Google has recently introduced its health app "Derm Assist". This app can be used to identify common skin diseases: The user takes a photo of the concerned skin area, answers a few questions, and receives - by means of AI - an initial diagnosis. Utopia? The app has CE approval for Class I in the EU. We take a look at the digital business model.
We reported in the spring on Google's development of an AI-based algorithm that allows to diagnose skin conditions. This algorithm was developed with dermatologists and has an accuracy of 90% compared to a group of dermatologists. However, this system was significantly superior to non-specialist physicians and nurses (Nature Medicine 2020). Now, a year later, Google has introduced the corresponding app, "Derm Assist." Users take a photo of the affected skin area with their smartphone, upload three photos to the app, and go through a questionnaire about skin type and symptoms. The AI algorithm uses the photos and patient information, compares them to 288 stored dermatologist-reviewed medical conditions, and lists several possible diseases that match the affected skin site, as well as additional information. The app is already certified as a Class I medical device in the EU (but interestingly, not yet approved by the FDA).
E-Health and Digital Business Models
There are now countless ehealth apps on the market, many of which are CE certified and more and more are also reimbursed by health insurers. Derm Assist is the next one, a few entries, a few data and then the diagnosis follows with the usual disclaimer that the app cannot replace a visit to the doctor. However, we find this app interesting from a medical point of view, because there are only few apps in the dermatological field so far and AI-based image analysis is primarily developed for radiology. Examples are the telemedicine apps, "Online Hautarzt - AppDoc" or "Dermanostic", developed by two German startups or "SkinVision" by the Dutch company of the same name. However, we are also interested in Google's app from the perspective of digital business models.
First of all, Google has not really developed any new diagnostics. Neither biomarkers for early detection or stratification of, for example, skin cancer. It's not even the first dermatology app at all. But what is interesting is that Google is not known as a diagnostic company like a Roche or Siemens Healthcare. Google is known to be an internet service provider and how does an internet service provider make inroads into the healthcare sector? The answer is, Google has used a typical pattern of digital business models: through combinatorics of innovation and ubiquity of knowledge, industry boundaries are dissolved. In other words, today's rapid technology development coupled with extensive knowledge available to everyone makes the barrier to breaking into a new industry much lower.
A test: does the adjacent picture show a harmless birthmark or already an early stage of melanoma?
The benefit of a dermatology app thus becomes clear. Either the all-clear can be given to the user, or a visit to the dermatologist is unfortunately unavoidable.
This pattern can be found in the iTunes business model, with which Apple revolutionized the music industry. Tesla another example. Google has not dedicated itself to researching innovative biomarkers for the early detection of melanoma, but has simply catalogued existing knowledge in dermatology and made it available to patients via AI-based diagnostics. Trivial or Smart? It says it is hitting a huge market defined by the more than ten billion searches on skin diseases.
The size of a company alone is not decisive; Roche has also started to develop an app for the diagnosis of neurodermatitis. And Google, with a development time of more than three years, is not significantly faster than other diagnostics companies. The problem is rather that companies focus predominantly on their core business and observe too few trends to identify opportunities. Investing in activities that are fraught with uncertainty or do not immediately make the same economic contribution as the core business is difficult. People prefer to wait and see. Then, when a new trend becomes established, the race to catch up begins. Experience shows, however, that this is more expensive and difficult - for example, a shortage of skilled IT staff - than perhaps investing in a trend that turns out to be a flop.
There are now numerous well-established agile or lean methods that allow early trends to be identified, prototypes to be developed and tested without large initial investments. This allows a company to gain experience and better assess risks. From this position, it can then react much more quickly when a trend begins to take hold and adapt or change its business model. In fact, introducing digital technologies inevitable changes the business model. Of course, it is better not just to react, but to actively shape the trend and take on a pioneering role. In a recent study by McKinsey, the health sector appears to be particularly vulnerable to digital technologies.
For example, in the first paper, Google developed the prototype of its AI algorithm with about 16,000 images. The final app is based on 65,000 images of diagnosed skin diseases alone, plus countless images of healthy skin or non-diseased skin lesions. Derm Assist is not yet on the market and it remains to be seen whether this app will achieve the desired success for Google. However, the future for digital business models that change the healthcare market has already begun.