The future of artificial intelligence and interpretative specialization in clinical biochemistry.

In a recent article in JAMA, Saurabh Jha and Eric Topol discuss the potential combined impact of digital images and artificial intelligence (AI) on the future of pathology and radiology [1]. The premise of the article is that these specialties are simultaneously inundated with information and amenable to automated interpretation. Automated interpretations are envisioned as an augmentation to the traditional model of radiology and pathology. Using AI (see Table 1 for key terms), the authors suggest that relatively simple diagnoses can and should be automated to free up these specialists to do more difficult work. By employing AI, it is suggested that radiology and pathology could be merged based on the fact that they do very similar things – interpret high volumes of image information. While the reality of AI taking over much or even any of the work of pathologists & radiologists on a routine and global scale is likely to be slow and incremental, the advantages of automation in terms of consistency, efficiency, and cost are tremendously powerful driving forces. These driving forces are quite different in biochemistry. In clinical biochemistry/clinical chemistry, there is comparatively much less new information and the interpretative role of biochemists/ clinical chemists is limited. This article explores how artificial intelligence and information specialization might be applied in clinical biochemistry in the near future. One of the tenets of the JAMA article is that the role of radiologists and pathologists as image interpreters will need to evolve. This evolution is being driven by the increased workload, cost, and the wealth of image data for each case, necessitating a reinvention of how information is extracted and assimilated. Information extractionwill be transformed by employing AI andmodifying their workflows using technological advancements. Such technological advances have long been embraced in biochemistry. As radiology transitioned to digital images, biochemistry had already implemented robotics and automation for routine tests [2], validated micropipetting advancements to use smaller sample