AI reshapes clinical trials and oncology care with data-driven precision, ethical design, and unmatched operational scale at the core.
Hi Jeff, welcome to AI Tech. Could you tell us about the journey that led you to be a board member of the Massachusetts Biotechnology Council, co-authoring a book, and now as Vice Chairman at ConcertAI?
I believe individuals are both shaped by their experiences and do some shaping of their own. In that spirit, some aspects of my journey were predetermined, others driven by deep passion and responsibility. Growing up with a biomedical researcher mother who helped bring new therapeutics to market, pursuing questions that mattered, and operating at the highest levels of scientific rigor were quite literally at my dinner table. The Massachusetts Biotechnology Council has played a crucial role in advancing biopharma innovation, and as vice-chair of the Board and Executive Committee member, I engage in key initiatives ensuring the vitality of biomedical innovation—especially as AI becomes central to its strategy.
My work with Clay Christensen on healthcare becoming part of his Disruptive Technology framework stayed with me, allowing me to assess what truly delivers value and what is ripe for disruption. AI has definitely arrived at that position. It is not just the technology, but the combination of Generative and Agentic AI to de-fragment and eliminate low-value adding activities and advance higher-value ones with greater efficiency. It was only natural, then, when Anne O’Riordan and I set out to publish a book that was at once a framework for biomedical innovation and a set of recommendations for how to bring this together for the greatest benefit of patients in all geographies that “disruption” would play a key role—hence, Healthcare Disrupted. We saw the early days of data moving into the cloud and the ability to bring advanced solutions and AI to fruition.
ConcertAI emerged from that vision, sitting between two ecosystems and bringing together 2,000 healthcare providers and nearly 50 life sciences companies to accelerate insights and treatments for the most challenging diseases. Building ConcertAI required defragmenting healthcare by aligning providers, molecular diagnostic, AI technology, and biopharma organizations. Just like my dinner table, I have healthcare providers and biomedical researchers as my customers and partners. ConcertAI could not have occurred any earlier—we needed advanced clinical cloud infrastructure, next-gen AI solutions, advanced GPU processing, and an unfreezing of models of innovation for both research sites and biopharma. Today, our work continues to push the boundaries of what’s possible in precision medicine and patient care.
With your intensive experience in this field, what was your original vision for the company when you started, and how has the vision evolved as ConcertAI has had remarkable growth since its inception?
We were born an AI company delivering clinical multi-modal data and some of the first AI Software as a Service (SaaS) solutions. When we started, there was a reduction in cancer deaths by 25% over 40 years of research and new medicines being made available. Though that was huge, it was not enough as real rates of new patients with cancer and deaths were still rising. Our mission was to advance solutions for the hardest-to-treat cancers, reduce cancer deaths, and improve the outcomes and quality of life for patients at a scale and pace 10x faster than the prior 40 years.
Early on, the focus was on data. To achieve high-trust, representative, and generalizable AI solutions, we needed large datasets. Our work with the American Society of Clinical Oncology (ASCO) and CancerLinQ was part of that, later expanding to include leading networks like ONCare and Exigent. In addition to data scale, we required source diversity, so we partnered with academic centers, regional health systems, and community providers. Then, we needed multiple modalities of data, including clinical records, molecular and imaging data, and medical claims. Finally, we needed to be able to develop concepts that came from the “unstructured” portion of clinical records. Between 2018 and 2022, we scaled rapidly and advanced some of the highest quality and highest trust data available for research and regulatory uses.
Today, we operate at the largest scale in oncology and hematology research and solutions. Our CARAai™ platform leverages small language models (SLMs), large language models (LLMs), and Agentic AI to process data with unprecedented speed, latency, and depth only imagined before. Rather than shifting our vision, we’ve accelerated every aspect of the company. Over the past 18 months, we’ve undergone one of the most significant re-platforming efforts in the industry. Every model, technical environment, workflow, and job definition has been impacted. The result is an exciting new era of solutions with transformative potential for patients, care providers, and their importance to biomedical innovation.
With the biggest multi-modal oncology dataset and smart oncology network, how are you balancing data ethics and privacy with developing AI capabilities in oncology?
Great question. We resolve the question of balance by noting that there is no balance. The ethical use of data is the foundation of everything. If you can’t explain the benefit of a data solution or an AI SaaS solution to a patient and their care providers, then it likely shouldn’t be built.
Our solutions fit within a narrow set of categories:
- Data solutions supporting research published in peer-reviewed journals and presented to regulators in support of a decision
- AI SaaS solutions that optimize access and performance of clinical trials
- AI SaaS solutions that assure that all patients are aware of newly approved therapeutics that might represent a better alternative than the current standard of care
- AI SaaS solutions that aid diagnoses and treatment decisions
What challenges did you encounter in attaining the goal of having ConcertAI’s automation capture 65% of the clinical trial data, and how does it change the dynamics of clinical trials and patient care in the future?
Automating clinical trial activities was challenging due to the complexity of trial structures, and the different concepts involved that are not part of the standard of care. Electronic Medical Records (EMRs) are different, with even the same EMR being implemented differently across providers. This means that some elements can be easily derived from EMRs, and others less so.
Most clinical trial solutions only automate 15-20% of required data—we approach this as both a workflow and a data challenge. Our goal was to reduce the research burden on providers, increase patient access to trials, lower the number of sites required for sponsors, and bring these aspects together in an approach that was 30-50% faster while reducing costs.
In just two years, we’ve advanced to our third generation of technology and are launching a new set of solutions based on our CARAai™ platform. This progress moves us closer to our goal of automating 90% of clinical trial data and better matching patients to trials. Earlier this month, a major oncology clinical trial network adopted our technology, reinforcing our commitment to expanding access and efficiency across research sites and sponsors.
Tell us how ConcertAI’s multi-tuned large language models are contributing to improving the accuracy and recall of AI in oncology.
In our experience with complex oncology concepts, we found that general large language models (LLMs) are 60-70% accurate to the question or calculation requested. This is acceptable for “search” or simple, general-purpose queries, but is too low to use for biomedical research or standard-of-care solutions. There is also the tendency for LLMs to drift and hallucinate.
To overcome these roadblocks and enhance precision, we have integrated small language model (SLM) pre-processing, oncology tuning of multiple LLMs for specific use cases, anchoring how LLMs initiate. These models are only working in our environments and are only exposed to the data within our network, which allows us to achieve exceptional accuracy, recall, and reproducibility across a range of defined use cases.
What do you look for in a strategic partner, and how do these partnerships help you amplify your mission of transforming healthcare?
Brilliant question. To put it simply, partnerships need to “amplify”—allowing you to boost the right “signal” amid the noise. Advancing healthcare and life sciences solutions is, in part, a defragmentation challenge, and the right partners—those operating with similar intent and models—can significantly unify efforts. Whether it’s healthcare providers, molecular diagnostic companies, or AI innovators, strategic collaborations help consolidate our efforts and drive immense progress. AI evolves rapidly, but new advancements don’t replace older technologies—they enhance them. The key is integrating improvements while maintaining a stable platform. Our partnerships with NVIDIA and SymphonyAI have proven this and have been fantastic accelerators.
What do you believe are the key factors that made you achieve a $1.9 billion valuation and the fastest time to double unicorn status?
Several factors contributed to this achievement. First and most importantly, we started with a clear and focused purpose: advancing medical innovations with the potential to improve outcomes and save the lives of cancer patients with the highest needs. This vision required a strong foundation of data, talent, partnerships, and technologies.
Second, we knew we needed to scale quickly. Cancer comprises hundreds of diseases that may appear and progress differently in each patient, so scaling ensures representativeness and generalizability, which in turn allows confidence in our analyses and conclusions.
Third, since day one, we have worked with leading healthcare providers and biopharma innovators. They knew our strategy and, in essence, endorsed it by agreeing to a path that was years into the future.
Lastly, we were born an “AI-first” company. Our approach integrated the best of data engineering, data science, and AI SaaS from the beginning. By validating our solutions through the research side of our operations and partnerships and moving them into standard-of-care workflows, we created a new model that can move at the pace of AI. These factors—clarity of purpose, speed to scale, strategic partnerships, and our AI-first approach—were key drivers in reaching a billion-dollar valuation and achieving the fastest time to double unicorn status.
What were some of the key initiatives that enabled you to scale efficiently while keeping annual recurring revenue high, where you had an impressive ARR of 77% in operational excellence?
While we are a relatively young company, our teams refer to our current operating model as Gen2, with plans to evolve into Gen3.
Gen1 consisted of digital workflows with staged transformations and expert human checks throughout. Gen2 introduces continuous workflows and processing, while Gen3 will bring together large language models, small language models and agentic AI. All of this is powered by the CARAai™ platform. These technologies allow us to do what would otherwise take hundreds of humans, and a lot of time.
This framework enables us to achieve operational excellence and deliver Data-as-a-Service and AI SaaS products with high recency, low latency, and exceptional intelligence to increase and accelerate decisions with precision and confidence.
How do you envision the AI ecosystem in healthcare evolving, and what are the next advances on the horizon for AI in medicine?
First, let’s focus on healthcare providers—we have more demand for healthcare services than the trained individuals needed to provide them. Any solution that can augment the productivity of clinical personnel will have rapid adoption. In the next few years, all health data will be prepared by AI and be readable, contextualized, and made accessible to a variety of solutions. New AI-driven tools will emerge to assist clinicians during patient appointments, creating a context for the decision that will be made, presenting relevant clinical information, providing the evidence and guidelines for decisions, and automatically initiating the implied actions, all in real-time. Equally important, AI agents can eliminate one of the most burdensome aspects of physician work—prior authorization.
For life science companies, their work will be comparably transformed. Translational medicine will evolve into a data science-centric set of workflows and assessments based on the use of multi-modal digital twins and trial simulations. As a result of these solutions, we anticipate a 20-30% improvement in early-phase trial success. With heavy documentation and administrative workflows, clinical trials are desperate for generative and agentic AI solutions. With AI advancements, site activities can be heavily automated, with agents supporting both site personnel and sponsors, and data can be automatically collected, processed, quality assessed, and made continuously available for interim analyses and decisions. Total end-to-end workflows can become 30-50% faster and more efficient.
Finally, the future of commercial enterprise will be patient-centric, ensuring eligible patients are aware of the best treatment options. Agentic AI will play a key role in sales activities, automated medical science liaisons, and patient assistance programs, to name a few. These changes, which are already under way, are “AI for Good” where the patients’ interests are always the North Star, at the forefront of every innovation.
What kind of advice would you give to entrepreneurs who are looking to scale a healthcare AI company while maintaining a strong focus on both innovation and patient outcomes?
All healthcare problems are inherently difficult; entrepreneurs should focus on solving the problems that truly matter. Asking themselves if anyone will care, and specifically who will care, is key. Since not all solutions have an obvious buyer, doing this will identify potential customers early on.
One of the key issues with the 1,000 FDA-approved AI models in medical imaging is that they don’t change workflows and generally are not coded or reimbursable events. For any new healthcare AI company, defining its “moat” is essential, as it will get increasingly easier to provide comparable functionality with superior performance. At ConcertAI, our competitive differentiator is having the largest and most diverse multi-modal oncology dataset worldwide.
Finally, healthcare AI entrepreneurs must follow the principle of “build it versus talk about it,” focusing on development over endless PowerPoints. Build, test, and innovate, rather than talking about it.
A quote or advice from the author: Entrepreneurs should focus on solving the problems that truly matter, asking themselves if anyone will care, and specifically who will care.
Jeff Elton
Vice Chairman of ConcertAI
Jeff Elton, Ph.D., is the Vice Chairman of ConcertAI, an AI SaaS solutions company providing research and patient-centric solutions for life sciences innovators and the world’s leading providers. ConcertAI is focused on accelerating and improving the precision of retrospective and prospective clinical studies using provider EMRs, LISs, and PACSs systems as the source for all study data. It is a long-term partner with the American Society of Clinical Oncology and its CancerLinQ program, US FDA, NCI Health Equity initiatives, and almost 100 healthcare providers across the US.
Prior to ConcertAI, Jeff was Managing Director, Accenture Strategy/Patient Health; Global Chief Operating Officer and SVP Strategy at Novartis Institutes of BioMedical Research, Inc.; and partner at McKinsey & Company. He is also a founding board member and senior advisor to several early-stage companies. Jeff is currently a board member of the Massachusetts Biotechnology Council. He is the co-author of the widely cited book, Healthcare Disrupted (Wiley, 2016). Jeff has a Ph.D. and M.B.A. from The University of Chicago.