Artera is a precision medicine company developing artificial intelligence tools to support personalized cancer treatment decisions. Its platform analyzes digital pathology images and clinical data to assess patient risk and predict potential treatment benefit. The company has developed FDA-authorized products for prostate and breast cancer and is expanding its technology across additional solid tumors.
What led you to found Artera, and why did you focus on guiding treatment rather than simply detecting disease?
Personalizing therapy is incredibly challenging. It is relatively straightforward to diagnose a patient, but determining the right treatment—especially for a complex disease like cancer—is much harder. When a clinician says one therapy is better than another, they are implicitly predicting that the patient will be healthier five, 10 or 20 years from now. Humans do not do that particularly well, and doctors will be the first to say this is where they could use support.
That was the genesis of Artera. AI can learn from data across time in a way the human mind simply cannot. Guiding treatment is the natural next step after detecting disease: once a patient has been screened and diagnosed, the patient and clinician must decide how to treat it, often repeatedly during long-term management. That is one of the hardest decision points in a patient’s journey, and it is where Artera sits.
How does Artera fit into a physician’s workflow, and how does it personalize treatment decisions in practice?
A patient suspected of having cancer is screened, undergoes a biopsy and has the tissue reviewed by a pathologist. Once cancer is diagnosed, the clinician can order Artera much like a send-out test. They can run the software themselves or have us run it, then receive a simple report addressing how best to treat the patient and what is likely to happen—whether the cancer may metastasize or progress.
In prostate cancer, doctors often place patients into a few coarse-grained groups—typically low, intermediate and high risk That cannot capture its sheer heterogeneity. For an intermediate-risk patient with localized cancer, a common question is whether to use radiation alone or radiation plus hormone therapy. We can help determine whether that patient will benefit from intensification. That is one of several key clinical questions we can help clinicians answer.
How can better information help doctors avoid both undertreatment and unnecessary toxicity?
Doctors walk a very fine line between undertreatment and overtreatment. If you undertreat a patient, the cancer may worsen and the patient may die. If you overtreat them—by giving too much therapy or the wrong therapy—you expose them to treatment that does not help but can produce very negative, toxic side effects. We have all seen what chemotherapy can do, and most cancer treatments are toxic in different forms.
Getting the treatment right is paramount to a patient’s long-term health span. AI may spare patients unnecessary therapy that would not help their cancer but would seriously affect their quality of life. Patients respond very positively, and we receive wonderful letters thanking us for that work.
What evidence do clinicians need before they will trust and routinely adopt an AI treatment tool?
Tools like Artera must be rigorously validated on high-level evidence and published in peer-reviewed venues. We partner with groups running long-term clinical trials and validate on phase three trials, the gold standard for high-quality evidence in medicine. The results are published in peer-reviewed journals by third parties, allowing clinicians to examine the data and decide how to use the tool.
For good reason, there are regulatory, reimbursement and community-trust steps before a new tool is broadly adopted. You do not want to cut corners; you want rigorous evidence and review by the appropriate bodies, whether the FDA or others. Today, Artera works with almost half of pathology laboratories nationwide and almost half of genitourinary oncologists in the United States. Clinicians are conservative by nature, as they should be, and trust builds in a compounding, longitudinal way.
How have you made Artera scalable and practical for both major cancer centers and community providers?
One of the wonderful things about AI is how scalable it is. It can run almost anywhere, quickly and easily, serving academic medical centers, community clinics and hospitals. In fact, 80% of the patients we help today are treated in the community rather than at leading academic centers.
Only a small number of steps are changed in the clinician’s workflow. We show medical centers the published evidence, explain the phase three validation and remain completely open and transparent. Clinicians begin using the tool, then use it more over time as trust develops.
Why did Artera begin with prostate cancer, and what has that taught you about applying AI across oncology?
Prostate cancer affects one in eight men and is the most common cancer among men in the United States. It gave us an opportunity to serve a very large number of patients. The central lesson is that AI is most useful when it answers the hard questions doctors cannot answer well themselves: whether to give one therapy or a more intensive version of it.
Artera has built an internal AI foundation model that has demonstrated the ability to scale across cancers. We have FDA clearance in prostate cancer and recently obtained FDA clearance in breast cancer, with work extending into bladder cancer and other solid tumors. The question remains remarkably consistent: I know the cancer, its stage and grade—but do I give this therapy, or this therapy plus another one?
What still stands between AI’s technical potential and its broader use in cancer care, and what should remain central as adoption grows?
There is a huge gap between what AI can do today and what is available in clinical practice. The challenge is rarely whether something can be built; it is whether it can be translated into the healthcare system. There are infinitely many potential use cases, from treatment selection to faster triage, but clinicians will have views, regulators need a framework and somebody has to pay. If you want scalable impact, you have to build a business. Those are the challenges holding back innovation, not the innovation itself.
Treatments need to be personalized. I lost a loved one growing up to a chronic illness, and it was not the illness that killed them; it was endless cycles of the wrong treatment being applied until the toxicities took their life. I hope policy clears potentially lifesaving tools in a reasonable timeframe while preserving patient safety. Leading cancer centers will remain early adopters, supporting validation, research and publication, but the goal is broader access—and I would like patient voices to remain central: what are they struggling with?
