Skip to main content
Premium Trial:

Request an Annual Quote

Geisinger Deploys 'Gia' Chatbot to Help Genetic Counselors Manage MyCode Participants


NEW YORK (GenomeWeb) – Geisinger Health last week began using a chatbot to communicate with patients participating in its rapidly growing precision medicine research program, known as the MyCode Community Health Initiative.

Geisinger, a health system that serves largely rural communities in Pennsylvania and New Jersey, has been working with San Francisco-based Clear Genetics for 18 months to develop the Genetic Information Assistant, or Gia. After market surveys and pilot testing, Geisinger last week introduced the chatbot within the MyCode research program as a technological aid to its genetic counselors.

To date, more than 200,000 individuals have consented to partake in MyCode, in which Geisigner and partner Regeneron Pharmaceuticals are sequencing the exomes of participants for disease research and returning results to subjects when testing detects genetic mutations linked to conditions that they could take some preventive or mitigating action against.

Regeneron has sequenced the exomes of approximately 92,000 individuals, and around 2 percent will find out they have mutations in one of the genes Geisinger is testing for. Currently, Geisinger is reporting back likely pathogenic or pathogenic variants detected in 76 genes and associated with 27 conditions. According to the most recent report from the MyCode genomic screening and counseling program, around 870 participants received an actionable result.

Although MyCode currently involves nearly 10 percent of the population that the integrated health system serves, Geisinger plans to continue to expand the project and even help healthcare systems around the country launch similar efforts. Critical to these ambitions are Geisinger's team of more than 25 genetic counselors, who already have a variety of responsibilities, including interpreting variants, consenting patients for MyCode, delivering results to participants, speaking to at-risk relatives, and working on variant interpretation — and who will need help.

There are more than 4,000 genetic counselors in the US and Canada, and the US Bureau of Labor Statistics projects that jobs in this sector will grow by 29 percent between 2016 and 2026. During the same period, the genetic testing industry will also experience rapid expansion. According to one estimate, there are currently more than 74,000 commercially available genetic tests in the US and 14 new tests entering the market daily, compared to 60,000 commercially available tests two years ago and 10 new tests launched on the market daily.

Anticipating continued demand for genetic testing in the coming years, healthcare systems like Geisinger and others are testing out chatbots and telegenetics services as a way to enhance counselors' ability to support patients as genetics becomes further integrated into medicine.

With the deployment of Gia last week, Geisinger is evaluating how well its patients react to the chatbot in limited settings. The bot is currently being used to check in with MyCode participants one month after they've received test results to gauge what actions they've taken. Additionally, patients with positive results can also use the bots to relay pertinent educational information to at-risk relatives.

According to Amy Curry Sturm, co-director of the MyCode genomic screening and counseling program, the idea is not to replace human genetic counselors with chatbots, but to use these tools to free them up to lend their expertise to patients who most need it. "Don't fear the chatbots," Sturm said she tells healthcare professionals at medical conferences. "I'm a genetic counselor. Do you think I want to be replaced by an artificial intelligence chatbot?"

When MyCode participants receive likely pathogenic or pathogenic results, Geisinger’s genetic counselors still call them up to relay that information, and discuss the implications for them and their families. During that time, counselors also explain how Gia can be used in the follow-up setting and to educate at-risk relatives, and ask if they want to interact with a bot in the future. So far, 63 percent have agreed to do so.  

Before introducing Gia within the MyCode program, Geisinger held six focus groups across communities in Pennsylvania and in parts of New Jersey. In these focus groups, 62 people provided feedback on Gia when it was used as a one-month check-in tool to find out if patients have scheduled recommended appointments with doctors and genetic counselors, and as part of the cascade testing program.

Geisinger has also evaluated Gia for consenting patients to MyCode, but hasn't deployed the bot for this purpose yet. "We will await the analysis of our planned data collection on how individuals respond to Gia for their check-in and for cascade testing communications, and then decide about other use cases," Sturm said.

Clear Genetics' cascade testing chatbot is also being used by Cleveland Clinic, the University of Virginia, Johns Hopkins University, and Stanford University. "We're learning at these places what's the best way for genetic counselors to introduce the family sharing tool among relatives who are spread out in terms of where they live," Clear Genetics CEO Moran Snir said.

Genetic risk information has implications not just for the individual with the genetic mutation, but for entire families. However, people may not feel prepared to discuss this complex and sensitive information with their relatives. By including Gia within the family sharing tool, MyCode participants can help their relatives who may harbor the same mutations learn about their risks and testing options.

"The proband, the patient that gets the result from MyCode, can use the family sharing tool to launch the cascade chatbot," Sturm explained. If there are relatives that the MyCode participant wants to relay this information to, the family sharing tool includes an option to send a link to the chatbot via text, email, or Facebook messenger.

When family members open this communication, they'll learn that their relative participated in Geisinger's MyCode program and received information that could impact their health, and that a chatbot called Gia could provide more assistance. "It's really automating that pretest genetic counseling session," Sturm said.

Gia Chatbot for Cascade Testing

With permission from Geisinger Health

In the pretest setting, the information shared with family members tends to be similar accross disease settings and genes, such as the risks and benefits of learning this information, and what it means to have a positive or negative result. Gia also relays information about what gene isn't working properly in the tested relative, what condition is associated with that mutated gene, specific information about the particular mutation, and what lab does the testing.

"Our genetic counseling team has scripted out the entire chatbot," Sturm said. "It's very authoritative information coming from the minds of genetic counselors, put into a chatbot, and delivered to an at-risk relative."

Research on cascade testing suggests that probands often feel ill-equipped to communicate this type of information their relatives. "They're not medical experts," Sturm said. "They want more assistance from the clinical service, [because] they're just getting this diagnosis themselves, and they're trying to manage their own health."

Gia also includes a frequently asked questions feature where the user can type in a question and the chatbot uses natural language processing to analyze the question and come up with an answer. The bot has been specifically trained to answer questions related to genetics and genomics, but can also tackle questions related to test costs and insurance coverage, which commonly came up during pilot testing. "People have similar concerns across different areas of care," Snir said.

Clear Genetics' partners specify answers to common questions that Gia can draw from. Each time Gia answers a question, the bot is designed to learn from the interaction and improve. When the bot doesn't know an answer, the partner institution, such as Geisinger, is notified and can respond to the patient.

When individuals interact with Gia, they're informed upfront they're interacting with a computer, information that other industries may not volunteer to customers. "We didn't feel comfortable giving people the impression that they're interacting with a human when in fact they're not," Sturm said, adding that during any chat session there is a link via which users can contact Geisinger's genetic counseling team. "We wanted the users of chatbots to know that they had access to real intelligence genetic counselors right there in the tool."

Within MyCode, participants learn genetic risk information that they can take some preventive action against. For example, a woman with a BRCA1 mutation can get more frequent screenings for breast and ovarian cancer and may even decide to surgically remove their breasts and ovaries. This is "somewhat high-stakes information" about disease risk, Sturm said, and noted that the geneticists, counselors, and bioethicists at Geisinger didn't feel right that people not be told they were speaking to a bot.

"Gia's role isn't to replace but to assist genetic counselors," Snir said. "So, we [at Clear Genetics] also didn't feel right that she poses as a human."

Clear Genetics is also evaluating Gia with Cleveland Clinic as a tool for identifying patients in the primary care setting who are at genetic risk for hereditary diseases and for communicating negative genetic test results to patients. For the time being, Geisinger has no plans to use Gia to communicate test results to patients. But the healthcare system is working on a six-month check-in tool and is codeveloping a patient and physician portal with Clear Genetics.

It was previously announced that Sturm would serve as a scientific advisor and David Ledbetter, executive VP and chief scientific officer at Geisinger, would be chief scientific officer at Clear Genetics. They both stepped down from these positions, Sturm said, recognizing the need to do away with conflicts of interest as Geisinger would need to independently evaluate the implementation and utility of these chatbots for its patients.

"Neither one of us have any financial or official relationship with Clear Genetics," she said, clarifying that Geisinger and Clear Genetics have codevelopment agreements in place around Gia and may ink others in the future.