Can you introduce yourself and tell me a bit about your current role?
My name is Jonathan Hall, I am one of the co-founders of Presagen; which has its first product Life Whisperer in the IVF space. We look at healthcare, specifically women’s health focusing on FemTech, AI technology, and producing new AI tech for a range of different medical problems that we’re really starting to solve.
One of our champion products is Life Whisperer. Life Whisperer has a focus on selecting healthy embryos for the IVF process. Health includes a number of factors; including viability; whether the embryo will lead to a pregnancy, and also the genetic integrity of the embryo.
IVF is a significant industry, and an industry that is a main focus for us to solve key challenges, and to help couples who are struggling to have children. In terms of my role in the company, as one of the founders, I wear a lot of different hats. I built the first version of the technology, I am the Chief AI Science Officer, Quality Manager, and worked with the patents and core tech.
Can you tell us more about Life Whisperer?
Life Whisper is an online web-based application that you can access through a browser. IVF Clinics can be subscribed to it; so a clinic might reach out to us or our distributors directly. They might hear about us from some other work we have done, or we might actively go out and make clinics aware of us and the product. Those clinics would then have a brief conversation with us to understand what value they are looking for, so we would work with them to work out where we would provide the most value to them.
Some clinics really like the fact that we can make their processes more efficient, some clinics have reported that it can help to standardize their processes, and some are more interested in the accuracy improvement. We work with them to understand what they’re looking for so that we can help both their clinic and their patients.
How it works
The patients are important, and are the ultimate recipients for this healthcare – they are not to be left out, they are an important part of the conversation. The way the product works is that staff within a clinic would be offered an account login, they would use a browser window, and then they would login with secure access to our system, which is fully compliant with all privacy laws in each of the regions in which we operate, which is roughly two-thirds of the world’s market. This includes GDPR for European regions; we pay a lot of attention to the regulatory details. It is a regulated medical product, so it’s treated to medical regulatory standards.
Clinics would then drag and drop microscope images obtained 5 days after IVF, receive an instant report as to the health of each emryo, and then can choose a suitable embryo with the patient, using the downloadable Patient Report. Receiving guidance from a peer-reviewed and tested algorithm as to which are the healthy and unhealthy embryos helps guide the overall clinical decision. The clinician and the patient together would make that choice and have a better outcome statistically for people going through the IVF process.
What inspired you to start Life Whisperer?
When I was working and studying in the university research sector, I became very frustrated at the culture of continuously applying for grants and scholarships, with little time to commercialize work and bring it out to the community. I was working with the ARC Centre of Excellence in Nanoscale BioPhotonics. The creation of new Healthcare technologies was the key focus of that Centre, and they helped researchers to try and find, and explore projects in this industry.
I ended up working with a group that was a fertility lab, the Robinson Institute at the University of Adelaide. We identified that embryos had interesting physical properties, and my background was in particle physics, and resonances in finite volume cavities. The core skills I had developed through working with the Centre for the Subatomic Structure of Matter at Adelaide University.
“People want non-invasive processes”
The shape of an embryo is round and symmetrical, and it was noted that it could have a lot of engineering properties that could really help and understand health, with this symmetry. While the initial directions and subject of my thesis was on generating Whispering Gallery Mode resonances, to make (deceased) animal embryos ‘lase’; the ideas that we championed in our group of colleagues in the medical area, stemmed from the fact that people want non-invasive processes.
There is tech out there being developed, but it can often be toxic, using chemicals and dyes to analyse animal embryos post-mortem. The idea was to do experiments in the lab safely on dead animal embryos, and then to turn the simulation into a computer programme, and then the computer programme is marketed.It is non-invasive, and it’s got the research in the science backing. We went through a number of entrepreneurial courses, notably Australian eChallenge, and CSIRO On Prime program,and workshopped the ideaTt became a cloud-based software.
It was through this journey that I met my Co-founders of the business, our CEO Dr Michelle Perugini and Chief Strategy Officer Dr Don Perugini; and soon after we hired our first senior software engineer, CTO Andrew Murphy. This was really driven by a strong desire to branch out and create something that made a change.
Can you tell us a bit more about Presagen?
Presegen came about after my co-founders Michelle and Don Perugini successfully exited their previous Predictive AI business to EY in 2015. Our goal was to become the most innovative AI healthcare company in the world.
As we began to collect data to train the AI for Life Whisperer, we realized that a continuous cycle of securely acquiring data from clinical partners would be difficult to scale without the use of a global network and collaboration. Our processes have allowed us to become an ‘AI Factory’, where we can quickly productionize medical-grade AI for solving a range of medical problems; but that requires a lot of high quality data, sufficient data diversity, and domain knowledge from experts in the field. Oftentimes though total amount of data required is not the main barrier, but data quality.
Medical data is often messy, and contains errors. Sometimes these are errors due to manual compilation, but more often than not, the errors are due to the fact that clinical work is subtle and it’s not often clear what the assessment or diagnosis should be. This can contaminate the dataset as a whole, very easily.
We struggled to educate clinics as to the safest and most secure manner to deidentify and compile the specific data needed for creating an AI application; but we realized early on that we had to create a format template that people could use to contribute data, but without us having to actually take any of the data from them. We needed to work out a way that we could extract only the specific learnings from the data that we needed to scale.
“Presagen is the Social Network for Healthcare”
We came up with the idea that the clinics would keep the data, we didn’t even want to see it; if they can provide just the format of that particular data, then we can use our unique (patent pending) Decentralized Training algorithm to send the AI to them – bring the learnings,which are not private data, back to us and we can then co-create a product with them for their industry.
Presegen champions the idea that the data does not need to be collected. Instead, the algorithm goes to the clinic, the learnings come back to us, and then they partner with us to co-create the product together
In many ways, Presagen is the Social Network for Healthcare. Clinics and hospitals around the world partner with us to solve medical challenges and to create products. In the same way as social networks, learnings are extracted from the data provided by these partners. The learnings are transformed into products, with a revenue share for partners that assisted in co-creating these products, which is then sold to other clinics so that patients can benefit in the improvements to healthcare with these cutting-edge applications. Everyone benefits – the patients receive better healthcare, the clinics get improved technology and workflow, with some clinics receiving a revenue share, and we are able to drive forward the next product for our AI Factory.
What challenges have you faced as a start-up?
There are a huge number of challenges. It is not easy to solve medical problems! There is often a misconception that you can “just throw data at an algorithm”, and it will make a product, but it is much harder to solve a medical problem when you have to make sure that all the proper controls are in place, making it reliable to the standard that is necessary for the industry.
The most crucial aspect of this process is the communication with domain experts, such as the clinicians and doctors who understand the problem needing to be solved. Often it is easy to jump in and start creating product designs without understand precisely what is needing to be solved, in a sufficient level of detail. It takes a long time and many detailed conversations to reach a point where the problem is understood crisply enough, that a solution can begin to be designed.
Defining your data requirements
One of the hardest parts is defining your data requirements. How do you understand your data? Also, how well you understand the different groups within the data and different demographics so that you really are aware of what problem you are solving, andwhat problem you are not solving.
Another challenge is understanding exactly how the product will roll out, and precisely what is itspathway to production, initial release, and scaling in the market-place. Going through the regulatory pathway is also very difficult. Not many people are able to do that because it is very expensive and time-consuming. Next, it’s how well you communicate the message to the doctors, scientists and the public in a way that they will understand. It’s really important to find the best way of communicating what what the product does and does not do. You want people to be educated in what it means, and what it will do for them before they use a product.
How has the way medical data been used, changed over the years?
I think there has been a rapid increase in machine learning methods applied to medical data, with total number of publications doubling over the last few years. I think it is being talked about a lot more in the general public I think there is a need for this now, as people can’t look through the number of medical records that are out there – there are simply too many!
Machine learning has the power to do just that; the algorithms are there, the servers and compute utilities needed to run the algorithms are there that can achieve this. People are now being able to leverage their medical data and transform the insights into products -I think that’s how it is being used differently to how it was before.
How does AI play such an important role?
It’s a disruptive technology, in that it is only now starting to be used in a commercial and enterprise way that’s really changing the face of healthcare in a big way. Having a disruptive technology of this kind and ways of creating new products, can potentially touch almost every area of healthcare.
Computers are simply better than us at looking across large datasets consistently. Having applications and tools that can take on the continual, repetitive tasks that take up many hours of a professional’s time means that we as people are able to spend more time being human. Doctors can spend time talking to their patients rather than filling out spreadsheets.
What other developments do you see emerging over the next 10 years in Health tech?
I believe there is a key role being fulfilled in machine learning that hasn’t been fully realized yet – because of its ability to look across different areas of research and subject matter. We live in a society where people are often isolated, unaware that experts in related fields of research could be reaching similar insights without ever knowing that they are converging into one topic, because the terminology used in different fields of healthcare is so varied and disaggregated.
Machine learning can act as a lens, giving the ability to see patterns across a broad range of subjects, and highlight when people are doing similar things so they can talk to each other and unify their understanding. I think that has a really strong collaborative effect; we might be able to see collaborations across industries that have stringent privacy laws, without ever exposing the privacy.We are going to need these tools to be able to cope with the amount of data that we have, both now, and in the future. In a sense, it is helping to boil down all this data into something that as humans we can understand.
What advice would you have to other medical and health tech start up founders who are just beginning their journey?
It can sometimes be demoralising at the start when it is difficult to get those first collaborations! Getting the first partnership or collaboration to access data is the first step and is very hard. But, once you find that first collaboration to help guide the product design and creation, you are then able to create and begin productionizing an applicationthat really works, and can potentially scale to many customers.
I also recommend that you prepare your product documentation for regulatory and compliance well in advance. If you are working in any type of regulated space, you can’t start that too soon! As soon as your design and proof-of-concept is established and has market interest; begin write up descriptions of your product and processes. Very few tech startups are able to say that they are compliant to a medical device standard and that they have the necessary regulatory documents, so that’s absolutely crucial as a separator from the competition.
The team is very important, it cannot be done without teamwork and trust. Having people on board who are able to work really close together and get through the hard times together is more than important; it is a necessity. That’s probably true of any start-up, whether in healthcare or not.
It’s all the more important however in healthcare; because having to prove continually to regulators and doctors (who can often be a critical audience) means you have to develop the way you communicate to them. The total amount of activity required of even a small business, will be beyond any one person’s ability to fully track, without trust in a team. The product development requires so much upfront workshopping, and back and forth conversations with stake-holders, to make sure it’s correct, and that is where the teamwork comes in.
Looking for help recruiting to your business?
Speak to our expert team about how we can help you to locate and employ, the top talent for your company.
See what Besik Mikhelashvili, Director of Talent at Happify Health said about us!