By Reza Malekzadeh
It is generally rare for venture funds to invest in both life-sciences and general hardware and software technology with the same team. The market dynamics and the investment cycles are very different. In traditional life sciences investments, the skill sets required for aiming at success were very different than those in traditional hardware and software technologies.
But the recent trends in software and hardware are helping with the emergence of a new segment in biotech that are far more about “tech” than “bio”. We can now fund what are mostly software companies but applied to the field of life sciences.
Compute and Storage Costs hit rock bottom and then some
We are seeing continued decrease of costs of compute. Some argue that Moore’s law is fading away but regardless, compute power is cheaper and cheaper by the day. If you look at the cost of compute on the major cloud providers like AWS, GCP or Azure, you see that the equivalent of super computers is now barely a few thousand dollars per month. Add to that the massive drops in the price of storage. Not a single quarter goes by without one of the major players announcing new drops in storage price. These trends have made big data possible with a huge realm of new possibilities arising from it.
Machine Learning and AI with a focus
We all know that machine learning and artificial intelligence are the new buzzwords that everyone touches one way or another. As my friend Alex Lebrun recently pointed out, when he started pitching his first entrepreneurial venture a few years ago, using AI or ML was rather perceived negatively and he had to remove those terms from his slide deck in order to be able to raise funding. And today, there is not a single pitch deck that does not contain those terms.
But for the most part, today, ML and AI do not have a true and full impact. The models are very hard to train generically and there is always a human trainer in addition to the machine. We are probably many years away from a generic model or a true autonomous vehicle (for example).
A focused area of training makes all the more sense then. If you have a limited vocabulary on a large data set, then you can see much greater return and benefits from the algorithms. Deep mind won the game of Go because it was specifically trained for that but could probably not drive a car for you.
A big part of healthcare and medicine involve data and images. A doctor will analyze your X-Ray, your test results, your MRI, etc. They will leverage their knowledge, expertise and experience. But this is a field where the machine can help. An algorithm can probably dissect images and detect weak signals that the human eye might have missed. So technology can very clearly help and augment the medical professional and help save more lives.
Startups with a purpose
So we think that this new wave of technology that can improve medicine for the greater number. My colleague Claire Godron and I are very excited about our latest investment in Cardiologs, a startup born out of the Paris-Saclay research centers. The entrepreneur, Yann, and his team have trained a neural network using more than 500,000 electrocardiogram recordings, and this training dataset keeps growing. The result is that Cardiologs will help both expert cardiologists and general practitioners better recognize patterns in a cardiac signal for fast and precise analysis of heart diseases. This is particularly powerful for long-term recordings that require today a very laborious human analysis process.
What particularly convinced us in Cardiologs is their close relationship with top-notch cardiologists that have been labeling the electrocardiogram database in the past 3 years. This is this combination between medical expertise and machine learning technologies that explains the technical performance of the tool, and builds a solid entry barrier to more generalist players.
Today, an estimated 70% of ECGs transmitted to ECG centers for being re-read are false-positives. Leveraging Cardiologs can improve their efficiency by at least 2X. That productivity gain will directly translate into being able to handle more patients per day with an unprecedented level of accuracy and provide them with better treatment.
Cardiologs received FDA clearance of its Cardiologs ECG Analysis Platform this past summer.
As you can tell by now, Claire and I are very happy and proud to welcome Cardiologs to the Partech Ventures family. This is a great example of a data-driven and computational approach to medicine leveraging the benefits of Cloud computing. It will hopefully save lives by improving the accuracy and quality of medicine.