Jeff Elton, Ph.D., is CEO of ConcertAI, an AI SaaS options firm offering analysis options and patient-centric options for all times sciences innovators and the world’s main suppliers. ConcertAI is targeted on accelerating and bettering the precision of retrospective and potential scientific research utilizing supplier EMRs, LISs, and PACSs methods because the supply for all research information. It’s a long-term companion companion with the American Society of Medical Oncology and its CancerLinQ program, US FDA, NCI Well being Fairness initiatives, and virtually 100 healthcare suppliers throughout the US.
Previous to ConcertAI, Jeff was Managing Director, Accenture Technique/Affected person Well being; International Chief Working Officer and SVP Technique at Novartis Institutes of BioMedical Analysis, Inc.; and companion at McKinsey & Firm. He’s additionally a founding board member and senior advisor to a number of early-stage firms. Jeff is at the moment a board member of the Massachusetts Biotechnology Council. He’s the co-author of the broadly cited ebook, Healthcare Disrupted (Wiley, 2016). Jeff has a Ph.D. and M.B.A. from The College of Chicago.
Because the founding CEO of ConcertAI, are you able to share your imaginative and prescient for the corporate at its inception? How has that imaginative and prescient advanced since 2018?
We began with the concept improved affected person outcomes come from deep and actionable insights. Gaining these insights requires information completeness, information scale, information representativeness and superior AI intelligence. So, we created a Knowledge-as-a-Service and AI Software program-as-a-Service firm. We focused AI that permits inferencing and prediction. This included predicting occasions to keep away from, equivalent to sufferers’ non-adherence to their remedy or discontinuation of care due to an absence of constructive response, which knowledgeable when scientific trials could be the subsequent choice.
Our imaginative and prescient has remained steadfast, and we proceed to anticipate extra out of our options. With the newest era of LLMs, agentic AI and different generative AI options, we will function at scale (and virtually in real-time—one thing we didn’t anticipate or anticipate in 2018). With companions like NVIDIA, we will advance our options to carry out higher than anticipated, acknowledge limitations and distinctive traits, and transfer on the tempo of your complete market’s improvements—the journey to date has been terribly productive and exhilarating.
We’ve opened up beforehand unimaginable efficiency in scientific trial automation options, automating the location of sufferers on evidence-based scientific pathways, superior workflows in radiological interpretation, and using digital twins as a decision-enhancing instrument for care and analysis.
Immediately, we serve virtually 50 biopharma innovators and a couple of,000 healthcare suppliers—so whereas not at fairly the dimensions of your complete market, we’re the broadest-reaching AI options for oncology within the trade.
What impressed you to give attention to oncology and hematology datasets particularly, and the way did you see ConcertAI making a distinction in these fields?
The US began the “Conflict on Most cancers” in 1971 with the Nationwide Most cancers Act. This catalyzed large-scale authorities funding, which generated insights into the mutations that drive cancers, new modalities for therapies, expanded Nationwide Most cancers Institute-designated therapy facilities, and extra. Underneath the Obama administration, funding elevated once more by $10 billion in digital stimulus going to the NIH and, in flip, to the NCI. Underneath Biden, the Most cancers Moonshot 2.0 program was launched in 2022, once more catalyzing a completely new era of analysis and seed funding funding for educational, neighborhood, and private-public partnerships.
I give this historical past as a result of few illnesses or areas of healthcare have the extent of information: genomic, transcriptomic, digital pathology, digital radiology, detailed digital medical information, and so forth., and the extent of analysis that contextualizes these information with validated insights via rigorous, multi-center, peer-review research. As additional proof, the American Society of Medical Oncology annual assembly is the biggest medical assembly on the earth, with the best variety of new publications, posters and abstracts of any scientific discussion board on any matter.
So, if you’ll be information and AI-centric, there are few higher areas to advance options with confidence and at scale than oncology. ConcertAI has the biggest assortment of research-grade information of anybody on the earth. It consists of a whole bunch of peer-reviewed publications enabled by that information, vital proof ensuing from these publications altering how sufferers are handled and assuring essentially the most constructive doable outcomes, and now AI SaaS applied sciences which are integral to the processes of care and analysis that convey the ability of that information and proof to bear in any respect factors and for all choices alongside a affected person’s care journey. What is basically essential right here is that we don’t do that unilaterally. It’s completed transparently with our healthcare supplier and biopharma innovator companions to engender the best confidence and use. So, we’re evolving towards real-time, superior, AI intelligence-enabled resolution augmentation.
ConcertAI has change into a number one participant in real-world proof (RWE) and AI know-how for healthcare. What have been a few of the early challenges you confronted in positioning the corporate as a frontrunner on this house?
It’s a must to be trusted and evolve in the direction of being the reference supply. That’s earned. The belief comes out of your supplier companions, believing that the information you might be accessing is in the most effective pursuits of their sufferers. Belief comes out of your tutorial and trade companions, who see the proof of and consider that your information is derived as an ideal reflection of the unique affected person information and that the ideas you advance are ‘true’ and reflective of present scientific and scientific observe. You even have to attain a scale that your information options characterize not solely your complete inhabitants but in addition produce conclusions which are confidently generalizable again to the complete inhabitants being handled with a specific medication. Know-how is analogous. Scientists and clinicians are inherently skeptical—as they need to be—and don’t belief black bins or algorithms they don’t perceive. So we wanted to ascertain belief there, too, via publications and being open about how our options work.
ConcertAI holds the world’s largest oncology and hematology dataset. What distinctive alternatives does this information create for reworking most cancers analysis and therapy?
I really like that query. We’re engaged on this daily! The alternatives to supply worth to suppliers, sufferers and innovators are virtually limitless. In early-phase trials, we’re evolving research simulation approaches with digital twins that may change the packages we take into scientific trials. Our information and AI optimizations will decrease the time required to go from finalized protocol to finalized submission to regulators by 30 to 40%—which means new medicines get to sufferers sooner. Our resolution augmentation AI options will advocate pathways for therapy which are evidence-based and particularly tailor-made to these pathways, monitor responses in step with the anticipated response, and search for doubtlessly useful scientific trials when response or profit is beneath expectations. Our imaging scientific interpretation options function on the stage of operational processes, scientific interpretation, and longer-term view of recent interpretations or new interventions that needs to be thought-about primarily based on insights and proof sooner or later. Not is an motion “as soon as and completed” however quite it turns into “as soon as, after which repeatedly” such that useful reassessments and future choices are an ongoing course of! What’s completely different right here is that the view is your complete affected person journey—this can be a horizontal view versus a collection of slim, deep, vertical views that need to be stitched collectively. That is an innovation enabled by AI and a profound course of change that gives new methods of working to the knowledgeable people concerned.
Are you able to clarify how ConcertAI’s Digital Trial Resolution works to match most cancers sufferers with life-saving scientific trials? What affect have you ever seen to date when it comes to affected person outcomes?
Medical trials are very advanced and require hours of effort by a variety of extremely knowledgeable people. For many organizations, scientific trials are supplied as a duty and dedication to sufferers the place the present customary of care could not characterize a viable various. Trials have probably not been very obtainable to sufferers in neighborhood therapy facilities, the place 80% of sufferers obtain their care. But, these are the sufferers who will in the end be receiving newly accredited medicines. This creates a double dilemma: the vast majority of sufferers who want entry to trials are restricted, and those that are reflective of the final word customary of care inhabitants aren’t within the trial dataset. We set a path to resolve these issues.
The outcomes have been nice—so constructive that we’re going to be increasing our variety of research underway by 10x in 2025. We printed this for the final American Society of Medical Oncology conferences and in different areas. Our method is how we predict AI needs to be carried out—as an augmentation of knowledgeable people the place there are giant capability and expertise constraints and the place lives are at stake. We’ve developed a set of orchestrated and tuned giant language fashions that entry affected person information, synthesize traits, and match sufferers to doubtlessly useful trials, doing precisely what the knowledgeable people would do—with a completely documented method to creating suggestions and assessments. Within the websites the place our applied sciences are deployed, we carry out on the stage of essentially the most knowledgeable people and accrue sufferers at 5x or extra relative to websites the place our applied sciences aren’t deployed—the analysis groups and biopharma innovators are each happy, and sufferers profit most.
How does ConcertAI’s AI-driven method to trial design and affected person recruitment tackle a few of the present limitations in scientific analysis, equivalent to affected person variety and trial effectivity?
I’m pleased with my staff—they informed me three to 4 years in the past that reaching variety is an obligation and the proper factor to do scientifically. Additionally they maintained that it’s exhausting to do whether it is guide however requires zero incremental effort if automated. So, we determined then that each dataset and AI SaaS resolution would combine variety and social determinants of well being traits as our customary method. It’s not an choice. It’s simply what we do. Subsequently, our CARAai™ supported scientific trial design and optimization options can assess what ethnic, racial or financial subpopulations could also be most adversely impacted by a illness, combine these issues into the trial design, be certain that these populations aren’t unwittingly excluded, and outline the scientific websites almost certainly to guarantee participation and representativeness. That is the place AI might be “AI for Good” and the place know-how doesn’t introduce a bias however assures that biases don’t enter the method, the final word design, or the operational processes across the scientific trial.
What function does ConcertAI play in decreasing the burden on healthcare suppliers and optimizing web site choice in scientific trials?
We combine the work burden into all points of our scientific trial options. First, there’s a burden on the affected person. This may be the place the positioning is positioned, the variety of visits required for a research versus the usual of care, or the scientific depth of a research versus the usual of care, as within the case of further biopsies. These items can decide whether or not the affected person—or the affected person in session with their supplier—can afford to take part or tolerate and full participation.
There may be additionally a burden on the supplier. If we will automate the identification of sufferers for scientific trial eligibility, reduce false positives that create work, and supply what we name “AI leverage” to the work of the Medical Analysis Related, Examine Nurses, and Physicians, then the burden is lowered. The identical is true of our AI Automation Resolution, which permits the analysis staff to keep away from doing guide information entry—sometimes 2 to 4 hours on the finish of the day, and infrequently accomplished at residence. Early on we appeared on the information within the EMR—digital—being manually entered right into a portal for the sponsor’s EDC. So digital information is being learn after which rekeyed to change into digital information once more! Right here, too, we’re utilizing our multi-tuned giant language fashions—this was an actual focus of the NVIDIA partnership from the start. We’re at 55% full automation at present, with a really quick path to over 80% within the coming few months. As these parts come collectively, we’ll get the employees time right down to 10% of legacy necessities and make these research extra accessible to extra sufferers.
Precision medication is a key space the place AI is making vital strides. How does ConcertAI’s know-how contribute to extra exact and personalised most cancers remedies?
We’ve not mentioned this an excessive amount of since final 12 months. In December 2023, we assumed duty for the American Society of Medical Oncology’s (ASCO) CancerLinQ program. It’s the world’s largest clever well being community, comprising tutorial facilities, regional hospital methods and neighborhood suppliers. A key a part of this community is implementing the ASCO Licensed® high quality and scientific pathway options. Since CancerLinQ is a ConcertAI initiative, we have now been rising the community, automating precision oncology pathways, creating new digital twin approaches for augmenting therapy choice for the suppliers, figuring out and messaging crucial diagnostic assessments that might inform therapy choices, and doing the identical for newly accredited medicines that characterize one other or higher therapy various. All of that is underpinned by our CARAai™ structure, once more a set of imaginative and prescient LLM and tuned oncology LLMs completed in collaboration with NVIDIA. It’s wonderful to see the progress being made, and we’re enthusiastic about what we’ll be publishing and presenting at subsequent 12 months’s ASCO 2025.
How do you see AI imaging options benefiting fields like oncology and radiology, particularly as these fields face clinician shortages?
Nice query! It’s true that each the variety of new oncologists and radiologists coming into the sphere is lower than the quantity retiring. Nonetheless, affected person demand is ever-increasing. So, it’s the preferrred space for offering AI SaaS options that assist doctor and allied care professionals in each workflow optimization and scientific resolution augmentation. Radiologists and oncologists will each cite the significance of those new clever options coming into their fields particularly. Imaging is a superb space for AI, and its efficiency is outstanding. Non-inferiority research replicate that AI fashions might be near or similar to knowledgeable people in slim areas. Orchestrated workflows can convey this all collectively. The identical is true in oncology, the place we’re bringing collectively molecular check outcomes with immune response information, predictive algorithms for resistance and different parts that may all inform the therapy resolution and allow response monitoring. I’ve been within the area for years and on completely different sides of recent improvements—what we will do now’s effectively past something we have been ever capable of do earlier than, and the tempo of change is wonderful.
As an skilled chief in healthcare know-how, what recommendation would you provide to new firms seeking to make a significant affect in healthcare via AI?
You possibly can’t be an AI firm with out entry to information at scale. Knowledge is the substrate for constructing coaching and monitoring fashions. Additionally, constructing AI options is a staff sport. You want area data at an distinctive depth matched with a brand new era of AI mannequin improvement capabilities that acknowledges the behaviors of various courses of AI options and may convey them to bear in opposition to slim goals, particularly tuned for human or above efficiency. Then, these approaches might be orchestrated in numerous methods to characterize a brand new system for working—that’s the place the modifications happen, and the worth will get delivered. Follow “AI Humility” as the whole lot is wonderful and reveals issues we couldn’t do even six months earlier than. But, ‘wonderful’ is just not essentially a product or a brand new means of working—it’s simply that, know-how doing one thing new. It’s the duty of the AI firm to make it a brand new means of working and a brand new method for delivering an astonishing stage of worth that was by no means accessible earlier than. Lastly, assume you want to show belief in enterprise practices, AI fashions, and resolution transparency. We’re nonetheless early in our societal journey, and we’re those who need to earn the belief to convey concerning the modifications we’re able to delivering.
Thanks for the good interview, readers who want to study extra ought to go to ConcertAI.
