Startup Jedi
We talk to startups and investors, you get the value.
Become an author
Settings
Sign Out
Two heads are better than one. This principle is even more relevant for the medical field than any other. A fresh look helps to find a mistake and adjust the treatment, so patients often pay visits to several specialists for consultations. The Medicalab App startup offers an option of automatic consultation instead. You can upload the results of your blood test and get a recommendation regarding the diagnosis. The product has got the team to the batch of the American Newchip accelerator. We talked to the startup founders to find out what it’s like to work in such a complex niche.
Startup Jedi
We talk to startups and investors, you get the value.
Let’s get to know each other first. My name is Artyom Trofimuk, I am the Co-Founder of Medicalab App and an android developer with an interest in machine learning technologies and experience in medtech projects.
There are two Elenas working in my team — an Internet marketer with experience in IT companies and a resuscitator, who’s a clinical intern. John — a PR specialist with successful experience in journalism, PR and business communications at the New York Times has also joined us at some point. It was our common interest in fresh breakthrough products that gathered us together.
We found each other quickly thanks to participating in Social Weekend. A team with all the necessary competencies gathered in mid-February 2020 at the first stage of the competition. This was when we began actively working on Medicalab App.
Our entire team had experience launching their own products from scratch, we managed to take lumps while keeping a burning desire to make our own products. It only remained to start working.
The idea for a startup was brought by a tick that infected our Co-Founder and marketing specialist Elena with encephalitis. It took the doctors over a month to make an accurate diagnosis and Elena had to study medical literature herself, delve into it, and ask specialists more clear and specific questions.
She tried to find modern and affordable ways to understand her current medical tests and reviewed many websites, forums, and medical applications. But there seemed to be no working way to get recommendations on the results of the tests.
I believe that such a method can be created in current machine learning technologies and the idea has potential in the market. Indeed, 100% of people living today have blood.
According to MarketsandMarkets analysts’ forecast, the blood test market will grow from USD 2.2 billion in 2019 to USD 3.3 billion in 2024. As the number of medical tests grows, so does the demand for them.
The scenario for using the Medicalab App will be as follows: we upload the blood tests data and get a recommendation in response: what diseases (in decreasing order of probability) these tests may be the indication of. You can also add gender, age, your chronic diseases, blood pressure, pulse, temperature, symptoms. The more data there is, the more accurate diagnosis you get.
Such service comes in handy for both patients and doctors as if substitutes consultations with several specialists. We do not make a diagnosis, but give recommendations instead and reduce the likelihood of a mistake.
...
A neural network operates in the doctor’s head, trained to recognize diagnoses by protocols. Scientific papers that describe the clinical picture of diseases serve as input data. This is where you can find the algorithm “if there is this, this and this the diagnosis is X. And if we have this, this and that it means the diagnosis is Y”. Monitoring the operation of this neural network takes place on an exam. And then every day a doctor works.
We automate this principle with machine learning. In the start there is a training dataset from the results of medical tests and diagnosis based on them. By comparing the downloaded test results with the “experience” obtained based on the training data, the neural network will make its diagnosis.
Of course, in our case it is impossible to find ready-made datasets. Therefore, we collect our own in three ways: through independent markup of data according to the medical protocols of the world’s leading medical organizations, partnership with private laboratories and the doctors community. Practicing physicians share data with us because they’ve been willing to use such a product for a while already.
We select the size of the dataset empirically. Now we have a small set of data and we are conducting the first tests on it. Elena, our medical specialist, is in charge of checking the results. The first results show that there is enough data to launch the alpha version. We’ll proceed by negotiating with laboratories in Belarus and the USA.
It’s worth mentioning that we focus not only on the biochemical analysis of blood. We are adding support for general blood and urine tests so that you can get a complete picture.
...
We plan to pack our neural network into several products. One for end customers, another for doctors, and a third for medical facilities. The products will differ in functionality and we will offer custom solutions for clinics’ needs. We plan to sell products simultaneously. Our monetization is based on the Freemium model. A certain number of checks (we will test the exact number) will be available free of charge with more offered by subscription.
Our main markets are the USA and the EU, due to the introduction of innovations fast enough for the medical field and the much larger market size and number of potential users than in the Eastern European market.
The second priority market for us is the CIS, in particular Belarus, where we have contacts with doctors and a common cultural space, which simplifies negotiations with partners. Even in Belarus, we see a big demand for such applications. There has been a benchmark case recently: after a story about us appeared on Belarusian television, dozens of people reached out to us on social networks with a request to have a look at their medical tests.
...
You need to get an 1A certificate which confirms compliance with basic quality requirements to work in the EU. It is technically free, you just need to pay € 5000 for the preparation of documents.
Yet, it will all get much more complicated starting from May 20, 2021. After this date we will have to get a 2A level certificate for the operation of our service in the European market. The procedure for obtaining one is much more complex and more expensive. We will have to undergo an independent study in three clinics, which takes about a year on average.
The average cost of this procedure is about € 300 thousand. There are grants available for this and we are working to attract such funding since the EU is an important market for us.
Expensive certification makes entering the market of ML-solutions for medicine very difficult, while implementing certified solutions takes years. Medicine is not the industry most open to innovation, because each change can potentially harm someone’s health, and to avoid this, many checks are needed on different levels.
This is why we do not see finished products that would provide a service similar to Medicalab App. Almost all medical-related applications from the USA, EU and CIS were included in our analysis, making it about three hundred applications in total. No solutions similar in technology were found in application stores, in search, or in doctors’ phones.
Most likely, there are projects that develop solutions that are essentially similar to ours. The lack of market players can be explained by a high entry threshold and difficult access to data. But it will definitely appear — we can see the interest.
The fact that we are moving in the right direction was also confirmed by the Newchip accelerator. We made it to a new batch. We are about to go through a three-month preparation program for launching the product on the main medical technology market in the world. We are very inspired to be able to get selected on the first try.
...
We are currently preparing to launch the alpha version of Medicalab App. It will recognize frequently occurring diseases in biochemical analysis. The plan is to release it in October 2020.
Once it’s been launched, we plan to attract 5,000 users each month, in this forecast we relied on the tops of medical applications in our target markets. If they can grow at such a fast pace, why set the bar lower?
We are looking for a half a million dollars investment to launch the first version of the service and promote it. This is a lot for a pre-seed round in Belarus, but the biggest part of it is certification costs, although we plan to attract a grant for obtaining a European certificate, we need to be prepared for different scenarios.
The development of technology in our case also requires a lot of effort and money. Therefore, we want to attract a sufficiently large amount for 1.5 years of steady work. We are looking for an investor who will see that the ocean is still blue enough to build a big business.
Facebook: facebook.com/StartupJedi/
Telegram: t.me/Startup_Jedi
Twitter: twitter.com/startup_jedi
Comments