Outsmarting Cancer: How AI and Digital Twins are Pioneering Precision Theranostics

United Theranostics
Nuclear Oncology Resource

Overview

Over the last several decades, medical imaging has evolved from producing static pictures to generating massive, multi-dimensional datasets, arming oncologists and their patients with more data than was ever thought possible. Despite this, most cancers are still fought using one-size-fits-all tools, often relying on standard dosing, larger doses than are necessary (or effective), and underutilizing recent advances in medical imaging that allow cancer to be tracked and treated more precisely.

The intersection of advancements in medical imaging, the ability to synthesize data in never-before-seen quantities, and the computing power to run thousands of simulations in a clinical environment is where Eliot Siegel, MD, FACR, FSIM sees the future of theranostics heading. As the Co-Founder of United Theranostics and an adjunct professor at the University of Maryland School of Medicine, UMBC, and UMCP, he has a firsthand view of how physicians will be able to use recent advancements in cancer research in conjunction with technological advancements to develop more personalized and precise treatment options in the future.

“We must move beyond standard dosing protocols and ecological clearcutting,” he explains, “and instead embrace the convergence of three foundational concepts:

• The Theranostic Digital Twin
• Computational Oncology
• The Practical Application of AI”

A Treatment Plan Built Around You: The Theranostic Digital Twin

Historically, radiopharmaceutical therapies have relied on fixed, population-based dosing, in which patients are given standard doses of radiopharmaceuticals based on clinical trial data across large populations, rather than how those individuals respond to treatment. But this approach ignores the immense biological variability between patients. One approach to overcoming this that is being considered in the field of theranostics is the development of a dynamic, mathematical replica of a patient’s specific anatomy, physiology, and disease trajectory referred to as a “digital twin.”

Dr. Siegel uses the example of an 80-year-old female patient diagnosed with a metastatic small bowel neuroendocrine carcinoid tumor that’s also affecting her liver, lymph nodes, and bone marrow, that is being treated with Lu-177 DOTATATE (Lutathera). “In this example, if we imagine that she also suffers from chronic kidney disease, we’re now operating in a dangerous clinical blind spot.”

The answer, he elaborates, can be found in running dosing simulations on a digital twin—functionally, a virtual model of the patient’s body that can predict how kidneys, bone marrow, and other organs might respond to varying dose sizes over fluctuating time periods. In this future scenario, the care team will be able to run thousands of simulations, varying doses, timing, and other variables to identify the best way to deliver radiopharmaceutical treatment.

Weaponizing Cell Competition

For decades, the standard strategy in oncology has been to deliver the Maximum Tolerated Dose (MTD). Dr. Siegel describes this approach as “hitting the tumor as hard as possible to achieve complete eradication.”

While well-intentioned, there are several critical issues with this aggressive strategy. Primarily, MTD often causes what Dr. Siegel refers to as ecological clearcutting: wiping out sensitive cells but leaving behind an ecological vacuum. When this occurs, cells that are naturally resistant to the treatment are allowed to thrive without competition, which can lead to a rapid, treatment-resistant relapse.

A Strategic Approach: Computational Oncology

A new frontier, known as computational oncology, is exploring powerful alternatives. By applying mathematical modeling, computer science, and advanced medical imaging, computational oncologists are discovering that strategic breaks in treatment can actually be highly beneficial.

Allowing a controlled number of treatment-sensitive cancer cells to survive and grow back maintains cellular competition, significantly decreasing the likelihood that purely resistant cancer cells dominate the ecological vacuum.

“We can already see this concept validated in the broader medical literature through the use of intermittent androgen deprivation therapy (ADT) cycling for prostate cancer,” notes Dr. Siegel. By taking strategic breaks in hormone therapy, a treatment modality typically managed by urologists, physicians have been able to successfully maintain this cellular competition. “In many cases, it’s allowing the field to transform metastatic cancer from an acute, failing battle into a manageable, chronic condition.”

Applying the Precedent to Theranostics

This established precedent paves the way for a similar strategic approach in nuclear medicine. While United Theranostics does not administer ADT, they plan to adapt this exact philosophy of strategic cycling when treating patients with targeted radioligand therapies, such as Pluvicto or Lutathera.

Moving forward, computational oncology will help physicians develop dynamic mathematical models based on a patient’s specific biomarker levels, tumor volume data, and treatment response history. These models will simulate exactly how cancer cells will react to targeted radiation over time, empowering theranostic specialists to fully personalize care and predict the absolute best times to cycle radiopharmaceutical treatments on or off to outsmart the tumor’s evolution.

AI as the Clinical Co-Pilot

United Theranostics is already utilizing AI to rapidly synthesize data, creating a holistic view of their patients’ entire health profiles by cross-referencing lab data, surgical history, and recent treatment outcomes. Moving forward, Dr. Siegel expects AI to evolve from a purely retrospective tool into a predictive simulator.

Paired with the digital twin, AI will soon be capable of running thousands of complex treatment simulations. A major catalyst for this leap will be the integration of continuous liquid biopsies and spatial transcriptomics alongside advanced dosimetry.

Liquid Biopsies: Playing Chess with Cancer

While liquid biopsies have been used for roughly a decade to test blood samples for fragments of tumor DNA, the strategy of repeatedly taking these samples to track treatment effectiveness in real-time is relatively new. Utilizing AI to interpret this continuous stream of data and actively update patient models, like the digital twin, is virtually on our doorstep.

Dr. Siegel finds this prospect particularly compelling. He believes United Theranostics “is very close to a future where liquid biopsies will help AI to play chess against the tumor’s biology and alert the physician to pivot therapies to prevent the cancer from developing resistance.”

Spatial Transcriptomics: Uncovering the Cellular “Why”

Similarly, spatial transcriptomics utilizes AI to map the gene expression (RNA) of microscopic tissue samples within their spatial context, precisely defining the type and quantity of cancer cells present. This paints a much fuller, more dynamic picture of the patient’s specific disease.

As Dr. Siegel points out, “The AI will help us to determine why one specific metastasis is resisting treatment at a cellular level.” This level of insight will empower oncologists to deploy multiple targeted treatments simultaneously, managing the cancer holistically based on the unique genetic makeup of specific cellular populations.

“Especially when utilizing imaging information from whole-body PET scans, oncologists will have the ability to treat a single lesion with external beam radiation, for example, while continuing combined systemic therapies (like chemotherapy, hormone therapy, and radiopharmaceutical therapy) for the rest of the body.” — Eliot Siegel, MD, FACR, FSIM

Precision Theranostics: Already Underway

United Theranostics is already laying the groundwork for these developments across its Nuclear Oncology Centers of Excellence, utilizing AI to do far more than just automate paperwork or track lesions.

As Dr. Siegel describes it: “Our AI platform acts as a clinical synthesizer. It doesn’t just look at today’s PET/CT; it simultaneously cross-references the patient’s entire reality: their genomic profile, real-time lab data, surgical history, and the toll of past therapies.”

As a result, United Theranostics’ Nuclear Oncologists can map out the precise trade-offs of various therapy cycles with unprecedented accuracy. This ensures clinicians maximize the radiation’s impact on the tumor while aggressively protecting vital organs.

“We’re already very close to implementing many of these advances directly in our clinics,” notes Dr. Siegel. “The bottom line is clear: while AI handles the immense complexity of the data, we, the theranostic specialists, will be able to focus entirely on what matters most—our patients.”

This article is for informational purposes only and does not constitute medical advice. Please consult with your physician or a qualified healthcare provider regarding your individual treatment options.