Medical-grade AI that reads the data hospitals already have - and predicts who needs care before the emergency arrives.
It is an ordinary Tuesday in an emergency department. Someone describes their symptoms into a form. Before a clinician has read a word, a Diagnostic Robotics model has scored the case, ranked the likely paths, and flagged whether this person is heading toward a hospital bed or a phone call home.
This is the quiet thing Diagnostic Robotics actually does. Not a robot in a white coat - the name is a little theatrical - but software that turns billions of insurance claims and nearly a hundred million prior patient visits into a usable forecast. Health plans and providers use it to find the patients who will get expensive and sick later, and to do something about it now. The company is headquartered in Boston, employs around 66 people, and counts Mayo Clinic among both its partners and its investors.
Healthcare has a budget problem and a workforce problem, and they feed each other. There are not enough clinicians, the ones who exist are exhausted, and a startling share of spending goes to crises that were visible months in advance - the heart failure that deteriorated, the avoidable ER visit, the mental health decline that went unmanaged.
The data to see these patterns mostly already exists. It sits in claims files and visit records, vast and largely unread. Diagnostic Robotics' founders looked at that pile and saw the obvious, uncomfortable gap: the system collects enough information to predict who is at risk, and then waits for them to show up in crisis anyway. Knowing is not the bottleneck. Acting on what you know is.
Figures reported by the company and press coverage. Treat the round ones as round.
Diagnostic Robotics was founded in 2017 by Kira Radinsky, Jonathan Amir and Moshe Shoham. Radinsky had been eBay's director of data science and chief scientist in Israel before most people finish a PhD program's small talk. Shoham is a Technion robotics professor - the academic weight behind the name. Amir rounded out the founding team on the business and product side.
Their wager was specific. Plenty of companies could build a model that predicts illness in a slide deck. Far fewer could build one a clinician would trust at the point of care - validated, explainable, and accurate enough to route a real human toward the right next step. They bet that "medical-grade" was the hard part, and the part worth owning.
The platform is less one app than a stack of predictive jobs. A patient answers a simple questionnaire; the system reads symptoms, history and results, predicts the likely course of care, and points them to the right site - ER, clinic, or home. Underneath, models score populations for who will deteriorate, and a virtual agent handles the flood of routine medical questions that would otherwise eat a nurse line alive.
Analyzes a patient questionnaire in real time, predicts the most likely care path, and routes patients to the most appropriate site of care. Built to supplement - not replace - clinician judgment.
Care-management platform with predictive models for avoidable ER visits, congestive heart failure deterioration and serious mental illness, so plans can target interventions where they matter.
An AI agent that fields members' medical inquiries at scale - reportedly covering more than 98% of them - and escalates the cases that genuinely need a human.
A pandemic-era risk assessment and remote-monitoring tool, rolled out statewide to Rhode Island residents and offered to governments worldwide.
Skepticism is the correct posture toward any company that says "AI" and "healthcare" in the same breath. So here is the evidence, not the adjectives. Mayo Clinic agreed to co-develop automated ED triage and later joined the Series B - a customer and an investor are two different kinds of conviction. Blue Cross & Blue Shield of Rhode Island runs care-management programs on the models. During the pandemic, an entire US state pointed its residents at the triage tool.
The $45M Series B in July 2022 was led by StageOne Ventures, with Mayo Clinic, the Technion, and Berkshire Partners co-founder Bradley Bloom among the backers - alongside a roster of Israeli institutions. Total funding sits around $84M. None of this proves the models cure anything. It does prove that serious, cautious buyers were willing to put both patients and capital behind them.
Bars are scaled for readability across different units, not a like-for-like comparison - the point is the orders of magnitude, not a horse race.
The mission is unromantic and, for that reason, credible: relieve strained health budgets and workforces by directing clinically-actionable care to the right patient at the right time. No promise to cure disease. A promise to waste less - less avoidable utilization, fewer crises that a phone call could have headed off, fewer clinician hours spent on cases an algorithm could triage.
There is an honest tension here that the company does not pretend away. Prediction in medicine is only as good as the action it triggers, and only as trustworthy as it is explainable. The bet is that "medical-grade" - validated, auditable, accurate at the point of care - is what separates a useful tool from an interesting one. The competitive field is crowded with healthcare AI and payer-analytics firms; the differentiator Diagnostic Robotics claims is the rigor, not the novelty.
Populations are getting older, chronic conditions are getting more common, and the supply of clinicians is not keeping pace. That math does not resolve itself with effort. It resolves with leverage - with software that lets a fixed number of clinicians reach the patients who need them most, sooner. If value-based care is the direction the industry is heading, prediction is the engine, and Diagnostic Robotics is building engines.
Return to that ordinary Tuesday. The patient fills out the same short questionnaire. But now the form is not a formality - it is the first read of a system that has seen a hundred million visits before this one. The model scores the case, the clinician sees the patient who actually needs them, and the heart-failure crisis that used to arrive by ambulance instead arrives as a phone call three weeks earlier.
That is the change Diagnostic Robotics is trying to make ordinary. Not a robot doctor. A quieter, stranger idea: a healthcare system that calls you first.