Implementing a chatbot to promote hereditary breast & ovarian cancer genetic screening in women’s health: identifying barriers and facilitators to screening adoption

Abstract

Background

To promote genetic screening among women at risk for hereditary breast and ovarian cancer (HBOC), the American College of Obstetricians and Gynecologists recommends that risk assessment be integrated into practice. Chatbots like the Genetic Information Assistant (Gia®) are increasingly implemented to expand access to hereditary genetic screening. Factors that impact chatbot implementation for HBOC risk screening and women’s uptake are not fully realized. To refine implementation strategies prior to full scale implementation, we sought to identify women’s perceived facilitators/barriers to adopting Gia screening in a rural population within a large healthcare system in the southern United States.

Methods

We recruited both women who agreed to screen using Gia (and then recommended for genetic testing based on National Comprehensive Cancer Network guidelines) as well as women who opted not to do the screen from three Women’s Health clinics (OB/GYN) in a northern rural region of Florida. We conducted in-depth, semi-structured interviews with 17 women (nine adopted the screen, eight did not). We conducted a thematic analysis to identify and further define barriers/facilitators to women’s uptake of Gia for HBOC cancer risk screening in obstetrics/gynecology care.

Results

Women identified six factors that inhibited and/or facilitated their willingness to use Gia for screening: 1) cancer risk perception, 2) communication with their clinician, 3) feasibility of screening, 4) fiscal and insurance concerns, 5) technology trust/distrust, and 6) previous genetic testing experience. Findings illustrate how each factor functioned as a facilitator and/or barrier in women’s uptake (e.g., technology being trusted for accuracy versus concerns for data privacy and security).

Conclusions

While chatbots can extend women’s cancer risk screening access, particularly in rural regions and with college-educated women, systems-level (cost) and individual-level factors (perceived risk, misconceptions about repeating genetic testing) should be addressed to promote adoption. Women’s interaction with a clinician may be a key implementation strategy for addressing these factors to personalize the screening opportunity and promote their chatbot screening adoption.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-025-23488-4.

Background

Every year in the United States nearly 300,000 women are diagnosed with breast cancer, while an additional 20,000 women are diagnosed with ovarian cancer [1]. The five-year survival rate for breast cancer is 91% whereas for ovarian cancer it is notably less at 51% [1]. Each year, nearly 9,000 women diagnosed with breast or ovarian cancer also carry an inherited genetic mutation, typically a variant of the BRCA1 or BRCA2 genes, which significantly increases their lifetime risk of developing hereditary breast or ovarian cancer (HBOC) by 60–82% [2–4]. Although much of HBOC is linked to pathogenic variants in these two genes, other genetic mutations are also linked to women’s elevated HBOC risk [5].

To reduce cancer incidence and ensure women have the opportunity to use risk-reducing measures, women’s access to HBOC risk screening, as well as genetic testing when recommended, is imperative. According to the National Comprehensive Cancer Network (NCCN) guidelines, when women better understand their risk, which includes identifying inherited mutations, they can make critical risk management decisions prior to or even after a cancer diagnosis, such as increased surveillance and risk-reducing intervention (e.g., risk-reducing surgery or chemoprevention) [6]. In addition, women can then communicate their known risk to other family members who may also be recommended for testing, thereby promoting cascade testing with elevated risk individuals—a targeted approach to further reduce cancer incidences [7]. While it is imperative women understand their HBOC risk, a recent study found that access to genetic screening is still limited as 1.3 million women who would qualify for HBOC genetic testing still have not received testing [8].

In order to promote genetic screening and testing awareness among women at risk for HBOC, the American College of Obstetricians and Gynecologists now recommends that HBOC risk assessment be implemented into clinical practice to reduce cancer incidence and provide personalized patient care [9, 10]. Personalized HBOC diagnostic and screening programs in obstetrics/gynecology (OB/GYN) clinical practice are still relatively rare, in part due to the lack of a streamlined screening assessment [11]. Instead, the current model “relies on a patient-provider interaction to start the testing process, an interaction that often occurs by chance” (p. 2) [11]. Across healthcare practice, barriers to risk assessment and genetic testing are also tied to a lack of awareness about HBOC risk among health professionals as well as lack of prioritization of (and, thus, reduced patient access to) genetic counseling resources and practitioners in clinical settings (i.e., certified genetic counselors, CGCs) [8]. Population-based, routine screening may address oversight of patients at elevated risk while also placing less responsibility for initiating this conversation onto clinicians who are balancing multiple priorities and challenging time constraints in clinic [11].

Innovative healthcare delivery approaches to genetic risk screening are vital to consider in reducing such barriers [11–13]. Implementing artificial intelligence (AI) strategies in the delivery of healthcare has surged in recent decades in response to barriers to care and healthcare needs [14]. In genomic medicine in the United States, chatbots like the Genetic Information Assistant (Gia® – a HIPAA-compliant and SOC2-certified electronic chatbot created by Clear genetics), have been increasingly implemented in healthcare practice to expand patients’ access to hereditary genetic screening [7, 15, 16]. Chatbots utilize aspects of AI (e.g., natural language processing) to engage in bi-directional communication with patients facilitating interaction about their family history by modeling realistic human behavior (emotions and language) to ask risk-related questions and exchange genomic information [17]. The innovative use of chatbots in genomic settings has begun to extend globally, such as the implementation of Rosa in Norway and Dr. Joy/KakaoTalk in Korea [18, 19].

Chatbots are showing promise among both patients and within the healthcare system and may be an especially viable tool in expanding access to HBOC risk screening in OB/GYN care to ultimately reduce cancer incidence. A recent systematic review reported high patient satisfaction when using chatbots [20]. Recent studies also demonstrate the utility and acceptability of chatbots among patients for a number of services, including delivering information about genetic testing, providing pre-testing genetic counseling, and for sharing genetic risk information with other family members, which in turn can lead to cascade testing and ensure more elevated risk family members are tested [7, 17–19, 21–23]. Patients also report appreciating the convenience of using a chatbot to receive information and to share it with their broader family system [7, 18, 24]. Furthermore, from a healthcare delivery standpoint, chatbots can supplement the lack of genetic screening resources for patients, reduce CGCs’ workload, and increase the number of women at risk for HBOC being screened [14].

Yet, factors that impact the implementation of chatbots for HBOC screening are not fully realized [21]. In genomic medicine, research is just beginning to examine and understand patients’ perceptions using chatbots [14, 18]. Although factors such as feasibility and utility have been highlighted in studies among patients who have used chatbots [18], little is known about factors impacting the first step in the success of implementation—women’s adoption of the genetic screening via a chatbot—a primary implementation outcome of concern, particularly with AI approaches to genomic care [25]. In addition to understanding factors that impact adoption prior to scaling-up implementation [26], the barriers and facilitators to adoption should be considered and contextualized within the target population to promote adoption with the distinct community or culture of women the health service will serve.

Target population for implementation

Interventions should be “tailored to different communities and health care settings to facilitate broader understanding and adoption of evidence-based services” [27] (p. 6). Although the context of the intervention is a core component of successful implementation and, thus, implementation science, rarely is it appropriately considered [28]. Not considering the unique values and concerns of the community ultimately inhibits the potential efficacy of the implemented intervention as well as the potential evidence it could hold for other populations [29]. Little attention is paid to the factors driven by the context of a distinct community that influence their adoption of an intervention. Given this concerning gap in implementation science, recent evidence-informed action steps have been proposed that include prioritizing research focused on the context of implementation from stakeholders’ perspectives to best tailor implementation strategies to promote adoption and health equity [30].

The catchment area served by UF Health is comprised of rural and small-metropolitan counties in North Florida with higher age-adjusted advanced cancer incidence and mortality rates compared to state and national rates. A 2021 report of the catchment area found that women’s healthcare, including cancer screenings, were among the most difficult services for women in the community to access [31]. To increase health equity by promoting access to HBOC risk screening, since July 2020, approximately 957 patients seen across two women’s health clinics within this catchment area were invited to use Gia® for genetic risk screening via email invitation or during an OB/GYN clinical encounter. As of October 2023, approximately 45% of women invited to screen via Gia adopted the screening opportunity. Of these women, 42% were recommended for genetic testing based on National NCCN criteria. We sought to understand what factors promoted or inhibited women’s adoption of the Gia screening to refine implementation strategies prior to full scale implementation.

Materials and methods

To capture stakeholders’ insights on what influenced their Gia adoption decision, we conducted semi-structured phone interviews from October to December 2023. All procedures were approved by UF’s Institutional Review Board (202301507).

Participants and recruitment

Women were patients at one of two Women’s Health clinics within the UF Health system where they received care from an OB/GYN. We sought participants who represented one of two groups: those who completed Gia screening (group 1), and those who did not complete Gia screening (group 2). Women were eligible to participate in this study if 1) they were 65 years or younger, 2) had been offered Gia screening through UF Women’s Health (per clinic records), and 3) either adopted Gia (group 1) with screening results that met NCCN criteria for genetic testing or did not adopt Gia (group 2).

Women were recruited through MyChart® (Epic Systems Corporation©), a secure online health management platform that allows patients to access their health information and directly connect to their care team. All of the eligible women were users of MyChart. Women who met the eligibility criteria were sent a recruitment message in MyChart containing study information, eligibility criteria, study contact information, and a link to an online form hosted on Qualtrics. Those interested in participating in the interview study were asked to follow the link to the form, which included screening questions to confirm eligibility, ascertain interview scheduling preferences, and obtain contact information (phone number and email address). Eligible women were then contacted via email by a member of the study team to schedule an interview (see Fig. 1 for additional information).

Fig. 1.

Recruitment flow

Materials and procedures

Prior to their interview, participants were emailed an informed consent form describing the study, participant rights, and study contact information. At the start of their interview, participants were asked to verbally agree to participate in the study (per IRB- approved procedures). Interviews were audio-recorded and professionally transcribed verbatim.

The lead and senior authors [ENW, CLF] developed the semi-structured, in-depth interview guide (see Supplementary Materials) to capture women’s experiences with genetic screening via a chatbot and to explore their perceptions of factors that played a role in their decision to adopt genetic screening (or not) using the innovative approach of Gia (i.e., a chatbot). Demographic questions including personal and family cancer history were asked at the end of the interview. Interviews were conducted by the lead author (ENW—who has extensive qualitative training and implementation science experience in health behavior) and overseen by the senior author (CLF—an expert in qualitative methodology, implementation science, and women’s HBOC risk management decisions). Interviews lasted on average 20 min. Upon completion, women were emailed a $50 e-gift card. Professional transcriptions resulted in 123 pages of data.

Data analysis

To promote interviewer responsivity and identifying the extent of thematic saturation, analysis was conducted concurrently with data collection, a best practice in qualitative analysis, to allow for both a rapid analysis during data collection to precede a comprehensive thematic analysis of transcripts [32–34]. During data collection, the lead and senior author used a matrix-type rapid analysis template (RAT) that corresponded with the guide and key study domains of interest [33, 35]. The RAT was updated to record findings immediately after each interview, along with thematic and operationalization memos [36]. The RAT was segmented by group 1 or 2 (those who did vs. did not use Gia) to triangulate findings to capture similarities and differences in factors that impact adoption, further promoting rigor in the analysis [32–34]. Through this iterative process of concurrent data collection and rapid analysis, we identified patterns and developed a typology of preliminary themes that were subsequently used to develop a codebook for a formal, in-depth thematic analysis. Established thematic saturation criteria (repetition, recurrence, and forcefulness) were used, and the extent of thematic saturation was also identified and confirmed by conducting additional interviews and concurrent data collection and analysis [37, 38].

The first author (ENW) used the codebook to deductively and inductively thematically analyze transcripts using a constant comparative method approach, meeting weekly with the senior author (CLF), who reviewed analysis of all themes, to validate the typology and refine the codebook [39, 40]. Closed coding was used with a priori themes while inductive coding was used to ensure patterns potentially missed in the rapid analysis could be captured. Axial coding was also conducted on data aligned with each theme to develop thematic properties to further define each theme (i.e., facilitator/barrier) [41]. A codebook with definitions, descriptions, and examples of themes and properties was developed and updated throughout formal analysis [42]. The extent of thematic saturation was also noted, with themes emerging on average by 56% of participants and by as many as 94%.

Results

Interviews were conducted with 17 women who represented two groups: 1) women who completed the Gia screening and were recommended for genetic testing based on NCCN guidelines (n = 9); and 2) women who did not complete the Gia screening (n = 8). Of the women in group 2, two reported refusing to do it, three forgot about it, and three had no recall of it. Women were on average 50.6 years (SD = 11, 31–64). Most (65%) identified as non-Hispanic white. All had completed some college education. Most women (76%) had children (on average 1.4 children), who ranged in age from 3 to 34 (M = 22.2, SD = 0.94). Nearly half (n = 9) of women had at least one daughter (6 women in group 1; 2 in group 2). Most women (82%) had a family cancer history, with 59% reporting a family history of breast and ovarian cancer, and 12% reporting a family history of another gynecological cancer (e.g., uterine, cervical). Three women (17%) had a personal history of breast cancer. Demographic information collectively and by group are reported in Table 1.

Table 1.

Participant demographics

Demographic variable	Both Groups	Group 1 (n = 9)	Group 2 (n = 8)
Age	M (SD)	M (SD)	M (SD)
Average participant age in years	50.6 (11)	50.9 (12.2)	47.7 (11)
Race & Ethnicity	% (n)	% (n)	% (n)
White/Not Hispanic or Latino	65% (11)	89% (8)	38% (3)
Asian/Not Hispanic or Latino	18% (3)	0% (0)	38% (3)
White/Hispanic or Latino	11% (2)	11% (1)	13% (1)
Black or African American/Not Hispanic or Latino	6% (1)	0% (0)	13% (1)
Employment Status
Employed full-time	53% (9)	33% (3)	75% (6)
Employed part-time	18% (3)	22% (2)	13% (1)
Retired	18% (3)	33% (3)	0% (0)
Other (full-time stay-at-home parent)	11% (2)	11% (1)	13% (1)
Education
Some college or 2-year degree	24% (4)	33% (3)	13% (1)
Bachelor’s/four-year degree	24% (4)	33% (3)	13% (1)
Master’s degree	28% (5)	22% (2)	38% (3)
Professional or another doctorate (e.g., JD, MD, PhD)	24% (4)	11% (1)	38% (3)
Marital Status
Married	59% (10)	67% (6)	50% (4)
Not married/Divorced	41% (7)	33% (3)	50% (4)
Children
Those with children	76% (13)	78% (7)	75% (6)
Those with daughters	50% (8)a	67% (6)	29% (2)a
Cancer History
Past cancer diagnosis	17% (3)	22% (2)	13% (1)
Family history of cancer (any)	82% (14)	100% (9)	63% (5)
Family history of breast or ovarian cancer	59% (10)	78% (7)	38% (3)

^amissing value for 1 participant

Factors impacting Gia screening uptake

Women identified six factors that inhibited and/or facilitated their willingness to adopt the Gia screening: 1) cancer risk perception, 2) communication with their clinician, 3) feasibility of screening, 4) fiscal concerns, 5) technology trust/distrust, and 6) previous genetic testing experience. Women’s lived experiences illustrate how each factor can function as a facilitator and/or barrier to doing cancer screening via a chatbot like Gia. Thematic properties for each factor identified in italics below further characterize or define how each is important to consider when implementing chatbot screening tools with similar populations.

Cancer risk perception

Women’s cancer risk perception both promoted and inhibited Gia uptake. Women shared three ways that their own perception of cancer risk promoted their adoption. Women described how having a known personal and/or family cancer history motivated them to do the screen to assess their own risk:

In 2018, I lost my dad to pancreatic cancer, which shares the markers for breast cancer. And in 2021, I lost my mom. … I had an appointment [with doctor] … and I was like, “Look, with what's gone on with my parents, can I get this BRCA testing?” (white woman, age 57, group 1)

Women also wanted to reduce risk, which motivated their uptake. They wanted to be able to manage their risk or take risk-reducing measures if the risk assessment indicated they should:

If I’ve got something, I’d rather know now, and so maybe I can be on the lookout for what it might cause, other than being hit off guard with oh, all of a sudden now I’ve got this kind of cancer. So, at least now I can do tests to check things. (white woman, age 63, group 1)

Finally, women were motivated to do Gia by altruism. They wanted to know their risk to inform relatives, particularly those with younger female relatives:

I have two daughters and three nieces. After I got the [Gia] results, I sent the information to all of them because I'm the only—my mother passed away—so, I'm the only female. … I thought if I'm carrying some sort of genetic trait, I'm the only one who could give them that information. (white woman, age 64, group 1)

Women also described two ways their cancer risk perception inhibited their willingness to do the Gia screening. Women who opted not to do the Gia screening acknowledged having a fear of risk results: “It's also, to be honest, kind of scary, finding that [risk] out, because then it's like, ‘Well, now, do I have to do something about it?’” (Asian woman, age 31, group 2). Additionally, women with a lack of perceived risk, opted not to do Gia. These women believed they had no cancer risk because they had no known history of cancer in their family. Thus, to them, Gia screening was not necessary.

Actually, I’m very fortunate. I have a pretty healthy family. … That's probably the other reason why I didn't complete [Gia]. … It's just I don't feel that it's a risk for me. … There's no concern for risk because I don't have a family history. … Now if I had a family history, I definitely would have brought it up [to my doctor]. (Asian woman, age 51, group 2)

It is also noteworthy that some of these women shared that openly sharing health information in their family was not typically practiced or that in their “culture … the older generations, they don't really share as much information” (Asian woman, age 48, group 2). Similarly, this woman explained,

I'm Asian, so it's kind of something we don't talk about because it's bad luck. So, it's cultural. … It's not just bad luck, but it's also a boundary kind of issue because it's private, especially about GYN stuff. (Asian woman, age 51, group 2).

Communication with their clinician

Women in both groups reported that they talked to their clinician about Gia or that they did not have a discussion (or couldn’t recall a conversation): five in group 1 and four in group 2 did not recall any discussion. Yet, women in both groups identified the importance of communication with their clinician in their decision to do/not do the Gia screen. Women described how having a conversation with a clinician promoted uptake. Women who talked to a clinician about Gia shared that the discussion helped to clarify personal risk, explain why Gia was relevant to them, and Gia’s purpose (to identify heightened risk). A woman explained:

I think it's really the doctor. … She's the one that went ahead and sent it [Gia] to me, based on [my] history. So, without her doing that, I don't know that I would have done it if a doctor hadn't said,"Look, maybe you should go ahead and do this so we can see if you've got any kind of genes there that has caused cancer.” (white woman, age 63, group 1).

In contrast, not talking to a clinician inhibited uptake. Women who did not do the screening acknowledged not being sure why they should participate in Gia without having more information about Gia’s relevance to them:

It would be helpful to know what the doctor's reasons would be [to do Gia]. Like, if it was based on my history, my family history, I think that could also hold some weight in encouraging a patient to go and do Gia. (Asian woman, age 31, group 2)

This seemed especially pertinent when women perceived they did not have cancer risk: “[Clinicians] always take the family history, but it's never been my mom or my sister [with cancer], so I don't think a genetic test has ever really come up” (white woman, age 31, group 2).

Feasibility of screening

The feasibility of screening via a chatbot enhanced both groups of women’s willingness for three reasons. Women appreciated that Gia was fast and convenient due to its digital format: “I mean, the more you could do that [cancer screening] electronically, the better. It's just efficient” (white woman, age 64, group 1). Gia was convenient and accessible providing women with flexibility as to when they could complete it, so long as they remembered to do it, as this woman remarked: “[The] online format probably is a bit more convenient, because some people who work and things, it might be hard for them to find the time, if it [was] a phone interview type thing” (Asian woman, age 31, group 2).

Women who completed the screen with Gia also found it to be “user-friendly” as it was intuitive, easy to navigate, and easy to answer screening questions (white woman, age 63, group 1). As another woman further explained, “[Gia was] pretty easy, and I think that the questions that [Gia] asked were valid” (Hispanic/Latina woman, age 51, group 1). Finally, women liked receiving results quickly. They received the assessment and recommendation from Gia (i.e., about doing/not doing genetic testing) in a timely manner. The importance of receiving time-sensitive results also transferred to getting genetic testing results, as a woman recommended for genetic testing, shared: “I did like a 23andMe at one point and that took weeks and weeks and weeks. And it [screening and testing through clinic] wasn't like that. I remember it being quicker than I expected it to be” (white woman, age 39, group 1).

Fiscal concerns

Women acknowledged that financial questions and concerns made them hesitant to adopt the Gia screening, noting three types of fiscal barriers. Women expressed concerns about future insurance coverage: “If mine [results] were positive, … my concern would have been that I would have had pre-existing condition in case something happened [I would want to make sure], that my insurance would be not denied” (white woman, age 64, group 1). Women also had concerns about other types of insurance coverage: “I was also worried about … how much my insurance would have to pay in addition to being able to get life and house insurance” (Asian woman, age 51, group 2). In addition, women disclosed concerns about costs for next steps based on the Gia screening, like the immediate fiscal impact of screening or (if needed) testing. A woman who started Gia chose not to finish, in part due to fiscal concerns:

I was also worried about if I did want to get genetic screening how much out of my pocket would I have to pay? … [And] about how to interpret the results, and if I had to need a genetic counselor, would my insurance pay for it? (Asian woman, age 51, group 2).

They also voiced concerns about long-term costs if risk-reducing measures were recommended or if they developed cancer in the future. A woman who tested positive for a BRCA variant said, “I know that the treatments and everything get really expensive, so we're [me and my husband] talking about maybe saving some money or something for that if that problem is to come” (white woman, age 51, group 1).

Technology trust/distrust

Women shared how their trust in technology promoted uptake whereas distrust inhibited it. Women noted the importance of data security and privacy. When women trusted that Gia was safe and secure, which was informed by their trust in the dissemination source (i.e., their clinic), it promoted their trust in Gia and subsequent uptake:

I feel more confident, the way that it [Gia invitation] came, that it's a secure database. I was just reading yesterday or today that 23andMe got hacked, and that I would never feel secure with, but something like this, I feel like it's in the proper privacy protections. (white woman, age 64, group 1)

In comparison, when unsure of how secure/private their data would be, women were not willing to do a screening via Gia: “If it's used on multiple people, how is that information stored in the chat? … The answers that you put, where does that information go? Where is it stored?” (Black woman, age 32, group 2). This included concerns about how their information would be shared: “I would want to make sure that it's not something that is being used by some—like, [whether] my information is going to some third party, or something like that” (white woman, age 47, group 2). Women also shared how their distrust in technology accuracy made them hesitant to adopt the screen. For example, a woman with a family cancer history shared her concerns about Gia:

[What if it] wouldn't really capture the increased risk and, therefore, it wouldn’t advocate or recommend the actual genetic testing. … That was my only hesitation was that if it wasn't, I guess, sensitive enough to catch it, that I would miss the opportunity for the genetic testing because of AI. (white woman, age 39, group 1).

Previous genetic testing experience

Women who had previously completed genetic testing at some point shared how this influenced whether or not they participated in Gia, but with divergent beliefs. For some women, they believed that they didn’t need to repeat genetic testing. They thought that genetic testing only needed to be done once: “I pretty much disregarded it [Gia invitation], because I had already done genetic testing, and I already knew my risk” (white woman, age 47, group 2). Other women were motivated to complete Gia to update previous genetic test results, either because they had a different panel or they understood that as science evolves, testing improves or changes: “I know genetics is developing rapidly, and so I think it's always good to check in. I know I don't have BRCA or any of the known genetic variants, but the genetic variants of unknown significance, I like to keep track” (white woman, age 64, group 1).

Interestingly, women with a previous genetic testing history also connected their adoption of the screen in relation to talking to their clinician. As one woman explained, “At the beginning I was confused because I said, ‘I think I did the BRCA [test] already.’ But she [my dr.] said, ‘No, this is different.’ [So] I said ‘Okay.’” (Hispanic/Latino woman, age 51, group 1). In comparison, a woman who opted not to do Gia because she had previous genetic testing shared, “I've read things that have said, if you've done genetic testing, it makes sense to do it again, because there's always new technologies and things changing [but] nobody's ever contacted me about that” (white woman, age 47, group 2).

Discussion

Implementing AI-assisted genetic screening tools into routine OB/GYN care can help streamline identification of women with hereditary risk, like HBOC, who may benefit from further genetic services to reduce risk and, ultimately, cancer incidence. AI approaches to screening may not only be more feasible from a healthcare system integration level, but as women in our study reiterated, this approach may extend access simply by providing women with a more convenient option. The feasibility factor or ability to access chatbots “anytime, anywhere” has been identified as important in previous research [18] and described as the “sweet spot” (i.e., “when the payoff [patients] received from the chatbot exceeded their energy investment” p. 325) [21]. Chatbots extend women’s access to screening, which may be particularly beneficial in rural regions similar to our catchment area where access to genomic medicine is notably limited [14]. However, it may also be important to provide nudges to remind them of the service (e.g., email or text reminders), as women in our study mentioned simply forgetting about it after receiving their initial invitation via email.

While AI may extend access to screening, women in our study identified key factors to consider when implementing chatbot genetic screening that inform whether women will adopt the screening opportunity. They described both systems-level factors (i.e., costs of services, insurance coverage) and individual-level factors (i.e., fear of results) that are known factors that collectively inhibit women’s uptake of genomic screening and testing [11, 14, 43]. In addition, while AI screening opportunities reduce barriers to access, women in our study shared how the innovation of this approach—AI—can function as either a barrier or facilitator in their uptake. Women stressed the importance of being able to trust the technology—trust that it is safe (i.e., their privacy is protected) and that it can also accurately assess their risk. It is noteworthy that their trust in the technology was informed by their trust in the source disseminating the chatbot. When they trust their local, personal medical provider/healthcare system, that trust translates to trusting the chatbot technology they disseminate, promoting women’s adoption of the screening opportunity.

Furthermore, women’s interaction with a clinician may be a key intervention strategy to implement with chatbots to reduce women’s inhibition and promote uptake by personalizing the screening opportunity. Women noted that when they had talked to a clinician about Gia (i.e., about its purpose and relevance to their health history), they were motivated to adopt the screening opportunity. When they hadn’t discussed it with a clinician, they were left with questions (e.g., Why should I do this?), which ultimately inhibited adoption. Our findings complement previous qualitative studies where patients reported wanting chatbots to act not as a substitution to clinical communication but rather as a supplement to communication with their clinician [18, 21]. Future research can move beyond identifying provider-patient communication as a factor that impacts genetic service by extending this to identify effective communication strategies that clinicians can use to facilitate screening adoption.

The findings from this study provide initial guidance as to the content of this clinical conversation or what clinicians should strategically communicate about to personalize this health service opportunity and promote adoption. Clinicians should address the facilitators and barriers that can inhibit women's adoption. One critical topic to discuss is women’s perceived personal risk. Women in our study highlighted the importance of their risk appraisal in whether they engaged in screening, a finding reflected in patients’ engagement in genomic medicine [44–48]. When women knew they had a cancer history in the family and, thus, appraised themselves as having risk, they were motivated to do the screening. However, women who did not adopt Gia shared how they believed they had no hereditary risk or did not understand how doing the screen was necessary for them personally. Interestingly, these women also at times acknowledged having limited discussions of health and illness within their families due in part to cultural norms, which may deter health information from being passed down generationally. Thus, their risk appraisals may not accurately reflect their risk and they may miss a critical opportunity to accurately assess personal risk.

Relatedly, this discussion of personal risk could extend into a second inter-related topic that motivates adoption: altruistic desires to reduce cancer risk among younger family members. The benefit that genetic information may have on other relatives, such as children particularly daughters, is one of the most common motivators for undergoing genetic testing (and a primary concern voiced during genetic counseling), as was reflected in our findings [7, 48–50]. In our study, the majority of women who completed Gia had at least one daughter (6/9 or 67%), compared to only 29% of those who did not complete Gia. Ultimately, while perceived personal risk is a motivating factor in women’s adoption of chatbot screening, they not only want to manage their own risk but also protect younger family members’ future health. When implementing chatbot screening into OB/GYN practice, clinicians can strategically integrate a conversation about personal risk (in relation to family history or lack thereof) while also including a discussion of their family members to promote altruistic motivation for screening adoption. Given chatbots are also now being implemented so that probands can more easily share their results with family members, this may also promote cascade testing among family members [7, 14].

Finally, when talking to their patients, clinicians should ask women about their genetic testing history. Although guidelines state that tests be updated periodically to account for changes in genetic testing and scientific knowledge [51, 52], women were confused about whether or not they needed to update their panel (or when they should). Some believed that their risk was known because they had previous testing, also believing there was no need for future screenings. Women in our study also validated the importance that their clinician address this, with one woman learning that not testing again was a misconception only after talking with her doctor and another acknowledging some confusion about whether it was needed given her clinician hadn’t addressed it with her.

Programs should be contextualized and tailored to the community served [27]. Our findings may be most transferable to similar populations, particularly rural-residing women as well as those who have some college education (all had completed at least some college or technical/professional training coursework and 12 had bachelor’s degrees or higher). UF Health is also part of a large state university with a medical school and, therefore, findings may be transferable to similar geographic areas (e.g., a university town with a medical school) as the context of the study described herein. The transferability of these findings (i.e., the extent to which findings may be applied to similar contexts of study) is an important criterion of trustworthiness in qualitative designs [53]. Still, while women represented diverse voices in race and ethnicity, the sample size was low in accord with targeted, in-depth qualitative study designs. Thus, future studies should explore which facilitators and barriers are reflected in a larger, more diverse population of women in OB/GYN care practice.

Conclusions

Chatbots can be an innovative way to bridge women in OB/GYN care with genetic screening. AI can extend the reach of genetic screening and, in turn, potentially contribute to reduced HBOC cancer incidence. To promote women’s adoption, particularly in rural areas, key enabling and inhibiting factors should be considered including addressing cost, trust in technology, and feasibility of use. Further, to promote adoption, clinicians should first talk with women during routine OB/GYN care to personalize the genetic screening opportunity and promote their adoption of screening via a chatbot by addressing their perceived risk, concern for younger family members’ risk, and clarifying the need for updating genetic testing.

Abbreviations

HBOC

Hereditary breast and ovarian cancer

Gia®

Genetic Information Assistant

OB/GYN

Obstetrics/gynecology

NCCN

National Comprehensive Cancer Network

CGCs

Certified genetic counselors

AI

Artificial intelligence

RAT

Rapid analysis template

Authors’ contributions

Conceptualization, CLF, EE, LRS, SMS; Methodology, CLF, ENW, EE, SMS, and DW; Formal Analysis, CLF and ENW; Writing & Revising– CLF, ENW, EE, LRS, SMS, and DW; Project Administration, CLF, ENW; Funding Acquisition, LRS.

Funding

Research reported in this publication was supported by the University of Florida Clinical and Translational Science Institute (CTSI), which is supported in part by the NIH National Center for Advancing Translational Sciences under award number UL1TR001427. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The APC was funded by UF CTSI.

National Center for Advancing Translational Sciences,NIH National Center for Advancing Translational Sciences,NIH National Center for Advancing Translational Sciences,NIH National Center for Advancing Translational Sciences,NIH National Center for Advancing Translational Sciences,NIH National Center for Advancing Translational Sciences,NIH National Center for Advancing Translational Sciences,Clinical and Translational Science Institute,University of Florida

Data availability

The data that support the findings of this study are included in part in the study presented. The full dataset was not approved by IRB for open access. Data requests would require IRB approval and should be directed to the lead author (eastonwollney@ufl.edu).

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of the University of Florida (IRB approval # 202301507, 9/28/2023). Informed consent was obtained from all subjects involved in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.