startup based in Paris Nabla just was announced that it has raised a $24 million Series B funding round led by Cathay Innovation, with participation from ZEBOX Ventures — CMA CGM’s corporate VC fund. This funding round comes just a few months after Nabla signed a large-scale partnership with Permanente Medical Group, a division of US healthcare giant Kaiser Permanente.
According to a source, Nabla reached a valuation of $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 copilot for doctors and other medical personnel. The best way to describe it is that it is a silent 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 Groll and Martin Raison. Lebrun, CEO of Nabla, was the CEO of Wit.ai, an artificial intelligence assistant startup that was acquired by Facebook. He then became head of engineering at Facebook’s AI research lab FAIR.
A few weeks ago, I saw a live demo of Nabla with a real doctor and a fake patient pretending to have back pain. When a doctor starts a consultation, he presses the start button on the Nabla interface and forgets about his computer.
In addition to the physical exam part, a consultation also includes a long conversation with a bunch of questions about what brings you 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 a written transcript. It works with both in-person consultations and telehealth appointments.
After the patient leaves, the doctor presses the stop button. Nabla then uses a large language model of medical data and health-related conversations to identify the important data points in the consultation – medical vitals, drug names, pathologies, etc.
Nabla creates a detailed medical report in a minute or two with a summary of consultation letters, prescriptions and follow-up appointments.
These reports can be tailored to the doctor’s needs with a personalized format for your notes. For example, you can add instructions to make the note more concise or more comprehensive. Or you can ask to create notes that follow the Subjective, Objective, Assessment and Plan (SOAP) note pattern widely used in the US.
During the demo I saw, I was extremely surprised by the effectiveness of Nabla in general. Even though we were in a packed room and Nabla was running on a laptop a few meters away from the demo presenters, the tool was able to produce an accurate transcript and a useful report.
With Nabla Copilot, as the name suggests, the startup is not trying to take humans out of the medical circle. Physicians still have the final say, as they can edit reports before they are filed in their electronic health record (EHR).
Instead, the company believes it can help doctors save time on admin work so they can spend more time focusing on patients.
“What we do know is that the near future is that we don’t want to try to replace doctors. You’ve seen companies – like Babylon in the UK – burn through $1 billion trying to build chatbots and trying to automate things immediately and take doctors out of the loop. And we have decided this a long time ago with Nabla Copilot [doctors] they are the pilots and we work alongside them,” Lebrun said.
“It’s a bit like automation for autonomous vehicles. We are still on the second level today. We will be launching level three very soon with clinical assurance support. Then level four is clinical decision support, but with FDA approval, because you’re making decisions that you can’t really explain,” he added.
At some point, you could even imagine a level five of autonomous healthcare, which would mean removing doctors from the room. But Lebrun is still very cautious on that front.
“For certain situations in certain markets, like in certain countries where they don’t have access to health care, it would be a relevant thing,” he said. In the long term, he sees the diagnostic process as a “pattern matching problem” that could be solved by artificial intelligence. Doctors would focus on empathy, surgeries and critical decisions.
While Nabla is based in France, most of the company’s customers are located in the US following an installation at Permanente Medical Group. Nabla is not just a work in progress, it is actively used every day by thousands of doctors.
Nabla’s privacy model
Nabla is currently available as a web app or Google Chrome extension. The company is well aware that it handles sensitive data. That is why it does not store audio or medical notes on its servers, unless the doctor and the patient give their consent.
Nabla focuses on data processing rather than data storage. After consultation, the audio file is discarded and the copy stored in the EHR that doctors already use for their patient records.
In more technical terms, when a doctor starts a recording, the audio is transcribed in real-time using a refined speech-to-text API. The company uses a combination of an off-the-shelf speech-to-text API from Microsoft Azure and its own speech-to-text model (a refined model based on the open source Whisper model).
“When you just have a normal speech-to-text algorithm, it may or may not be good on medical data. But we have a perfected one. And, as you’ve probably seen, the text is very light at first and then turns dark. And when it gets dark, it means we verified it with our own model and corrected it with drug names or medical conditions,” Nabla ML engineer Grégoire Retourné said during the demo I saw.
The copy is initially aliased, meaning that personal information is replaced by variables. Pseudonymous transcripts are processed by a large language model. Historically, Nabla has used GPT-3 and then GPT-4 as the major language model. As an enterprise customer, Nabla can tell OpenAI that it cannot store its data and train its large language model on these consultations.
But Nabla is also playing with a refined version of the Llama 2. “In the future, we envision using more and more narrow models as opposed to general models,” Lebrun said.
Once LLM processes the transcript, Nabla de-anonymizes the output. Doctors can view the note, which is saved on the computer in the local web browser’s save file. Notes can be exported to an EHR.
However, doctors can give their approval and ask for patient consent to share medical notes with Nabla so they can be used to correct transcription errors. And since Nabla is on track to process more than 3 million consultations annually in three languages, chances are Nabla will improve very quickly thanks to real-world data.
Image Credits: Romain Dillet / TechCrunch
