Diagnostic Robotics, which has developed an artificial intelligence system for clinical prediction, has announced that it has raised $45 million. Diagnostic Robotics was founded by Yonatan Amir, together with Dr. Kira Radinsky, who became famous when, at a very young age, she sold artificial intelligence-driven prediction company SalesPredict to eBay. A third founder is Prof. Moshe Shoham, inventor of the technology of Mazor Robotics, which developed a robotic guidance system for spine surgery and was sold to Medtronic for $1.6 billion, and that of several other companies. Radinsky is CEO of Diagnostic Robotics, after replacing Amir, who became company president.
The investment was led by StageOne, with participation by Migdal Insurance, Clal insurance, Bank Hapoalim, and US medical center The Mayo Clinic. Private investors Bradley Bloom, who is a founding partner of private equity firm Berkshire Partners, and Gigi Levy also participated in the current round. Among past investors in Diagnostic Robotics are Accelmed, Maverick Ventures Israel, Alpha Capital, and Dr. Kobi and Dr. Judith Richter. So far, including the current round, the company has raised $84 million.
Diagnostic Robotics was founded in 2017 with the aim of using artificial intelligence to improve processes at hospitals; for example, to predict which patients were liable to deteriorate rapidly, or to manage patients’ visits for tests and appointments with doctors, in order to make more efficient use of resources and shorten waiting times.
In the past few years, however, the company has changed its model, and it now focuses more on clinical prediction in the community rather than at hospitals. The company has agreements with several insurance companies in the US, and it monitors some 28 million users. Its revenue, according to Radinsky, is "substantial".
The company hit the headlines at the start of the Covid-19 pandemic when it sought to use its artificial intelligence-based predictive capabilities to predict where new centers of Covid-19 cases would develop and the pattern of the disease’s spread. Radinsky had in the past dealt with prediction of disease spread in the course of her academic work. With Covid-19, however, Radinsky explained to "Globes", "There was no great need for forecasting. We didn’t need to identify focal points of the spread of Covid-19 because it was everywhere, and the trends could be understood from looking at the numbers, even without a prediction system." The current spread of monkeypox too is not an interesting market for a system such as that of the company, she says. But making the process of identifying patients who could benefit from taking Paxlovid to prevent severe Covid-19 symptoms more efficient could be interesting.
This is part of what Radinsky calls "triage in the community". The term triage generally refers to the process of prioritizing patients who come to hospital emergency departments. The new approach is to catch them before they arrive at the emergency department. "The state needs to recognize people at risk, and approach them proactively, to prevent them from deteriorating," Radinsky says.
At this stage there is the best chance of preventing deterioration, Radinskty explains. "We work on the basis of information from the patient’s personal file and also from questionnaires that we send to patients from time to time, in order to identify people who are on the brink of deterioration in their condition, using our artificial intelligence system, which has already seen tens of millions of files. We have shown that we succeed in preventing unnecessary hospitalization. In some cases, our agreement with the insurance company enables us to receive payment derived directly from the saving we produce, and the insurance companies recognize the economic value that the company’s tools yield them, and they reward us accordingly."
We don’t look for the sickest patients
The company has products relevant to some 25 different diseases, of which five have undergone the entire process of validation, and the ultimate intention is to combine them into a single predictive tool that takes into account the patient’s state of health in a holistic way. Meanwhile, the company’s leading product is designed to deal with congestive heart failure. The financial cost of hospitalization following congestive heart failure is very high. Attacks can be prevented outside the hospital with simple drugs, if the deterioration is caught in time. Several companies, including Israeli companies, are developing wearable, or even implanted, devices to predict an attack.
"The problem is the low take-up by patients of the wearable devices," Radinsky says. "Our approach is to send questionnaires, or for a nurse to contact patients and ask about symptoms, and we also have one monitoring device - a digital scales. We aren’t looking for the sickest patients, who will agree even to an implanted predictive device, but for those in the middle, who live their lives and don’t pay attention to deterioration, but who are liable to be harmed by it in the most significant way, and in their case prevention is especially important."
Among other products of the company is a system for identifying endometriosis. Through the medical files of female patients it identifies who may be presenting a clinical picture of endometriosis, and starts to send her questionnaires designed to capture the cyclical character of the disturbance. "Today, it takes three to eight years to reach a diagnosis of endometriosis. Our aim is to shorten that significantly," Radinsky says. Additional products are designed to diagnose COPD, complex psychiatric diseases, or deterioration of metabolic diseases.
"We see that a patient has received a diagnosis of pre-diabetes, and his data are not improving; perhaps it’s necessary to hold him by the hand more. Or a patient with a chronic disease who we see has stopped going to the doctor and taking drugs; perhaps he has also developed depression alongside the chronic disease. We look for patients who can be helped by intervention. There are also those that we identify will not be helped by intervention of this kind, and then we don’t try. For example, we can predict that for a particular patient there is no point is repeatedly sending messages about nutrition and diet. For that patient, it would by now be better to invest in drug treatment."
"We started the fund raising round after the market had started to collapse"
On the face of it, a family doctor who knows the patient and is in continual touch with him or her should be able to spot some of these things and to adapt treatment in accordance with knowledge of the patient’s characteristics. Particularly in the US, however, where the company’s main activity is concentrated, the family doctor who really knows their patients in depth is a disappearing phenomenon. "All these programs are currently operated by nurses, while some of the activity is digital only," Radinsly says. "As part of the great resignation in the US, there is a shortage of nurses. In some clinics, as many as a third of the nurses have resigned at once, or the cost of employing them has risen. So our system also tries to make use of nurses’ time more efficiently."
In the US, the activity in the community is the company’s leading product. In South Africa and Israel, the other two countries in which the company is active, the hospital management products are still the ones most in demand.
Raising $45 million at this time is no simple matter, Radinsky says. "We started the fund raising round after the market had started to collapse. This time, it wasn’t straightforward," she admits. "Many investors had not just become more selective or more demanding, but had simply frozen investment." Nevertheless, the company completed the round fairly quickly. "It happened because we are a growth company that already has a track record in the market," says Radinsky.
Published by Globes, Israel business news - en.globes.co.il - on July 25, 2022.
© Copyright of Globes Publisher Itonut (1983) Ltd., 2022.