RamaOnHealthcare June 11, 2024
Breaking through Cancer Treatment Faster
Today, RamaOnHealthcare talks with Dr. C.K. Wang, an Oncologist and Chief Medical Officer at COTA Healthcare. There he leads the medical team to meet the interest and demand for real-world cancer care data. Previously, he served as Manager and Oncology lead at IBM’s Watson Health. There, he supported their Global AI-driven, oncology efforts. He was the Medical Director at USMD Health System, a large cancer center in Dallas-Fort Worth.
COTA recently announced a partnership with Sanofi, to use real-world-data and AI to dramatically improve the speed of cancer clinical trials, including for Multiple Myeloma. About 35,000 new cases of Multiple Myeloma will be diagnosed this year. Reducing the time it takes to perform a trial and improving the quality of the trials gives patients and their families the ability to be treated with breakthrough treatments faster and with more confidence. About 97% of cancer clinical trials fail. Most of them are slow, expensive, and complex from their earliest stages. It’s possible to accelerate trials and improve their design with high quality data, but today this is the exception – not the rule. COTA is advancing real-world data use in cancer treatment research and creating data sets outside of clinical trials.
Dr. C.K. Wang, Chief Medical Officer at COTA Healthcare
RamaOnHealthcare (ROH): Can you explain what COTA does, who you reach, and an example of a real-world use case?
Dr. C. K. Wang (CKW): COTA uses high-quality real-world data (RWD) and artificial intelligence (AI) to help people with cancer benefit from the right treatments, faster. We work with life science companies, oncologists, payers, regulators and others to accelerate clinical research and drug discovery in oncology.
We were founded by oncologists who wanted to bring more clarity to cancer with RWD. Today, their expertise remains essential to our work. Clinical experts curate our data set with high quality, multimodal information from oncology practices across the country, building a representative view of cancer care in the US. COTA oncologists and clinicians are also training our AI models to ensure clinical accuracy while speeding up the time needed to make data usable for research. This input from clinicians helps to build trust and reliability in COTA’s AI models.
Life sciences teams use our data and technology to accelerate clinical trials, meaning patients can benefit from medical innovations sooner. Oncologists can also consult RWD as a sort of “second opinion” for their patients to understand how other patients with similar characteristics have been treated, then use their findings to inform a customized care plan for their patient.
…patients can benefit from medical innovations sooner.
ROH: How do you define RWD and how it differs from traditional data used in cancer research and clinical trials? How will the implementation of RWD improve oncology practice?
CKW: RWD describes any of the information gathered from a patient during routine clinical practice – in other words, from their real care journeys outside the confines of a clinical trial. This includes data from electronic health records, insurance claims, genomic sequencing, wearable devices, and many other sources. When high quality, complete RWD sources are linked together and analyzed, they can offer comprehensive views into patient care and outcomes almost as they happen in the real world.
…comprehensive views into patient care and outcomes almost as they happen in the real world.
Today, most cancer researchers use data from published literature and clinical trials to understand unmet needs and trends in cancer treatment. But these findings can be limited to certain demographic populations, and, depending on when the data were collected, may be out of date with the most recent standards of cancer care. As a result, life sciences companies and oncologists lack a comprehensive view of what treatment truly looks like for today’s patients. RWD studies can be as accurate as traditional research methods, and faster and more representative of diverse patient populations and current clinical care.
ROH: You recently announced a collaboration with Sanofi. How will your combined work accelerate cancer drug discovery?
CKW: COTA and Sanofi will work together to use RWD and AI to make cancer clinical trials faster. We’ll focus on multiple myeloma, a cancer that affects the bone marrow, using data and technology to better understand how patients with the disease are treated today. Our findings will provide a foundation of information to guide future clinical trials, or to contextualize ongoing ones. This means researchers can move faster when designing and executing clinical trials, ensure that the trials will meet patients’ needs, and help patients access breakthrough treatments quickly and confidently.
…help patients access breakthrough treatments quickly and confidently.
ROH: What are the benefits of RWD to life science research and how do you see this sector evolving over the next 5 years?
CKW: Currently, the vast majority of clinical trials fail, and most of them are slow, expensive, and highly complex. As a result, patients are left waiting for the treatments they need. Life science researchers can use RWD and AI to accelerate trials and improve study design, ultimately bringing drugs to market faster and making cancer treatment safer, easier, and more personalized for patients. While the use of RWD and AI in trials is the exception today, not the rule, in the next five years I expect to see life sciences companies continue to invest in data and analytics to complement their clinical trial design, reduce trial costs, and deliver even more impact to the patients they serve.
…making cancer treatment safer, easier, and more personalized for patients.
ROH: You were previously at IBM Watson Health. From your time there and through your role at COTA, what are the key industry learnings from previous technological advancements that should be applied to AI implementation?
CKW: The healthcare industry, rightfully so, is a conservative one that does not easily adopt new solutions or technologies. The burden is on the technology companies to prove, through quality publications, that their solution is trustworthy and of high quality. It is also critical to involve not only the right stakeholders, but also the end users, when an institution is seeking to implement new technologies.
The burden is on the technology companies to prove….
One of the concepts I learned early on was any AI model needs to be unbiased and demonstrate external validity. Otherwise, it should not, and cannot, be deployed for generalized use. Whether one calls this a hallucination or bias, this concept, in recent years, has become increasingly visible due to rapid widespread interest and use of AI technologies. And the only solution to mitigate bias and avoid perpetuating mistakes is to ensure that the datasets used to train AI models are complete and accurate. In addition, to make AI truly useful in cancer care, we need cancer experts in the loop to train and refine AI systems. Cancer is highly complex, and we need these experts to make sure the AI is as accurate, transparent, and reliable as possible to impact patient care.
…any AI model needs to be unbiased and demonstrate external validity.
…to make AI truly useful in cancer care, we need cancer experts in the loop to train and refine AI systems.
ROH: AI regulation is top of mind for many healthcare leaders – what policies or guardrails should be in place to ensure the safe and effective use of AI in healthcare?
CKW: No area of healthcare is looking more seriously at the use of AI to accelerate therapeutic advances than oncology, but there’s work to do to validate and build trust in these emerging AI systems. Reliable oncology AI will require a unique set of cancer-specific guardrails and considerations.
Given the complexity of cancer and its treatment landscape, a standard large language model (LLM) trained by data available on the internet will not cut it. We must ensure that any data used to train cancer AI is high quality and fit for use – in other words, the data is specific to oncology, fully accurate, doesn’t have any missing information, and represents a broad range of demographic groups. These are all essential components of a trustworthy AI output. We’ve already seen examples of AI discriminating against Black patients because the data used to train it didn’t include a diverse enough population. It’s crucial that we train models with the best data to receive the best output.
It’s crucial that we train models with the best data to receive the best output.
As I pointed out earlier, it’s important to note that advances in oncology AI won’t happen without cancer specialists. We hear the phrase “human in the loop” used to describe the need for people to place checks on the responsible use of AI. In cancer, we need people with oncology expertise to test, train, and develop AI systems. They can draw on their clinical knowledge and experiences to double-check AI findings, then tweak as necessary to improve accuracy.
For healthcare institutions looking to implement AI tools, not only is it critical to understand the data from which the tools were trained, it is also vital to establish comprehensive change-management processes during the implementation phase. It’s also essential to understand the risks and liabilities surrounding the use of AI tools, especially when the AI tools give a wrong recommendation.
ROH: What is on the horizon for AI in oncology, and COTA specifically?
CKW: I expect we’ll continue to see diagnostic and pharmaceutical companies explore ways to use RWD and AI to reduce the time and costs associated with developing groundbreaking cancer diagnostics and treatments. We will also see continued adoption of AI tools in clinical care, whether it be for patient engagement, clinical trial operations or decision support. Additionally, we will see an increasing need to validate AI tools utilizing RWD. COTA is proud to work with partners like Sanofi, PreciseDx, and oncologists across the country as we explore new ways to use AI to bring diagnostics and treatments to patients faster and improve care for all.
About Dr. C.K. Wang
Dr. C.K. Wang is an oncologist and Chief Medical Officer at COTA Healthcare where he leads the medical team to meet the interest and demand for real-world data in cancer care. Previously, he served as Manager and Oncology lead at IBM’s Watson Health where he supported their Global AI-driven, oncology efforts. He was Medical Director at USMD Health System. He received his undergraduate training at Washington University in St. Louis and his M.D. from the University of Texas Health Science Center at San Antonio before completing his residency at University Hospitals in Cleveland, OH and his hematology/ oncology fellowship at the University of Texas Southwestern Medical Center.