00:00:00
Voice-over: "One hundred days to success". This is the podcast from G_NIUS, the Guichet national de l'innovation et des usages en e-santé. Around Lionel Reichardt, meet e-health innovators and key experts to help you succeed in your projects.

00:00:20
Lionel Reichardt: Hello everyone and welcome to the podcast "one hundred days to success" the podcast for innovators and entrepreneurs in digital healthcare, but also for anyone curious about this field. This podcast is produced by G_NIUS, the Guichet national de l'innovation et des usages en e-santé. In this episode, we talk about recruiting a technical team with e-health expertise. To do so, we welcome Théo Ryffel, co-founder of Arkhn, a company that helps healthcare establishments to better manage the data they collect in order to facilitate the work of medical teams, improve patient care and foster innovation in healthcare. Théo Ryffel is also a doctoral student in federated learning and cryptography at ENS, ULM and developers at Open Mined. We also welcome Jorge Soto Romero, director of the ISIS engineering school, the first to train engineers specialized in e-health.

00:01:21
Lionel Reichardt: Théo Ryffel, hello and thank you for sharing your experience with us. Can you first tell us about your training and background?

00:01:29
Théo Ryffel: Hello. I have an engineering background. I went to Polytechnique with a background first in theoretical physics and then in data science. At the end of my studies, I did a year at Imperial College where I specialized in data sciences applied to the medical field. It was there that I became familiar with the notion of privacy, particularly in relation to medical data. At that point, I joined the Open Mined community. It's an open source community that develops libraries to make data confidentiality tools accessible to all. I actually started a thesis at ENS and Inria in machine learning and cryptography to continue doing research in this field.

00:02:07
Lionel Reichardt: You co-founded the company Arkhn in 2019. Can you introduce it to us?

00:02:11
Théo Ryffel: Arkhn works with hospitals to help them regain sovereignty over the healthcare data they collect. The idea is that when a patient arrives in a hospital, he'll go through a bunch of departments, through the emergency room, maybe afterwards through cardiology or the imaging department. In this case, it means that each department has its own specific software, which will store information on the patient and replicate, for example, his or her identity. So, if we want to have all the information on a patient, we need to have access to all this software. In practice, this is very difficult. So it poses a problem internally for care when we try to retrace a patient's care path. It's quite complicated, especially if you want to exchange data between different centers, because the information isn't accessible in the same place. What's more, we transform the data present in the databases of each software package, in a proprietary format, into a standard, open format called Fire, published by an American non-profit organization. In this way, a data warehouse is built up within the hospital, with all patient data in a standard format. This obviously facilitates the day-to-day work of practitioners to be able to easily access patient records, clinical research because we have much less time to give and also collaboration between centers because when data is standard, it's much simpler to exchange it between different centers. Arkhn was founded in 2019 and we are now a team of 17.

00:03:37
Lionel Reichardt: You're tackling a number of topics related to healthcare data, including the sea serpent, which is the interoperability of information systems. It's a highly technical subject. What profile do you need?

00:03:48
Théo Ryffel: Yes, indeed, we're a very technical team. We have three main types of profiles. The first are the developers. We do the hospital's plumbing, we connect with different databases, we extract data flows and transform them. So the major profile we need for this are developers from specialized schools. We can mention 42 Lenzi mags from Epitech and Epita, either just out of school or with some experience. That's the first part. The second part is that we need to understand this data. We need healthcare profiles, such as pharmacists or specialized schools. We have the ISIS school, for example, which specializes in medical informatics, or profiles who already have some experience in hospitals with handling medical software, such as clinical research associates, who are very familiar with these tools from which we want to extract data. The third component is data scientists. We tend to look for them in engineering schools like the one I came from. I'd like to add another section on recruitment. The first people we took on in our team had an appetite for managing a team, because the very first recruitment we're going to make is a very important one. The first person we recruit is likely to be at the head of a team of five or six people within six months or a year, as was the case for us. So, she needs to have the appetite to manage the team a little, to make sure that everyone is progressing well and that everyone is happy with their missions. These are the profiles we're looking for. Then, once we know what we want. How do we approach these people? The first thing is obviously the school network, which is quite powerful through private groups. You can just post ads, and that works pretty well. Various people from our respective schools have joined us. The second is mainly programs from the social economy. For example, we participated in the Goode data. These are people who often have a project and people join us once a week to work on it, or a program called Fridays. These are projects we do with a company that allows its employees to work one day a week on a project of social interest. It's very interesting because it allows people to get to know us and understand our issues. For us, it's an excellent way of spreading the word through all the related events. Then there are the more conventional events, such as those focusing on AI and healthcare. There are specialized medical conferences. These are places where you'll come across people who are already in their jobs, or students at the end of their studies. It's a good way of telling them about what we do. It usually echoes what they're looking for. And the third way, which also works, is simply to hunt on Linkedin, identify the right people and contact them. That's generally quite effective too. Once you've approached these people, how do you go about interviewing and selecting them? At the beginning, we had nothing at all. We were very bad on our first hires, but we were very lucky. It was just informal discussions. Today, we've formalized it into four phases. All our applications go through an application form. It's a simple Google Form and it allows us to process a large number of applications. If we have 50 people applying for a position, all the responses will arrive standardized in an Excel spreadsheet, enabling us to review them very quickly. Once we've identified the candidates with the greatest potential, we make a series of fixed 30-minute calls. We talk to the person and try to see if he or she is a good match in terms of values and desire for what we do. And that gives us a chance to explain a little about what Arkhn does. Depending on the number of applicants, we hold an initial interview, which may take place at home. I don't really like interviews at home, so I try not to have too many. In general, it's a fairly straightforward technical interview, followed by a more in-depth interview if ever we have candidates who are very strong and whom we're having trouble sorting out. Once we've identified these candidates, they join us. We've formalized the embording process quite a bit, mainly because our project is fairly technical and therefore a bit difficult to get to grips with. So it was important to us that the two or three sessions a week that are the most difficult should be well structured, so that people don't get lost. So it consists of three things a fairly simple entry path where we explain how to create an account for our tools. We use Google, Slack, Notion, Bitwarden and Guidel. For the discovery of the place, we're hosted at the Liberté de Living Lab, in the 12th arrondissement. It's a place that hosts lots of other start-ups and is very interesting. It's important to understand the people you're working with. And then, we often have a submission read out when we've put out a call for proposals, because it summarizes the company's strategy and current level of development in a concise, well-thought-out way, which is always more complete than a 30-minute discussion. The second element we put in place is enboarding sessions, i.e. 45-minute sessions on key points. These can be about our tools, Arkane's history and objectives, technical architecture presentation, strategy, operations... It's very important for us to be open about our strategy, to explain what we're doing so that no one gets the impression that we have a secret mission or that we don't know where we're going, about our values and, of course, about our data standard, which is called Fire. Because it's so complicated, it deserves its own session. And finally, we introduced friend limitings during the first confinement. These are 15-minute discussions with each team member, just to get to know each other in a somewhat programmed way. But it allows us to meet everyone quickly. Because some people work on such different subjects that otherwise they wouldn't talk to each other, except maybe at lunchtime. So, it's a way of forcing links within the team, so that everyone knows a little better who they're working with.

00:09:33
Lionel Reichardt: What type of profile do you prefer to recruit? Very expert and experienced profiles, or more generalist and versatile profiles?

00:09:43
Théo Ryffel: That's a good question. We choose profiles that are relatively beginners, for a reason that is quite simple. We're a startup that hasn't yet raised any funds. We don't necessarily have the resources for the most expert profiles. On the subject of the most expert profiles, I'd like to draw your attention to a small point, which is that it's actually very interesting, for a sales team for example, to have an expert profile because he or she arrives with a very strong address book. A technical expert can be a double-edged sword. Either they have a great deal of experience and can bring a lot to the table, or they can make very firm choices because they've made them. He's made certain choices for 20 years, and he's not going to change them now. That's people's point. That's why we've chosen people with two to four years' experience in development, a little less expert, but who are more open to changing architecture overnight, if ever we realize we need to pivot.

00:10:33
Lionel Reichardt: Have you thought about using recruitment agencies?

00:10:37
Lionel Reichardt: We've thought about it, actually. We have start-up friends who have tried it and it worked well. We didn't need to because we were very lucky. We found profiles that matched us pretty quickly. It's a bit like a choice you have to make when you can't recruit and you have a real need, an emergency or you know it's a good way. After that, it's expensive. It depends on our ability to recruit. For the moment, we don't need to. If one day we have an emergency and we can't manage to recruit the right people, we'll go through a firm

00:11:05
Lionel Reichardt: Do you encourage co-option within the team?

00:11:08
Théo Ryffel: We encourage co-optation because it's always quite interesting to say "We've managed to bring back a schoolmate." We also encourage them to pass on information about their various school networks. After that, we don't necessarily go beyond that. Everyone manages their own social networks as they see fit. We're not going to force them to communicate about the company. It's not necessarily a very healthy practice, but we encourage co-optation. We don't necessarily have big bonuses for people who manage to recruit people from their school precisely because we manage to recruit well. But it's something we could put forward a little later, when we need to recruit more.

00:11:43
Théo Ryffel: You use a particular standard, Fire, which is becoming widespread throughout the world. How do you train your staff in this standard? Do schools provide Fire training?

00:11:56
Théo Ryffel: En france, Les écoles ne forment pas sur Fire, donc il y a différents moyens de faire ça. The first is a bit on the job by attending specialized Fire events. For example, in November 2020, there was a conference for Fire developers, or it can also be by talking to other Fire experts. There's a very active community online, or also in France, at the Institut Curie, at the APHP. A lot of people are pretty sharp. In these exchanges, we share our problems or we share the solutions we find, and we get feedback that helps us, that helps us progress.

00:12:36
Lionel Reichardt: Do you feel that engineering schools in France provide training in topics related to digital health? Do you find it easier to find engineers specialized in healthcare?

00:12:45
Théo Ryffel: So, from the feedback we get on general engineering schools, they don't train very well in e-health. On the one hand, medical knowledge is still fairly limited in these courses, and on the other, the development aspect is not necessarily highly valued. We're more likely to value the data science aspect, with an understanding of theoretical concepts linked to artificial intelligence, or the managerial aspects that are important to us. In any case, we need these profiles, but not necessarily because of their e-health aspect.

00:13:16
Lionel Reichardt: To conclude, Théo Ryffel, what advice would you give to a healthcare entrepreneur who wants to surround himself with technical profiles?

00:13:23
Théo Ryffel: The first piece of advice is to be able to have good advisors. We were very lucky to have contacts who didn't work at Arkhn, but who are experts in the healthcare field, either in Fire or in software architecture development, and who enabled us to make the right choices upstream to go in directions we'd have to come back to later. That's the first thing about the team itself. Recruitment is essential. You need to recruit someone who is not only highly talented, but also a team leader, because he or she will necessarily have to manage a team. So they need to be comfortable communicating with others, and enjoy working with others. Otherwise, you're shooting yourself in the foot right from the start. And thirdly, I think the reason we've been so lucky at Arkhn in our recruitment is that we've created a good image. How do you make a company look good? The first thing is to have a subject where people say "OK, this makes sense to me. What I want to do when I grow up, that's got to be a good point." Secondly, we've shown that we're open and transparent. All our code is open source on GitHub. THE strategy is open within the company and it's clear our decisions are made as horizontally as possible. There are 17 of us, so we're not entirely horizontal. But the idea is to say, "OK, I'm going into this structure that's paying attention to me and tackling a subject that's close to my heart." And that's what's going to attract people and make it easier to recruit.

00:14:53
Lionel Reichardt: Théo Ryffel, thank you for your testimonial. Are you wondering about training engineers specialized in e-health? Elements of an answer with Jorge Soto Romero, director of the ISIS engineering school, the first to train engineers specialized in e-health.

00:15:12
Lionel Reichardt: Jorge Soto Romero, hello, can you tell us about your background?

00:15:16
Georges Soto-Romero: Hello, first of all I'm an INSA doctor from Toulouse, specializing in microsystem electronics. I switched to the world of healthcare in 2003 by joining the Lisiers Génie biomédical team in Besançon and doing several projects with the Toulouse University Hospital. And in 2013, I came back to Occitanie to be a researcher at ACNRS and projects with Toulouse University Hospital. Since 2017, I've been head of ISIS Castres, which is indeed the leading e-health school in France and has been training specialist engineers for fifteen years.

00:15:51
Lionel Reichardt: So the ISIS school is located in Castres. What specific e-health courses are taught there?

00:15:56
Georges Soto-Romero: We have several specific teachings. First of all, we have immersive hospital internships, because we believe that, indeed, digital health is developing in hospitals and in the service of healthcare staff. Other speakers on the podcasts have already said as much. We have specific courses on connected medical devices, artificial intelligence and Big Data in healthcare. And last but not least, we have a lot of project-based learning, which enables us to bring together several audiences: nursing students, engineering students, doctors and veterinary students. All this adds up to a compilation of teachings that put the student at the center of these healthcare systems.

00:16:39
Lionel Reichardt: You also have an Innov Lab and a Fab Lab. Can you introduce them to us?

00:16:44
Georges Soto-Romero: All right. We have an innovation platform dedicated to e-health, called the CHL. The Aide Lab collective is one of the Occitanie region's living labs, and one of the first in France, which enables students to develop projects using an innovative approach, to put themselves in situations with healthcare or other staff who also come to the site to be able to compare best practices and, for the men, explain what their needs might be, and for the others, try to develop demos, proofs of concept, innovative devices that meet the needs of the former.

00:17:21
Lionel Reichardt: In your opinion, what are the skills to develop when you want to become an engineer in the healthcare field?

00:17:27
Georges Soto-Romero: So, I would say that being an engineer in healthcare is not something you improvise. It's not an engineer. So in our case, for example, it's not an IT engineer where we'd put 50 or 100 hours of health in the last year of training, we'd make it an e-health engineer. It's really an educational program that extends over several years and in which there are not only digital disciplines, i.e. not just computer science or information systems. But there's also ethics, law, health, the organization and financing of the healthcare system, which are things we tend to learn in public health schools. And it's this combination that makes it possible to train an e-health engineer. It's also the fact that, as I said earlier, we bring together several audiences in our corpus. We'll have medical students, nursing students and veterinary students, and we'll turn them into e-health engineers because they're acculturated to the field. And finally, it's an apprenticeship program, so we put these groups together and get them to try and respond to problems that come directly to us, from hospitals that come directly to us, from companies. In the final year, for example, almost a third of the graduating class are like a mini research office, responding to concrete problems.

00:18:46
Lionel Reichardt: So you develop skills we call Hard Skill. What about Soft Skill?

00:18:51
Speaker 1: This is essential and I'd say it's essential in an engineering curriculum today, since we know that they're going to change jobs several times, if not professions, at least companies. So, there are several Softskills that are stimulated or practiced: entrepreneurial spirit, innovation and entrepreneurship is one of them. We also have specific courses on interpersonal skills and business communication. And we also benefit from the presence of a number of professionals from the healthcare sector, enabling us to position an engineer in a hospital system, in a clinic, not just within an IT department, but really within the healthcare services.

00:19:35
Lionel Reichardt: How do you ensure that your training is still relevant to the rapidly evolving e-health sector?

00:19:41
Georges Soto-Romero: On the one hand, in relation to the specificities or needs of healthcare professionals, we have a high percentage of healthcare professionals involved in our school. We even have a hospital relations manager who is a geriatrician at the Toulouse University Hospital, who is quite well known and who comes to challenge our teams to effectively ensure that our training is always in line with the reality on the ground. That's for the healthcare environment. For the corporate side, we have bodies such as the Conseil de perfectionnement, where our board of partner companies come to challenge the teaching team, saying "Here's the skill we need next year, because there's been a regulatory change, and this is the one that was the norm. So, you're going to have to acculturate your engineering students." This allows us to be quite reactive. We're also in a school that's the size where, in fact, in two phone calls to three e-mails, we still manage to get hold of the Director of Studies and set up from one year to the next, or even in the same year, professional seminars where someone comes for half a day or a day to explain a trend or a particular skill that's linked to their activity or a need that then generates a project, an internship, an apprenticeship contract or a hiring.

00:20:52
Lionel Reichardt: How and when can we recruit students specializing in digital health?

00:20:56
Georges Soto-Romero: Then we can already recruit them through apprenticeship, because we have an apprenticeship stream. So, in the event that a company would like to participate in the student's training process, it can recruit a student engineer through apprenticeship. Otherwise, it's mainly through internships, whether it's in fourth year, it's more around April, if not in fifth year, it's from January, for six-month periods, and then hiring, as in other French engineering schools, of course.

00:21:28
Lionel Reichardt: What advice could you give to a project owner or a company that would like to recruit engineers in the e-health field?

00:21:35
Georges Soto-Romero: Specifying the need, if it's indeed an e-health engineer we're talking about, we need to make sure that the engineer's training or professional background brings with it this dual skill set. In other words, one that combines digital technology and healthcare, and not necessarily only digital technology or only healthcare. It goes both ways. Secondly, you need to have, or at least be able to test the candidate's knowledge of the business and the various players involved. There may be engineers who call themselves e-health but don't know G_NIUS, or the ARS, or who don't know the various organizations that are trialing in our business sector. Finally, something you've already asked about is SOFT Skills. When recruiting, it seems essential to me that an engineer should be open-minded, and that the values of the company recruiting him/her should be in line with his/her own values. Particularly those who are capable of surpassing themselves or finding meaning. Today, we have many engineering students and graduates who need meaning. And if they work in the healthcare field, it's precisely because they want to contribute their skills to something that is of service to someone who is useful. It's a bit like humanizing computer science as opposed to just a school of computer science, and taking a kind of societal approach. If these are the values advocated by the company wishing to recruit an e-health engineer, it must be able to test the candidate on these skills.

00:22:59
Lionel Reichardt: Concretely, how can a start-up find the right profile? What channels are there to put them in touch?

00:23:05
Georges Soto-Romero: The usual channels, for example, we work a lot on LinkedIn. Obviously, we get direct applications from our corporate network comes in regularly via email addresses [inaudible] so when we're asked for specific profiles, we try to source to be able to match them. We don't have any particular procedure. But it's true that we're working more and more with social networks and LinkedIn, which spearheads it for professional insertion.

00:23:32
Lionel Reichardt: Are hackathons interesting meeting places?

00:23:35
Georges Soto-Romero: All right. We don't directly organize a hackathon in Castres. But we do have a very fine event that's well-known in the industry where our industrialists meet at the start of the summer called the Université d'été de la e-santé, which will be in its fifteenth edition this year if Covid allows us to do so, and in which we actually have the whole French professional network coming together here in Castres, and our engineering students are in direct contact with them. This enables us to exchange best practices and identify trends. Our students have a sort of open-air mega-course for three days on everything that's being done in e-health in France and abroad.

00:24:18
Lionel Reichardt: Our episode is coming to an end. Thank you for listening. We thank our two guests for their availability. Don't hesitate to subscribe to the podcast on the listening platforms. We look forward to seeing you very soon for a new episode of "One hundred days to success".

00:24:38
Voice-over: Those who make the e-health of today and tomorrow are on the G_NIUS podcast and all the solutions to succeed are on gnius.esante.gouv.fr.