The Physician and His A.I. Will See You Now
We have all heard of the I.B.M Watson, playing doctor, letting people wonder if all these A.I. things, “deep learning” and “big data”, combined with genomics, will somehow be poised to “replace” physicians? Will we be diagnosed by machines? This reminds me of “E.T. phone home”…
There is a competitor of I.B.M. which we hear little about, and yet impresses “insiders” by its technological as well as business performances: getting Doctors to come down from their ivory tower, or entering the forbidden tower, depending on your perspective. Sophia Genetics, a company founded in Switzerland by a French biologist, Jurgi Camblong, based in Geneva and soon in Bidart, in the French Basque country, trying to attract, for its new R & D center, the “data wizards” working in California, but who are starting to find that it is not so easy to surf inside of the wave, or maybe they felt that they have reached the end of their tether… They can still move their surf board from San Diego to Biarritz. Did you know that MIT Tech Review has ranked Sophia Genetics as one of its 50 smartest companies in spring 2017, ranking 30th out of 50, together with Alibaba or Tesla? Sophia is ranking among the best, among companies with a promising technology, a solid and operational business plan, enabling a rapid and sustainable growth (source).
More than 200,000 patients worldwide have been diagnosed with the A.I. from Sophia Genetics — an aid to medical diagnosis. Areas of application: paediatric cancer, oncology, innate and acquired cancer, solid metastatic cancer, diabetes, etc. Today, this company helps diagnose a patient every five minutes. According to Tarik Dlala, in charge of marketing (Geneva, Switzerland, statement dating from May 1st, 2018), the company has set up the goal of reaching a patient every five seconds. Its primary purpose is to significantly enhance the time to diagnosis and treatment of each patient, knowing that medical diagnosis are often achieved late, with a (sometimes major) delay that may be detrimental to patients.
Sophia Genetics and its diagnostic aid product lead to the most adapted and fastest possible care for beneficiary care centres. Their business plan is simple; in a nutshell: gain the trust of hospitals and clinics. And they have been successful. Their A.I. device is now used in nearly 500 hospitals and clinics in 60 countries, with an international growth, outside of France and Switzerland, that started in summer 2016: Canada, Africa, USA, Australia, etc. The prices of the tests may vary between 50 and 300 EUR per unit (price range given by way of indication only, can be subject to change, plus I’m quoting from memory), and can be partially reimbursed in France by the Social Security. Patients wandering such as Ulysses and Telemachus in search of an “actionable” medical diagnosis may visit the company’s website and try to find out, depending on their country and location, the nearest hospital or clinic, using Sophia Genetics’ A.I. device ; alternatively, they may inquire with patient associations, which is not easy when the illness that you suffer from has never been diagnosed correctly, which happens more often than one might think. The sequencing of the patient’s genome is done beforehand in the “Sophia” user care centres, or in centres authorised to do genetic sequencing for some of these centres not having (yet) machines (Illumina sequencers).
“- Our business model is B to B.”
The Sophia Genetics business model consists in selling an A.I. that helps diagnosis in clinics and hospitals, mostly in cancer. They do not sell data, but knowledge.
My question: how to turn your product into a benefit for a client (the clinic) knowing that advertising is prohibited for the client of the client (the patient)? Sophia Genetics has the right to do B to B, but not B to B to C. Which brings us back to ethics: care centres will treat their patients better with the “Sophia” A.I., but they have no right to say so explicitly to the client or patient, because of prohibited medical advertising in some countries, like France. I’m curious to know how Jurgi Camblong and his team tackle this problem: the B to B to C business plan cannot be implemented. It has to be B to B, from a legal point of view.
I was part of the tiny Intuitive Surgical (Inc.) European team between 2002 and 2007, at sales and marketing, to help install and develop computer-assisted, made-in-California minimally invasive surgery. In California, Intuitive Surgical offices were located right across the street with offices of the Apple management; if you were lucky you were able to wave hello to Steve Jobs. Of course this is also a problem we experienced first hand with Intuitive Surgical Europe: a B to B to C business plan, with no access to C, and we did not manage to solve the problem — apart from the time one of the film production assistants contacted me to ask me to lend her a computer-aided surgery da Vinci TM for the James Bond “Die Another Day”. I thought it was a hoax but no. We were able to complete this project.
So to push the “Genomic Precision Medicine” innovation in France, I see the same problem as for minimally invasive, computer-aided surgery … Generally speaking, it is the same problem that we encounter with health innovations. Big Pharma is the only one able to get around this problem. They manage to push innovation. But smaller players, who have strong or sharp innovation, face this problem. In particular, I suppose, because these small innovative actors bring a major plus to a few patients. For Big Pharma, you have to flip the issue around, as it is a question of bringing a little plus to many patients.
“ — We have a B to B business model,” said Tarik Tlala, the marketing manager of Sophia, who on Tuesday May 1st was in videophone conversation with French blogger and specialist in digital economy Jean-Michel Billaut:
(The interview is in French.)
"Very interesting e-interview in my video studio with Tarik Dlala, Marketing Director of Sophia Genetics ..
Sophia Genetics is a Swiss startup, created by a French (Jurgi Camblong) who offers hospitals around the world (almost 500 today, including 40 in France) an aid to the diagnosis of cancer. Tarik Dlada explains in detail all the mechanics: operation, business model, future developments (sequencing "direct to consumer"?). Sophia has already assisted in the diagnosis of 200,000 patients in the world (1 aid to patient-diagnosis every 15 minutes or so) Objectives: 1 patient every 5 seconds.Why has Sophia just invested in France in the Basque country?
Comparison with the program "genomics 2025" of the French government officials?My comment
Does this French genomics program (by the French government) have a future? Compared to China, the Americans, the Brits, the least we can say is that we are late to the game... Why not have an agreement with Sophia Genetics? While the "cost of a genome" as experts say today is in the range of $ 500 in China? And new platforms, with blochchain technology, are trending (Zenome, Nebula, Shivom .. etc). Always a bit fuzzy, the French elite ...Sophia remains in the field of curative medicine. Will the company evolve towards predictive medicine? Facebook Wall of Jean-Michel Billaut, May 3, 2018.
“ — Their business plan confines them to large regional hospitals,” commented Patrick Merel, a French biologist, from Bordeaux university, and founder in 2010 of the startup Portable Genomics, San Diego. Mr. Merel was present during Mr. Tlala’s excellent presentation. I would say that’s probably the case at the moment, but if Mr. Merel, who knows all the ins and outs of made-in-California precision genomics medicine, wants to partner with Sophia Genetics, then some acceleration could happen, a movement of democratisation, besides claimed by Sophia, which means wisdom and knowledge in Latin …
Among my clients, some computer savvy folks working for Tencent. I asked them if it would be ok for us to trust pharmaceutical companies or industries to make applications dedicated to diagnosis and health, used directly by patients. After all, why use a bunch of mathematicians to develop an AI helping doctors to diagnosis, mainly in oncology for the moment, in a secure cloud, that requires neither maintenance by dedicated medical teams, nor specific training? You do not have to be a geneticist or a computer scientist to use the “Sophia” A.I. No need to know the computer language used in genomics, Python, or the R-stats and MATLAB analysis tools, for example. Let’s not forget that genomics boils down to calculations of probabilities and statistics, that the R-stats and MATLAB languages can do. What if the patient could have an application that does the job? He would recover his own data, the app would crunch the data and tadaaaa… diagnosis, spot on. It would be so much simpler, right? Well, it turns out no, and here’s why:
“ — If the application is made in support of a drug manufactured by a pharmaceutical lab, there is a chance that there will be a budget to monitor, update the application, do the maintenance etc. But this is not the most common case. And without this budget, your app is the Titanic. “
“ — And what is the most common case?”
“ — Employees of pharmaceutical companies are on permanent contracts, but not the applications that they develop and distribute to their patients and customers via the AppStore. These apps are on short-term contracts.”
“ — Marketing department makes an app — communication is a crucial milestone. But their budget is just money that is burned, as no return on investment is expected, and especially, their budget is set each year, with objectives based on communication; not product monitoring. And so it varies every year. An app made by a marketing service will not have the budget to update the iOS environment of the app, or if there is the plan of an update, it will have to be in Swift, a rapidly evolving computer language, or C sharp (Microsoft), whereas some languages are progressively lagging behind, because they are a real labyrinth that is incomprehensible, like C and C ++. Java, bought by the Oracle group and their armies of lawyers, want clients (Java users) to pay royalties to the patent holder (Oracle). Java users like Google have to pay royalties to Oracle; by the way, this might apply to Android software in Chinese smartphones, and other Google-branded products as well. Apple coded everything in Objective C language, which is quite ancient, without going through the C and C++ labyrinth. Now they have shifted to Swift, which is a very young language and is evolving at a quick pace.”
“Employees of large pharmaceutical companies are on permanent contracts, but not the applications that they develop and distribute to their patients and customers via the AppStore. These apps are, very often, on short-term contracts. In many cases, they have a life span of less than five years, and even less in some cases.”
I happened to hear this at Tencent, as the group is working with myriads of startups in e-health: “Your apps are on a long-term contract; you are not.”
I guess that in Big Pharma, it’s quite the opposite: employees are on a long-term contract, but not the apps they make…
So the apps for diabetes or multiple sclerosis, developed by such and such pharmaceutical laboratory may be sustainable, but perhaps not, especially if it was designed and pushed by the marketing department of the lab. Patients might not want to invest time and efforts and money (my medical data is money, right?) on some app that will probably go extinct in three years, because some marketing department will not have the budgets to update the iOS environment or / and evolve the app in Swift, because of the lack of trained staff… Keeping that in mind, Sophia Genetics probably has the good business model, even if they are not able to circumvent the medical profession to target clients or patients (B to C), and thus massively democratise genomic precision medicine (a very young science, evolving every day, like the Swift computer programming language) at the speed of light. Slowly but surely, as the saying goes in Switzerland. It’s a safe bet that in the near future, Sophia Genetics will be on the list of the 50 Smartest Companies of the MIT Tech Review 2018, in any case, as a potential patient, this would be my wish.
Jean-Michel Billaut: “Electronic Health Record? We’ve been on it since... 2004!
We spent money on it! Sources familiar with the matter are talking about sums on the order of more than 1 BILLION EUROS (public money — so our money)! Mind you, friends of friends might have taken advantage of the loot… EHR is still a bit of a mess. How come there was never a live demo in planetary session at the assembly? Nobody knows for sure what EHR really looks like. And, while they’re at it, our deputees should ask (and know) where we stand with the government’s “Genomic Program 2025" (in the middle of nowhere, same as with the EHR?) Why not sign an agreement with Sophia Genetics, which is achieving (with a solid business plan) precisely what the genomic program claims it is aiming at? Opacity, when you hold us… I do hope Macron and his team will sort out this whole mess.” (Source: Facebook, May 4th 2018).
