Open roles

</aside>

Founders Associate @ Radical

<aside>

Problem

</aside>

Every cancer patient is different. Personalising treatment is the only cure.

There are thousands of potential treatment plans for a cancer patient. We don’t know what’s right for each individual so clinicians use a one size fits all approach.
We have millions of data points on each individual patient (radiology, pathology and genetic data). Doctors aren’t computers, they can’t computationally process all these data points to make the correct treatment decisions.
Learning how to pick the right treatment plan for a patient will improve outcomes more than any new drug that will come to market (fun fact: the average new cancer drug only improves survival by 2.8 months)

<aside>

Solution

</aside>

AI Cancer Doctor to Make Personalized Treatment Decisions

We’re building general-purpose models to make every cancer treatment decision. We have sole access to the highest quality healthcare dataset in the US (6mill multi-modal patient records and and 146mill clinical notes).

In the short term, here are some example clinical questions the model will tackle:

What is the optimal first-line treatment for stage 4 colorectal or pancreatic cancer? Currently, there are two main options, but humans don’t know the optimal one. We believe we can extend life by 20% just by learning how to pick the right for a patient.
What is the optimal dose? We’re able to suggest dose reductions that both save money and extend life, e.g. 30% dose reductions for pancreatic cancer patients.
Which patients will not respond to immunotherapy in metastatic lung cancer? We can avoid giving therapies that cost over $200k per year to patients who would do better on standard chemotherapy.