Skip to main content
Premium Trial:

Request an Annual Quote

Chatbots Could Help Human Genetic Counselors Focus on Patient Concerns

SALT LAKE CITY – Artificial intelligence approaches to genetic counseling like chatbots won't replace human providers but could instead help tailor counseling sessions to patients' concerns, according to a panel discussion at the National Society of Genetic Counselors annual meeting last week.

Companies like Clear Genetics, which Invitae is about to acquire, have developed chatbots for genetic counseling. Geisinger Health, for instance, is using Clear Genetics' Genetic Information Assistant, or Gia, to handle some aspects of genetic counseling that arise during its MyCode Community Health Initiative. Meanwhile, GeneMatters, a telehealth genetic counseling service provider, has also partnered with Clear Genetics to offer a mix of AI- and human-powered genetic counseling. Other companies, like DNAFeed and Metis Genetics, offer both AI and other tools to aid genetic counseling.

Panelists at the NGSC meeting delved into how chatbots and similar tools could be used in genetic counseling. They argued that, despite some fears, these types of tools could augment genetic counselors' practices by, for instance, triaging patients or providing initial educational information. The panelists cautioned, though, that chatbots or other AI counseling needs to be tested responsibly, and added that these tools would complement traditional genetic counseling.

"Our field relies on and requires human capital. ... It requires our ability to connect, our ability to reflect and to respond to that patient in front of us," said Tara Schmidlen, a clinical investigator and genetic counselor at Geisinger. "Our chatbots are great, but they're not that great. They can't do that."

Artificial intelligence is increasingly making its way into medical fields, and issues around its implementation and use aren't specific to genetic counselors. Clear Genetics' Shivani Nazareth, the session moderator, noted that radiologists, for instance, are currently also grappling with the use and implementation of AI.

"Genetics is not going to be exempt from this trend" [of increasing use of AI], said Nazareth, who is the head of clinical development at Clear Genetics.

The question, though, is how to best employ it. She and other panelists said AI approaches like chatbots could help lessen genetic counselors' load or free them up to focus on more complex aspects of counseling. For example, technology like this could be used to help genetic counselors educate patients — a part of counseling that can be repetitive — or triage them."We could use that technology to get a sense of where your patient is, of a starting point ahead of the visit," Schmidlen said.

Cathy Wicklund, director of the graduate program in genetic counseling at Northwestern University, noted that even now, genetic counselors don't always do a good job triaging patients. Much of the process is arbitrary, she said, or based on whether or not patients had a positive or a negative result. For instance, everyone who had a positive result would receive a phone call or face-to-face genetic counseling session and those with a negative result  would not.

But she noted that not everyone processes information the same way. "It might be routine for us. It is not routine for that person sitting across from us," she said.

AI and chatbots, several panelists noted, could enable genetic counselors to better tailor their sessions based on their patients' needs, whether it's going over complex cases or answering additional questions a more routine patient might have.

These new approaches, though, need to be studied. Kaylene Ready, senior director of product development at GeneMatters, noted that quizzes are often one of the go-to approaches to assess knowledge following genetic counseling, but she added that she hopes the field can move beyond that type of measure. "I think what we're more interested in is: can they make an educated, informed decision consistent with their values?" she said.

Schmidlen added that the type of testing needed depends in part on what the goal of using these AI approaches it. She and others noted that there are many possible outcomes that could be analyzed, ranging from cascade testing uptake to decisional regret. One approach she said she favors is non-inferiority testing to make sure this type of approach is, at least, no worse that typical genetic counseling.

The panelists also noted that what works in one population might not work in another, necessitating tweaks to the program.

Another question is whether patients would engage with chatbots or similar tools, or whether it would be popular just among younger, likely tech-savvy patients. However, Schmidlen said that the patient population at Geisinger skews older, with an average age of about 55, and patients there were willing to interact with the chatbot. While their analyses have indicated that patients who were 52 rather than 60 years old were more likely to consent to using chatbots,  patients across the age spectrum might embrace their use.

But as Ready added, genetic counselors could also just ask patients what they prefer.

"We should actually just ask our patients these questions," she said.