3 Innovative Approaches to AI-Driven Healthcare Implementation

Framework 1: Technological Innovation System (TIS)

This model, which is highly recognised amongst scholars, was originally conceived by researchers Carlson and Stankiewicz (1991) [1] using the “technological systems” denomination, but, it was Bergek et al. (2008) and Hekkert et al. (2007) [2,3] who put the focus to the functional patterns that learning flows of knowledge presented for innovation models, rather than just focusing on mere goods and services. Or, in other words, focusing on a dynamic network of actors engaged in the creation, dissemination, and application of technology within a given institutional framework, while operating in a particular economic or industrial domain.

In this sense, following the previous Swedish line of innovation research, a recent article by Petra Apell and Henrik Eriksson titled “Artificial intelligence (AI) healthcare technology innovations: the current state and challenges from a life science industry perspective” [4] used TIS conducting a mixed-research method approach to understand the processes for AI innovations, and their findings were quite promising, not only in terms of their contextual study to enhance the Swedish health system, but also as a chance to export this framework to suitable health systems globally involving numerous partners (Academia, Market, Governance) that could make use of it.

Following a functional approach, the researchers of this study used a combined TIS framework (Table 1)  establishing seven functions for analysing and extracting the next conclusions:

(Table 1: Technological Innovation System (TIS) framework) 

In this regard, the authors found some constraints and obstacles related to healthcare AI, which provide more insight into the need for a more compelling investment case. First of all, it was identified how not all decision-makers and managers had sufficient knowledge to understand AI innovations, which is opposed to not all AI experts having the business vision to deliver a solid proposal for a large AI investment. This, in my opinion, is a symptom of a breakdown in communication or a lack of connections across, supposedly, disconnected disciplines.

Second, a crucial aspect of any AI solution is data accessibility, and this aspect has always been controversial in the healthcare sector system configuration. This restriction results from a lack of resource mobilization and justification for building data-fetching infrastructure. Restrictive guidelines governing industry-healthcare collaboration frequently lead to issues with data responsibility, privacy, and consent forms for data reuse. However, in practice, accessing massive health datasets is far more difficult than, say, obtaining individual patient data for a business with no government participation. However, these are only safeguards against unauthorized access to patient information, which might lead to problems.  For this reason, I concur with the authors' assessment that one possible alternative would be to establish a "shared project portfolio platform" wherein government agencies oversee an interdisciplinary partnership with the aforementioned actors: academia, the market, and governance. This might be a means of expediting the process by granting access to additional resources, which is another way of saying “More data please”.

However, there is a fundamental aspect that sometimes is overlooked, and, from my experience working with AI-based DSS implementation based on multidisciplinary experts, I could identify that F1, through the lack of connectedness between technological knowledge and domain knowledge, is the initial cause of not building solid proposals for AI development in healthcare. One example is also provided in this study when a healthcare professional declared that: “Just because I am knowledgeable in various fields of medicine, does not make me capable of developing innovations relating to my field. We need to collaborate with the companies”.

This is a solid statement on how the nature of AI in healthcare should aim for multidisciplinary collaboration and knowledge sharing, instead of non-connected accounts from isolated experts. I based this conclusion on that innovation and collaboration go hand in hand, so these two concepts need to be properly initiated, communicated and connected to articulate concrete solutions that justify the commercialisation of the AI solution, otherwise, AI in healthcare will remain the domain of startups and small businesses, who are unable to scale their solutions unless they are acquired, merged, or bought by medium-sized or large businesses with potentially larger budgets. This is because, as of right now, decision-makers in large companies still do not see the right return on investment.