Some of the Deadliest Diseases Can Be Prevented by Using Modern Technology Solutions – Interview with Michał Sosiński, CEO at Soma Healthcare
Machine learning solutions have been used in healthcare for decades, but they only became popular a few years ago, mostly due to developments in deep learning.
According to PwC, in the next four years, the market for AI in healthcare will grow from $760 million to $6 billion.
Machine learning in healthcare not only allows providers to automate administrative tasks and collect information, but also improves communication between doctors and patients.
We decided to take a closer look at patient management apps and get to know how the conservative medical environment sees the development of technology.
Michał Sosiński, co-founder and CEO of Soma Healthcare, a consulting company focused on supporting the daily activities of doctors and patients, answered our questions and explained why it’s important to take a closer look at the implementation of machine learning for healthcare.
Are you interested in the full Patient Management apps report? You can find it here.
Natalia Chrzanowska: Technological solutions are currently present in many areas of life, including healthcare. Can you tell me a bit more about patient management platforms? What is the biggest challenge for a person who wants to develop a patient management system?
Michał Sosiński: It’s a very broad topic but, due to my experience, I will focus on 2 markets - the US and the UK. Before we start to implement any of these platforms we have to take into consideration 4 aspects.
The first is the legislation process, so all the aspects necessary to allow us to start using patient management platforms. Secondly, data protection, as we are talking about really sensitive areas. Thirdly, visual aspects - each platform has to be really intuitive.
We have to remember that older people will use these applications as well and they are not as fluent in using online tools as younger ones. Luckily, today’s 60-year-olds are more digitally savvy than their peers ten years ago.
Last but not least - convincing the conservative doctor environment to use these platforms.
What are the biggest benefits of those platforms?
First of all, from my point of view it allows me to be in touch with my patients.
I wish to avoid as many unnecessary visits as possible, because they mean unwanted losses for both the patient and me. I prefer to receive their results in advance so I can analyze them and decide if a visit is needed.
Secondly, I see great potential in all kinds of apps that send reminders to take medicine or help people to track how they feel.
What do you think about machine learning solutions in medicine? Are they our future? Where can they be used?
Machine learning is the next natural step after data digitization. We’re switching to digitized data, which provides us with extensive training sets for ML.
Having them makes it easier to analyze and predict if a patient might be at risk of getting sick, so it’s easier for doctors to contact him or her in advance, before the illness develops. Diabetes, hypertension, cardiovascular diseases - these chronic, multi-organ conditions are currently widespread among people of all ages.
They lower quality of life as well as life expectancy, while effective management, rather than expensive pharmacotherapies, is enough to significantly stop their progression.
Technology will help patients to check their health on a regular basis. It will also support therapy individualization, taking into consideration more characteristics needed to choose the proper pharmaceuticals.
Are there any areas where machine learning might not work as well as expected?
It’s a very good question. I will get back to the regulatory aspects of patient management solutions.
Before something gets through regulatory bodies it has to be tested clinically. After a number of tests we can verify if the tool will give us suggestions or decide about the illness. The whole process is very strict and if the tool doesn’t work properly it wont go to the market.
Doctors do not make decisions based on what they’ve read in books written a long time ago. They have clinical practice guidelines that support their job. Something like - if this parameter equals this then you have to do this, and technology supports them in this matter.
The only area that comes to my mind that might not yet be good for machine learning is mental health.
Doctors diagnose the patient based on either free talk or various standardized questionnaires in which patients talk about their frame of mind. Their descriptions of how they feel might be very subjective, so machines might not interpret them correctly.
Nevertheless, I’m not saying that this area is always be bad for machine learning. The technology is always improving, so it’s only a matter of time when it becomes suitable also for this area.
Do you think that the conservative market is blocking new machine learning solutions?
Yes. Everything has to be tested properly and it takes time, so strict rules have to be in place.
But American and British healthcare markets are very well-funded, so machine learning processes are in place and will be developed. Furthermore, there are some ways to make it faster.
Let’s take the example of the Food and Drug Administration in the US. If we prove that one device works in a similar way to another device that has been already accepted, the whole legislation process is way easier and I get certified faster.
There are even special lawyers specializing in that kind of legislation. In case of the United Kingdom, the NHS has a special grant programmes for technological companies that supports technological development. I’m not saying that implementing new technologies into healthcare is easy. But it’s definitely worth it.
I couldn’t agree more. Thank you for the interview.
If you are interested in broader analysis of patient management apps, take a look at our latest report: https://www.netguru.com/patient-management-apps-report.