Saventic Health

At Saventic, we are solving the problem of misdiagnosis of patients with rare diseases. Up to 10% of the world's population suffers from those diseases and over 80% of them are misdiagnosed or never have their disease addressed at all.

We propose a data-driven approach that leverages big datasets of electronic medical records, combines them with state-of-the-art machine learning algorithms and provide doctors with a list of patients with the highest risk of certain rare diseases. It simplifies the process, makes it cheaper and most importantly increases its accuracy.

Rare diseases are indeed rare but their patients are numerous. Up to 10% of the world's population suffers from rare and ultra-rare diseases, however only one out of eight from this group is correctly diagnosed and treated.

Today's approach is very basic and backward, but the industry is aware of the big change coming soon. Currently, pharmaceutical companies offer educational courses on rare diseases, provide materials, field workforce meet and talk to doctors on a regular basis to increase the awareness of rare diseases. Furthermore, pharmaceutical companies sponsor programs for doctors to review medical documentation (often in paper form) to diagnose and identify patients with rare diseases.

In practice, the approach to diagnosis of rare diseases described above does not work for the following reasons: lack of medical specialists today and the problem will expand in the future, limited know-how in rare diseases, natural focus on common diseases, high cost, misdiagnosis, mistreatment, deaths, no technology tools supporting doctors on a daily basis in the process. Finally, the number of medical records is increasing so quickly, so that it gets very hard for the doctor to timely analyze the existing medical data, conclude and react properly.

We propose a data-driven approach that leverages big datasets of electronic medical records, combines them with state-of-the-art machine learning algorithms (supervised learning based on historical data) and provide doctors with a list of patients with the highest risk of certain rare diseases. It simplifies the process, makes it cheaper and most importantly increases sensitivity and specificity of the diagnostic process.

We've already managed to collect a database consisting of 500,000 patients by reaching out directly to hospitals in Poland. We've also successfully built and tested our first rare disease algorithm, which is currently being deployed inside two Polish hospitals in order to diagnose patients on a regular basis. We've also started working on two more algorithms and reached out to the next couple of hospitals in order to expand our database even further.

Our algorithms are being deployed to a stationary computer at a given hospital but we've also built a web API in case a given patient management software is capable of sending HTTP requests, which will allow us to scale our services even faster.

Our target population is very large, as around 10% of the world's population suffers from rare diseases. This could be literally anyone. We have established very good relations with a number of rare disease patients. We've talked to them numerous times and they've been generous enough to provide us with their medical data so that we can use it to build even better AI algorithms.

Talking to those patients helped us understand that very few of those patients get correctly diagnosed and treated. They often spend years being sent from one specialist to another while never receiving a satisfying diagnosis and treatment. Our solution will help doctors diagnose those patients as early as possible and spread awareness of those rare diseases.

Rare diseases are a problem in the healthcare industry that has not been properly addressed just yet. As we mentioned earlier, rare diseases are indeed rare but their patients are numerous. Our AI-based solution is a very modern way of approaching this problem and we're one of the very few companies that are trying to address it in such way.

Today's approach is very basic and backward, but the industry is aware of the big change coming soon. Currently, pharmaceutical companies offer educational courses on rare diseases, provide materials, field workforce meet and talk to doctors on a regular basis to increase the awareness of rare diseases. Furthermore, pharmaceutical companies sponsor programs for doctors to review medical documentation (often in paper form) to diagnose and identify patients with rare diseases.

In practice, this approach does not work. We, at Saventic, propose a data-driven approach that leverages big datasets of electronic medical records, combines them with state-of-the-art machine learning algorithms (supervised learning based on historical data) and provide doctors with a list of patients with the highest risk of certain rare diseases. It simplifies the process, makes it cheaper and most importantly increases sensitivity and specificity of the diagnostic process.

As of today, we have a small number of potential competitors but no other solution integrates directly with patient management software, which we believe is our biggest asset.

Our solution, from a technology perspective, consists of three parts:

- machine learning algorithms that help detect patients with rare diseases

- web API for integrating with modern patient management software

- containerized standalone application that's ready to be deployed on any machine in any hospital, in case a given patient management software does not support sending HTTP requests

We have managed to deploy our application already in two Polish hospitals but it would be very hard to show a proof without reaching out to hospital officials directly.

We are daily engaging in the below activities:

- building better machine learning algorithms

- deploying our algorithms in hospitals

- talking to rare disease patients

Our expected outcomes that impact the problem:

- better accuracy at diagnosing patients with rare diseases

- having our algorithm running at hospitals in order to regularly diagnose rare disease patients

- understanding the problems of patients with rare diseases better

We scan 10,000 patients in each hospital for each rare disease monthly, currently operating in two hospitals and diagnosing one rare disease. We're planning to expand to 10 hospitals with the next year and add four more rare diseases, which would equate to 500,000 patients per month.

Our five year plan is 50 rare diseases and 100 hospitals, which would mean 50,000,000 patients scanned each month.

We're planning to cover 5 rare diseases and operate at 10 hospitals in two countries by this time next year. Our five year plan is 50 rare diseases and 100 hospitals in 5 different countries.

We currently only obtained data from Polish hospitals, which means that if we scaled outside of the country, we'd need to rebuild our algorithms because different populations have different feature distributions.

Another barrier is that we don't have very well established relations outside Poland, so convincing hospital executives to let us deploy our algorithms at their hospitals will take more effort.

Before deploying our algorithms at hospitals outside Poland, we will collect sample datasets from these hospitals in order to rerun the learning process and make them work with more accuracy on that given population of patients.

We have managed to already partner with a global pharmaceutical company, which has well established relations at hospitals all around the world, so we can leverage that partnership in order to help us convince hospital executives to let us deploy our algorithms at their hospitals.

9

We have a team of:

- our CEO has 15 years of experience in healthcare consulting

- our CMO has 17 years of experience in medical practice

- we have a team of three great programmers advised by a PhD of computer science

- our medical team consists of experienced hematologists and oncologists

We have already managed to partner with an international pharmaceutical company that produces drugs for rare diseases. We are working together in the process of deploying our AI-based algorithms at hospitals.

We have three groups of customers:
a) pharmaceutical business: pharma companies pay Saventic a subscription fee for every single algorithm in a single clinic implemented. Our solution diagnosis new patients for expensive treatment,
b) CRO business: CRO companies pay Saventic for identification of patients to carry our clinical trails in developing new orphan drugs to treat patients with rare diseases in the future
c) medical providers: clinics / hospitals pay Saventic for using our algorithms that diagnose patients at early stage what provides significant cost savings of the diagnostic & treatment process (the faster the diagnosis the cheaper the treatment)

Our funding is a combination of:
a) investors' funding in the future
b) commercial grants from the pharmaceutical business
c) public grants (Poland and Europe), and finally
d) revenue from operating activity:
- monthly subscription model per algorithm, per clinic (pharma & CRO)
- monthly subscription model per user, doctor (medical clinics)

We are a young startup with a great vision but we would greatly benefit from a partnership with experienced advisors in order to help us polish our business model as well as help us broaden our professional network in order to receive more opportunities for business partnerships.

We are using machine learning algorithms in order to improve the process of diagnosis of rare diseases.