CHICAGO (GenomeWeb) – At a genomic informatics conference in February 2018, Optra Health Cofounder and Executive Chairman Abhi Gholap demonstrated an artificial intelligence-based clinical report interpretation and guidance system that he billed as a "digital genetic assistant." The technology allows consumers and healthcare professionals alike to verbally query genetic data via Amazon Alexa and Microsoft's Cortana.
Initially called OptraGuru when publicly launched early last year, the system has since been branded by San Jose, California-based Optra Health as GeneFax.
Meanwhile, at the American College of Medical Genetics and Genomics (ACMG) annual meeting in Seattle this month, Addison, Texas-based Metis Genetics announced that health systems would be able to license a white-labeled version of Genetics Maven, a proprietary, patient-facing platform to help consumers make sense of genetic test results.
Both are among a number of new tools on the market that employ AI, chatbots, telehealth services, educational videos, and other consumer-friendly technologies to address the well-documented shortage of genetic counselors and burgeoning demand for better patient understanding of molecular tests.
In general, the tools are intended to augment rather than replace genetic counselors and other clinicians by automating triage, shifting some of the administrative work to patients, and making a scarce human resource more efficient.
"We founded the company with the mission of really enhancing the value of genetic testing by improving the access to genetic counseling creatively," said Metis Founder and CEO Amanda Elms.
"How can we augment the human, which in our case is the genetic counselor or pharmacist?" said DNAFeed Founder and CEO Ahmed El-kalliny. "We want to take out all the repetitive components of their job and give them more time to focus on the human component of their job."
DNAFeed, a two-year-old San Diego company, offers a mix of telephone-based counseling, video, and online chat backed by both AI and human genetic counselors and pharmacists — the latter for pharmacogenomic applications.
"We are not a chatbot. There is always a licensed counselor or pharmacist on the other end of the line," El-kalliny said. "What they are seeing is recommendations that are coming in by the software."
For example, if the software sees a keyword indicating that the chat is about Alzheimer's disease, DNAFeed can show a recommended response or two to the pharmacist or counselor. The counselor would have to click on a response to send, and can edit the computer-generated text before sending. "It's never being sent without a human reviewing it," El-kalliny said.
For clinical-grade tests, which make up the majority of DNAFeed's work, templates help counselors complete documentation for physicians who order tests. "We are trying to automate a lot of components of that," El-kalliny said.
After DNAFeed schedules a call with a patient, any family history collected by the testing laboratory or ordering physician shows up in the system to guide the conversation, and the counselor will confirm that information. Following the session, the counselor writes up consult notes, which then get sent back to the ordering lab or clinician. At that point, DNAFeed can open up an asynchronous chat line for the physician to ask follow-up questions of the counselor or pharmacist.
"For a lot of providers, they are more likely to adopt pharmacogenomics if they feel that a pharmacist is going to be involved," El-kalliny said.
DNAFeed's system has several AI components, including a library of responses. For this, the company has a HIPAA business associate agreement with Google to use an API for scanning keywords in the chat, then matching the keywords with the DNAFeed library to propose responses.
The company has direct links to ClinVar and other databases to help counselors find citations and evidence. El-kalliny said that that feature is more often used for inquiries related to direct-to-consumer tests than for professionally ordered screening.
DNAFeed's own machine learning technology comes into play when a counselor chooses from a list of responses. "We feed that information to our software to learn" why the counselor may have made that particular selection to improve the algorithm, according to El-kalliny.
Another component is translation because DNAFeed offers the service in every language that Google Translate supports, which is based on Google AI.
Google Translate is notoriously inaccurate at times, so DNAFeed provides disclaimers whenever a translation is needed, and also keeps the source language in its documentation. "It actually works very well, much better than one would think, especially given the medical terminology and genetic terminology. It's actually gotten pretty good at knowing when you're talking about a gene [that doesn't need to be translated] and something else that does need to be translated," El-kalliny said.
DNAFeed also partners with a live-translation company for phone translation. The company is working on being able to automatically transcribe phone calls to provide counselors with near-real-time recommendations during phone consultations.
"We're trying to think of ways to make our experience more consistent, regardless of which counselor is doing the service," El-kalliny said.
He said that some counselors want to use the DNAFeed platform for their own clinics. El-kalliny said that the company is exploring the possibility of opening it up for telemedicine in that manner.
One company that already has incorporated telemedicine is Clear Genetics, a San Francisco-based chatbot startup that partners with GeneMatters for telehealth services.
Patients signing on to a health system's genetic telehealth portal will first go through Clear Genetics' Gia automated triage engine, a name that is short for Genetic Information Assistant. Complex cases can get handed off to counselors at a health system using Gia, or to live GeneMatters counselors.
GeneMatters has counselors available to serve patients in all 50 states. Some other automation companies offering telehealth-based counseling have similar services.
As of January, 22 states licensed genetic counselors, according to the National Society of Genetic Counselors, and four more had enacted legislation that will require licensure. A fifth, Georgia, passed a licensure bill this month.
Metis CEO Elms called state licensure a "moving target" that any company involved in genetic counseling must constantly stay on top of. "That's one of the headaches that Metis takes away," she said.
Metis wants to answer the question of how to do right by the patient in a way that works for labs, health systems, and payors as well. "In our opinion, really the only way to do that is to marry technology with the clinical expertise that counselors bring," according to Elms.
"We're not focused just on a chatbot or just on a platform, or just on educational videos, but combining those in a way that first and foremost makes sure patients get what they need in a responsible manner, but also takes into consideration what the test is, what we're trying to communicate, and where we're using the technology so that we can choose the right tool to increase the efficiency, increase accessibility and availability, but also do it in a way that lowers cost," she said.
When she first started the company four years ago, Elms thought that customers would mostly be labs or health systems that did not have genetic counselors. But about half of the company's clients do have counselors, including its flagship provider customer, Kaiser Permanente.
"Because we've created both that personal genetic counselor experience that's available as well as the automation, our customers can use us in almost any way that makes sense for their program," Elms said.
Kaiser Permanente did not need Metis for genetic counselors, but uses the company's automation services. Smaller health systems might not have the volume to require automation, but occasionally need to call in a genetic counselor, she said.
Clear Genetics also has pivoted since it came through Y Combinator, a prominent Silicon Valley business accelerator, particularly with the addition of telehealth and chatbots.
"Gia is a product of a lot of iteration," Snir said. "We're really dedicated to understanding how to solve this problem of scale. When we started the company, we looked at all kinds of different ways to reduce the load from genetic counselors."
Clear Genetics wanted an end-to-end offering to help clinicians identify patients who would be good candidates for genetic testing, educate those patients about the testing, obtain results, and connect those individuals to post-test consulting, according to Cofounder and CEO Moran Snir.
"We didn't think about utilizing chatbots in the beginning. We thought to do all kinds of different apps and we had all kinds of different ideas to communicate information," Snir said.
"What we learned is that there is so much content that's provided by genetic counselors that when we tried to communicate it to patients in different ways that we were exploring, it was overwhelming or boring or hard to understand and complex." Offering different options helped to personalize the experience and shorten waiting times for counseling, she said.
"The patients that really are high-risk and need the expertise and the experience of a genetic counselor can be connected to one as easily and quickly as possible," Snir said. Clear Genetics also can help clinicians manage those with pathogenic variants or are in some other high-risk category.
Optra Health's GeneFax is another example of adapting to a rapidly evolving marketplace. In a recent interview with GenomeWeb, CEO Gauri Naik said that "digital counseling" is just one aspect of that product.
"GeneFax is a full-fledged, AI-powered system, and we have a huge genetic knowledgebase that comes bundled along with this whole GeneFax infrastructure," Naik said.
Optra Health curates and builds its own knowledgebase from many different data sources, according to Medical Director Ashwin Kotwaliwale. The company now has a genetic dictionary containing about 450,000 terms, which are fully integrated with about 15 genetic databases, he said.
Ontologies make up a key part of the knowledgebase, covering 65 million different concepts. "A combination of all of these things in different layers of abstraction [comprise] what we call the GeneFax knowledgebase. We continually update that," Kotwaliwale said. The AI engine mines the knowledgebase to answer questions during the counseling process.
Users can connect with live counselors as needed.
On the back end of the enterprise version of GeneFax, counselors can configure dialog with patients, such as determining which types of patients should only hear about noninvasive testing, Kotwaliwale explained. "When patients come in and log into GeneFax, they will only see what was meant for them to see," he said.
"It has to be a meaningful dialogue," Kotwaliwale said.
"GeneFax leverages the technology that we have today to deliver genomic results in a most concise, precise [way] in the shortest time possible," Kotwaliwale continued. "That way, it becomes a productivity tool for genetic counselors to become more engaging."