Despite extensive theoretical research in the field of medical artificial intelligence (AI), few developed models are evaluated on patients and implemented, which creates a gap between expectations and concrete penetrations of AI. A dedicated clinical AI department may help to bring the two sides closer, an article in BMJ Health & Care Informatics argues (Cosgriff et al. 2020).
You might also like: How Healthcare Can Unlock the Promise of Big Data
The authors point out that if a true disruption by AI in medical field is to be achieved, “it should be as painless as possible to circumvent further, or perhaps even reduce, clinician burnout.” Otherwise, the path of AI applications in clinical practice may resemble that of EHR, which aimed at corporate and administrative benefits but sidelined improvements in process optimisation or patient outcomes.
With AI being widely implemented in other areas, such as advertising or banking, in healthcare its application is hindered by the “lack of coherence, leadership and vision.” The process of a new tool adoption may be facilitated if a cross-discipline AI department assumes coordination and supervision.
Establishing a clinical AI department has several implications:
- due to various implementation and policy issues, a clinical AI department should be charged with making health centres AI Ready. This can be achieved through guiding and directing the development community, adequate data design and interoperability, adjustment and continuous revision of applied models, and so on.
- such departments may play a key role in the process of changing the reimbursement models in line with the increasing AI deployment. Here, however, the authors warn about the need to calibrate ready-made solutions available in the market, to the needs of specific patient populations and to negotiate these calibrations with payers.
- AI adoption in clinical practice will require standardisation, hence best practice guidelines should be developed and implemented. In addition, AI-driven results must be transparent and explicable so that physicians could integrate them into their routines easily.
In conclusion, the authors acknowledge that AI will be “a defining technology” in clinical practice. To achieve any meaningful clinical outcomes with this technology will require efficient cross-disciplinary collaboration under the auspices of a department of clinical AI, and the question they ask is – who will establish the first department of clinical AI?
Source: BMJ Health & Care Informatics
Image credit: Cosgriff CV et al. (2020)