs of African ancestry.The research team included social scientists, genetic counselors, and transplant nephrologists from NU and GU, genetic counselors from Invitae, and a Scientific Advisory Board comprised of experts in nephrology, ethics and genetics, and race and genetics.

Community Advisory Board (CAB)
We established a nine-member, Chicago-based CAB, comprised of one living kidney donor, two kidney transplant recipients, and community leaders of African ancestry from organizations with expertise in kidney disease, organ donation and transplantation, and Black cultural values and community needs.During seven meetings between June 2021 and January 2022, CAB members provided feedback on Gia's content, appearance, wording, tone, and cultural sensitivity.

Adaptation
Gia allows users to interact by selecting one of the programmed response options to progress through the chat.Gia has one main chat flow path.Users can select additional information in optional chat flow paths and return to the main path, based on their selected responses.Thus, the information presented in Gia is based on the user's learning needs.
The research team adapted Gia in two ways: (1) topically, on APOL1 and living donation, and (2) culturally, to the intended group it aimed to serve-Black LKD candidates.Overall, adaptation entailed presenting APOL1 information in a conversational, neutral, unbiased, and culturally targeted manner.The topics, data, and resources included were obtained from peer-reviewed articles, the National Kidney Foundation, the National Institute of Diabetes and Digestive and Kidney Diseases, our research team, and CAB (Parsa et al. 2013;Gordon et al. 2018).Gia included information on basic genetic concepts and kidney disease, how APOL1 relates to kidney disease, prevalence of APOL1 risk variants, what APOL1 test results mean in relation to kidney disease and donation, reasons for APOL1 testing, the logistics of undergoing APOL1 testing, and the Genetic Information Nondiscrimination Act.
Cultural adaptation also involved making health communication culturally targeted, which refers to "a single intervention approach for a defined population subgroup that takes into account characteristics shared by the subgroup's members," in order to increase receptivity and acceptance of messages (Kreuter and Skinner 2000).This step is important because ensuring that health educational programs are relevant to the intended population is a core principle of public health practice (Glanz et al. 1997).
We used several strategies including (1) stakeholder engagement (i.e., involving members of the intended group, and the scientific advisory board, to provide input on chatbot development and insights into cultural dimensions); (2) addressing "deep structure" (i.e., embedding APOL1 content, design, and appearance of Gia in the context of shared cultural values, norms, and beliefs of individuals of African ancestry) to generate meaning about and enhance the salience of health messages; (3) presenting epidemiological data on the health impact of APOL1 variants on individuals of African ancestry (i.e., "13 out of every 100 (13%) people who identify as Black or African American have 2 APOL1 risk variants") to raise awareness and perceived relevance of APOL1 and promote taking preventative action; (4) addressing "surface structure" by using an avatar representative of the intended group to make the chatbot seem more familiar and comfortable; and (5) ensuring that the chatbot used language and phrasing commonly used by the intended group (Resnicow et al. 1999).

Scope
We adapted Gia as part of a multi-level intervention to integrate APOL1 testing and counseling into LKD candidate evaluation.Gia was designed for use by LKD candidates as a supplemental tool providing foundational information prior to their clinical evaluation with the transplant nephrologist.

Focus groups
Eligible individuals included English-speaking adults (age 18 + years) of self-identified African ancestry who were LKDs, community members, or living donor kidney transplant recipients.Participants were recruited via two methods: (1) LKDs or transplant recipients who donated or received a kidney within the previous 15 years were identified by medical record review, mailed a letter and/or email, and called a week later to screen for eligibility; and (2) CAB members distributed study flyers; interested community members contacted the research staff.All participants provided verbal informed consent.
Focus groups were conducted remotely and led by female moderators [PW and RY] trained in focus group facilitation by a seasoned qualitative researcher [EJG].Moderators used standard procedures to direct the discussion (Krueger 1994), published elsewhere (Smith et al. 2023).Focus groups began by providing background information about living kidney donation, relationship between APOL1 and kidney disease among individuals of African ancestry, and the purpose of Gia to ensure that participants had sufficient knowledge to respond to questions.Focus group questions evolved based on iterative updates to Gia.Thereafter, we provided participants the link to Gia to use for 8 min, and then invited initial impressions about their experience.We sought feedback on Gia's cultural sensitivity, content, design, function, organization, appearance, and tone.Research staff took field notes.After each focus group, participants completed an online survey on selected, relevant attitudes about Gia's cultural sensitivity (Guidry and Walker 1999) and demographic questions.Each focus group lasted approximately 78 min [range: 37-140] and was audio-recorded.Participants were compensated with a $50 e-gift card for their time.

Thematic analysis
Focus group audio recordings were transcribed verbatim and analyzed for emergent themes using the constant comparison and inductive and deductive coding methods (Guest et al. 2012).An a priori list of defined codes corresponding to focus group questions was developed.All responses were coded using NVivo qualitative software (QSR International).The research team [EJG, MJG, PW, JG-S, RY, ZA] used a rapid, iterative process to independently review and openly code the first set of transcripts (Keith et al. 2017).Thereafter, the research team revised the coding scheme by adjusting for new responses, with modified codes applied to prior transcripts, until reaching thematic saturation (when no new themes emerged).Transcripts were coded after establishing inter-rater reliability (kappa > 0.80) (Guest et al. 2012).We reviewed all segments pertaining to each code to develop analytic summaries that synthesized emergent patterns (Keith et al. 2017).

Descriptive statistics
Post-focus group survey and demographic data were analyzed using descriptive statistics.Medians, interquartile range (IQR), and frequencies were calculated using IBM SPSS Statistics for Windows, version 28.Fisher's exact test was used for categorical variables, and Wilcoxon rank sum test was used for continuous variables.

Culturally adapting Gia
Three interrelated themes emerged about making Gia culturally targeted to Black community perspectives: (a) ensuring transparency; (b) designing Gia as an informative, rather than persuasive, tool; and (c) avoiding stigma related to APOL1 and kidney donation within the Black community.Representative, illustrative quotes are presented in Table 2.

A. Transparency
Participants explained that transparency meant presenting information in Gia accurately and comprehensively (including positive and negative information), and reflecting LKDs' experiences.Transparency was considered beneficial for ensuring that LKD candidates of African ancestry feel comfortable proceeding with the LKD evaluation process, and for dispelling myths in the Black community about genetic testing and donation.Participants recommended the addition of testimonials or "stories" of Black LKDs discussing their APOL1 test results and their health several years postdonation.LKD testimonials could help to normalize the idea of Black individuals donating and enhance LKD candidates' comfort with APOL1 testing and the prospect of donating.
We addressed the need for transparency in several ways.First, we fostered trust by including questions and answers in Gia that were community-driven.Originally, the question "How long has the medical community known about APOL1?", which derived from our prior research (Gordon et al. 2018), was set up as an optional pathway.However, our CAB recommended making this a mandatory part of the chat that every LKD candidate would see, so that Gia did not come across as hiding information from the user.
Second, given the well-documented importance placed by people of African ancestry on protecting patient privacy and confidentiality, we addressed the disposition of APOL1 test results by stating in Gia that "Your test results will be shared with you, your transplant team, the genetic testing lab, and with authorized research personnel only."During an early focus group, a participant suggested greater transparency about who "authorized research personnel" are.Accordingly, we added a hover-over definition to this term: "This is the person who asked for your informed consent for this study, the principal investigator of the study, and other staff working to collect and analyze study data." Third, Gia highlighted the usefulness of genetic testing in LKD candidate evaluation to show people of African ancestry that this genetic test will help their donation process.Specifically, Gia states: "Genetic testing for APOL1, along with all the other donor tests, helps the transplant team better understand your personal risk of getting kidney disease after donating.Genetic testing for APOL1 may also help you feel better prepared and better informed about deciding about donating."The usefulness of genetic testing in LKD evaluation is highly relevant for Black LKD candidates because Gia presents that information to the donor to make them aware of how genetic testing fits into the entire donor evaluation process.

B. Informative
Participants emphasized the importance of making Gia "informative" to educate Black LKD candidates, rather than aim to convince them to pursue genetic testing.
We addressed the need for Gia to be informative in several ways.First, we included a pathway option "Why do APOL1 risk variants occur mostly in people of African ancestry?".To answer this question, Gia stated: "The APOL1 risk variants evolved about four thousand years ago in Sub-Saharan Africa to protect against sleeping sickness."That statement was well received by focus group members because it recast APOL1 risk variants in a positive light by highlighting a strength of people of African ancestry.We piloted the additional explanation of why APOL1 gene variants existed primarily in people of African ancestry by saying how ancestors to modern Europeans had migrated out of Africa, prior to when APOL1 evolved.However, CAB members and focus group participants suggested that we remove this additional information.CAB members expressed that the information may not be received well or may come across as insensitive because, for example, White Europeans had the choice to leave Africa.One focus group member remarked, "I don't think [the additional information] is necessary to have.Because [the additional information] does make [the chatbot] a little longer and I don't think that donors necessarily need that information." Second, Gia describes the APOL1 testing process: "APOL1 testing is voluntary, free to study participants, and involves taking a saliva sample."These three points were initially presented separately and appeared later in the chat.When participants came upon the information, they consistently indicated that the information answered questions early in the chat.As a result, this statement was deliberately presented early in Gia.Presenting this information early fostered trust by addressing user-driven information needs.
Third, Gia describes how discrimination is prevented: "There is a law called GINA (Genetic Information Nondiscrimination Act) that protects patients against discrimination by health insurance providers and employers based on genetic information."

C. Avoiding stigma
A pervasive concern was the need to reframe APOL1 in a positive manner, and to minimize comparisons between Black and White LKDs, which were perceived as off-putting.Avoiding stigma entailed framing the higher frequency of APOL1 risk variants in the population of those with African ancestry in a positive way so that LKD candidates did not feel "inferior to the White [LKDs]."Participants spoke about addressing the stigma related to living donation within the  [Gia] might be as an information tool for their health as well."(NU FG#3, Female) "Yes, I think Gia should try to convince our, the consumers of the information, that-who are the people with African ancestry-that it is OK to have a gene variant because as a result of adapting to the African savannah, people-or rather, their ancestors-tried to change their genetics [unintelligible] to that environment, which is not a bad thing.So I think Gia should try to convince them that even though this problem-or rather, the risk variant to the APOL1-is basically for the people with the African ancestry, it is not a bad thing.It is not a bad thing to them.Try to convince them, try to not make them seem inferior to the White [LKDs].Yeah, that's my take."(NU FG#2, Male) Not persuasive "Yeah.I agree.I think this should be informative, but it shouldn't be like it's a campaign, or you're trying to push people.Give the information.Sometimes a lot is too much.And as I think somebody said before you, you don't want to feel like you're trying too hard, and then you're turning people off.I think the way it is, I don't feel like there's any stereotypes or anything in the presentation, but I don't think you need to add any more as a way of persuading."(GU FG#6, Female) "What I was saying is I think that it's better to keep it the way it is.So it don't seem like you're trying to beg one race or the other to do it.So I think what you're doing is being informative.Rather be informative and let it be that because you don't want to feel like it's swaying one way, or I really have to get this group in here.You don't want to do that."(GU FG#6, Female) Avoiding stigmatizing language "I know that research came up with these numbers and there was much research and data and all of that, that came up with these percentages.But it's always been, not a sore spot, but it always gives me pause.When I see that, Black people are more likely to have this, this, this, or that.And, White people are less likely to have this, this, or that.It still gives a negative connotation.Those comparison numbers, percentages, it still gives a negative connotation to who I am as a Black person."(GU FG#8, Female) "I think it's gotten better, but some of the older generation, there is stigmas because of things that have happened at other hospitals, where things have happened and there are instances where it's true and they hold onto that.So if they're in the younger generation's ear, they've kind of planted that seed and it perpetuates from [generation] to generation.So I think having people who have already done [living donation] and gone through, having them available for people to talk to maybe, would be helpful."(GU FG#10, Female) "Do you think it would be helpful to have information in Gia about the benefits of kidney donation?Because even though I'm a donor, I still feel that there is a very big stigma within the community about organ donation."(GU FG#5) Black community by presenting statistics about kidney donation and the benefits in Gia.Participants suggested adding testimonials and statistics on "those who have donated… as well as those who are really on a waiting list, waiting" who are Black as it would be "pretty impactful" information that could encourage people in the Black community to become living donors.

Perceptions of Gia
Participants' ratings of Gia's cultural targeting indicated that our adaptations were appropriate (Table 3).Most participants "agreed" or "strongly agreed" that Gia is "neutral and unbiased" (82%), trustworthy (82%), and that "words, phrases, and expressions are familiar to the intended audience" (85%).We asked focus groups about their initial impressions immediately after using Gia and at the end of the focus group.Participants consistently regarded Gia positively and described Gia as follows: "informative," "engaging," "interactive," "straightforward," "personable," "user-friendly," and "excellent."

Incorporating CAB and focus group feedback
The CAB and focus groups provided extensive feedback and recommendations for improving Gia chatbot content, design, functionality, and appearance.Supplemental Table 2 presents the prominent, commonly shared recommendations, whether each recommended change was implemented, the rationale for not implementing the change, if applicable, and how Gia was changed in response to the recommendation.
We incorporated as many suggestions as possible to be user-centered.For example, several participants said that the tone of some response options (i.e., "I see" or "Okay") could be perceived as "abrupt" or "curt" within the Black community.Alternative responses (i.e., "That makes sense" and "I understand") were considered appropriate in subsequent focus groups and used instead.Additionally, participants across several focus groups desired supplemental resources in Gia about Black LKD testimonials of donation experiences and self-care strategies, which led to the inclusion of links to websites and educational videos.To make Gia sensitive to LKD candidates, CAB members suggested including a strong phrase to convey gratitude to the LKD candidate for considering donation, which became: "Thank you very much for considering donating the gift of life." However, not all suggestions were accommodated due to technical constraints, study cost constraints, idiosyncratic or contradictory feedback, or our intended use of Gia in clinical practice.For example, many participants suggested that Gia provide an audio function for those who cannot or prefer not to read, but implementation of this feature was not in scope for this project.Some participants recommended including information about whether the LKD candidate would be able to donate with two APOL1 risk variants.We deliberately did not include this information because Gia's intent was to provide foundational information about APOL1, and to relegate the shared decision-making process to the transplant nephrologist's meeting with the LKD candidate.

Discussion
This is the first study to culturally adapt a chatbot to individuals of African ancestry and to the APOL1 genetic testing context.We found that individuals of African ancestry were highly receptive to our chatbot for informing LKD candidates about APOL1 genetic testing in a culturally targeted manner.By adapting Gia to the cultural considerations of individuals of African ancestry through community engagement, our chatbot overcomes limitations of prior studies which designed chatbots for the end user rather than with the end user (Kramer et al. 2020).Three themes that emerged from making Gia culturally targeted to LKD candidates of African ancestry included making content transparent, informative, and non-stigmatizing.Transparency entailed providing balanced statistical information, while being informative entailed providing testimonials of Black LKDs, which fostered trustworthiness.A pervasive concern was the importance of reframing APOL1 in a positive manner (i.e., as an evolutionary protection from sleeping sickness), and to minimize comparisons between Black and White LKDs.Our prior research among Black LKDs similarly found that framing APOL1 in a positive manner was important to avoid stigmatizing language about Black individuals and their health (Gordon et al. 2019).Other research suggests that chatbots could become more trustworthy by disclosing the data sources, accuracy of predictions, and how diagnostic reports are produced (Fan et al. 2021).Our participants' positive perceptions of Gia suggests that our chatbot was viewed as trustworthy and holds promise for use in the genetic testing and counseling process.
Bombard et al. suggest, digital genomic tools can reinforce disparities in access to genetic services if tools are not developed in a trustworthy manner (Bombard et al. 2022).Chatbots are appealing for increasing patients' access to genetic information, thereby fostering equity in precision medicine (Nazareth et al. 2021).Ensuring that Gia's content and design are culturally targeted for use among LKDs of African ancestry enhances Gia's ability to reach and resonate with people who have been traditionally underserved, and are most at risk of kidney disease and APOL1 risk variants, and thereby increase its uptake.Our chatbot advances The National Human Genome Research Institute's call to leverage digital health-enabled genomics to increase public genomic literacy (Bonham and Green 2021) and address ethnocultural gaps in application of digital health technologies (Bombard et al. 2022).
However, as participants recommended, providing a voice-over feature would further increase access to the chatbot content for those who cannot read, which companies designing chatbots should address.The digital divide is closing whereby comparable rates of smartphone use were reported among Black (83%) and White (85%) Pew national survey respondents (Atske and Perrin 2021).
Gia was designed to foster the integration of genetic testing and counseling for LKD evaluation.Future research will examine use of the APOL1 Gia chatbot in real-world settings using self-reported perceptions and log data.Our research team is implementing Gia into donor clinics to evaluate its impact on LKD candidates' decision-making about APOL1 genetic testing and donation (Smith et al. 2023).Critics' fears that chatbots will dehumanize healthcare delivery and replace genetic counselors (Reddy et al. 2019) are unfounded because chatbots aid, rather than replace, human decision-making (Nazareth et al. 2021).Conversely, it is possible that providers may come to over-rely on chatbots.While some commentators fear that physicians' growing reliance on artificial intelligence such as chatbots will crowd out physicians who do not use artificial intelligence (AI), this is unlikely as it will likely take concerted effort by healthcare organizations to invest in strategies including human-AI collaboration to repurpose roles for increased efficiencies (Sezgin 2023).Accordingly, our intent is to implement this digital health technology to increase healthcare efficiencies in genomic medicine by releasing time for transplant nephrologists to discuss APOL1 genetic testing and living donation and foster patient-centered care delivery.Our future research will disseminate Gia to scale APOL1 genetic testing and counseling for LKDs.
There are several strengths of this study.This was a multi-site study among LKDs, living donor kidney transplant recipients, and community members of African ancestry, which can foster transferability of findings.Our study provided novel insights into attitudes about chatbot use among individuals of African ancestry, about whom relatively little is known.The chatbot was rigorously developed in a patient-centered manner: Individuals from diverse racial/ethnic and educational backgrounds were involved in the process of designing Gia so that the chat would culturally resonate and be accepted by a broad population.The Scientific Advisory Board provided input that supported face and content validity, while the Community Advisory Board provided feedback and focus group participants of African ancestry supported face validity and cultural insights.We followed established methodologies for survey development, and conducted rigorous pre-testing of the chatbot content to ensure its clarity and comprehensiveness.
This study may have limitations.We cannot exclude the possibility that participants did not understand information.Because the research occurred primarily with individuals in urban areas, the views may not be transferrable to those in rural areas.Although more females participated in focus groups than males, the proportion of female LKDs mirrored that of LKDs nationally.However, the average age of our focus group participants was higher than the average age of LKDs.Some focus group participants may have been hesitant to fully express their views in a group setting.Although moderators were trained, some biases in moderating style across sites could have occurred.

Conclusion
Our findings suggest that individuals of African ancestry perceived Gia as informative about APOL1 genetic testing, neutral and unbiased, trustworthy, and culturally targeted to individuals of African ancestry.Future research will assess the impact of APOL1 Gia on informed decision-making about APOL1 genetic testing among LKD candidates of African ancestry.As a supplement to physician medical care, Gia may augment APOL1 counseling and testing in LKD clinical evaluation.

Table 1
Focus group participant demographic characteristics

Table 2
Representative illustrative quotations about Gia-APOL1 and using chatbots by major theme and subtheme Not to sound like a broken record, but I agree with everyone.I think the presentation was informative.It was easy to read, it flowed, it provided information and I think it is a[n] excellent tool to present to someone who is considering donating.(GU FG#6, Female) "The most important part for me was knowing that the genetic testing results are going to be kept safe.Gia has] information that is accurate and is, so you just want the information that's given you to be accurate and to be truthful…" (NU FG#2, Female) "I would suggest a statistic of how many African Americans have donated, but I don't know if that's a good number to put out there, just to make others comfortable, and let them know that there are African Americans that donate."(NU FG#4, Female) "That's the main option.I think if you're expressing how many people who've actually done it, and are living, and surviving, and things like that because most people jump into it because it's a relative.They don't necessarily do it because, 'Oh, that's what I really want to do.'But they do it because they love their family members, and they want them to live.So I think open transparency is a good way of marketing it."(GU FG#1) "That's a tough one because the reality of it is that, there, like you said, there's a level of distrust.And I mean, the only way that you can really get away from that is, I agree with the stories and other testimonials, but I think what's important is that has to be a clear message being sent out that, transparency is there."(GU FG#10, Male) Informative "All of this, just more information, needs to be at the beginning, and that may even be because if someone just knows it's a saliva test versus a blood test, you know, 'OK, I might be willing to do that.
' Then you say, well, then it's no cost to you, if you're willing to do this as a donor.That might nudge them on, you know, 'Well, let me just see what's going on.'And not only for them being a donor.

Table 3
Focus group participants' attitudes about cultural sensitivity of the APOL1 Gia chatbot, N = 54