Startup in Paris dart Just Announce It has raised a $24 million Series B funding round led by Cathay Innovation, with participation from ZEBOX Ventures – CMA CGM’s corporate venture capital fund. This funding round comes just months after Nabla signed a large-scale partnership with Permanent Medical Group, a division of US healthcare giant Kaiser Permanente.
According to a source, Nabla has reached $180 million after today’s funding round. The company could also end up raising more money from US investors as part of this round.
Nabla is working on an AI-powered co-pilot for doctors and other medical staff. The best way to describe him is as a silent work partner who sits in the corner of the room, taking notes and writing medical reports for you.
The startup was originally founded by Alexandre Lebrun, Delphine Grohl, and Martin Raison. LeBron, CEO of Nabla, was the CEO of Wit.ai, an AI assistant startup acquired by Facebook. He then became head of engineering at Facebook’s artificial intelligence research lab, FAIR.
A few weeks ago, I watched a live dart demonstration with a real doctor and a fake patient pretending to have back pain. When the doctor starts the consultation, he presses the start button on the Dart interface and forgets about his computer.
In addition to the physical examination portion, the consultation also includes a long discussion with a set of questions about why you came here and your medical history. At the end of the consultation, there may also be recommendations and prescriptions.
Nabla uses speech-to-text technology to convert conversation into written text. It works for both in-person consultations and telehealth appointments.
After the patient leaves, the doctor presses the stop button. Nabla then uses a large linguistic model optimized using medical data and health-related conversations to identify data points that are important in the consultation – biomedical factors, drug names, diseases, etc.
Nabla prepares a comprehensive medical report within a minute or two with a summary of the consultation, prescriptions and follow-up appointment letters.
These reports can be customized to your doctor’s needs with a custom format for your notes. For example, you can add instructions to make the note more concise or detailed. Or you can request that notes be created that follow the Subjective, Objective, Evaluate, and Plan (SOAP) pattern widely used in the United States.
During the demo I watched, I was very surprised by Nabla’s overall effectiveness. Even though we were in a crowded room and Nabala was working on a laptop a few feet away from the presenters, the tool was able to generate accurate text and a useful report.
With Nabla Copilot, as the name suggests, the startup is not trying to take humans out of the medical loop. Doctors still have the final say as they can modify reports before submitting them into the electronic health record (EHR) system.
Instead, the company believes it can help doctors save time on administrative work so they can spend more time focusing on patients.
“What we do know is that in the near future we don’t want to try to replace doctors. You’ve seen companies – like Babylon in the UK – spending a billion dollars trying to build chatbots and trying to automate things on the fly and take doctors out of the loop. We decided that a long time ago With Nabla, the co-pilot [doctors] They are the pilots and we work alongside them,” said Alexandre Lebrun, co-founder and CEO of Nabla.
“It’s a bit like automation for self-driving vehicles. We’re still at level two today. We’ll start level three very soon with clinical assurance support,” he added. “Level four is clinical decision support, but with FDA approval, because you’re making decisions that you can’t really explain.” “.
At some point, you can even imagine a fifth level of independent health care, which would mean removing doctors from the room. But LeBron is still very cautious on this front.
“For some situations in some markets, like in some countries where they don’t have access to health care, that will be relevant,” he said. In the long term, he sees the diagnostic process as a “pattern-matching problem” that can be solved using artificial intelligence. Doctors will focus on empathy, surgical procedures, and critical decisions.
While Nabla is based in France, most of the company’s customers are in the US after launching through Permanente Medical Group. Nabla is not just a work in progress, it is actively used every day by thousands of doctors.
Nabla Privacy Form
Nabla is currently available as a web application or extension for the Google Chrome browser. The company is well aware that it is dealing with sensitive data. For this reason, it does not store audio or medical notes on its servers, unless the doctor and patient agree to this.
Nabla focuses on data processing rather than data storage. After the consultation, the audio file is discarded and the transcript is stored in the electronic health record that doctors already use for their patient files.
In more technical terms, when the doctor starts recording, the audio is transcribed in real-time using a fine-tuned speech-to-text API. The company uses a combination of Microsoft Azure’s out-of-the-box speech-to-text API and its own speech-to-text model (an enhanced model based on the open source Whisper model).
“When you have just a regular speech-to-text algorithm, it may or may not be good on medical data. But we have a fine-tuned one. As you’ve probably seen, the text is very light at first, and then it gets dark.” And when it gets dark, that means We verified it with our own model and corrected it with the names of drugs or medical conditions,” said Gregoire Retourne, Nabla ML engineer, during the demo I saw.
The text is first aliased, which means that personally identifiable information is replaced with variables. Aliased texts are processed by a large language model. Historically, Nubla used GPT-3 and then GPT-4 as its main macrolanguage model. As an enterprise customer, Nabla can tell OpenAI that it cannot store its data and train its large language model for those consultations.
But Nabla was also using a modified version of Llama 2. “In the future, we envision using more and more narrow models instead of general models,” Lebrun said.
Once LLM processes the text, Nabla aliases the output. Doctors can see the note stored on the computer in the local web browser storage file. Notes can be exported to electronic health records.
However, doctors can give their consent and request the patient’s consent to share medical notes with Nabla so that they can be used to correct transcription errors. Given that Nabla is on track to process more than 3 million consultations per year in three languages, Nabla will likely improve very quickly thanks to real-world data.
Image credits: Roman Delete/TechCrunch